KATHOLIEKE UNIVERSITEIT LEUVEN Faculteit Psychologie en ...inner+change.pdf · Onderzoeksgroep...
Embed Size (px)
Transcript of KATHOLIEKE UNIVERSITEIT LEUVEN Faculteit Psychologie en ...inner+change.pdf · Onderzoeksgroep...

1
KATHOLIEKE UNIVERSITEIT LEUVEN Faculteit Psychologie en Pedagogische Wetenschappen Onderzoeksgroep Psychodiagnostiek en Psychopathologie Onderzoeksgroep Psychotherapie en Dieptepsychologie TOUCHING INNER CHANGE Psychoanalytically Informed Hospitalization-Based Treatment of Personality Disorders. A Process-Outcome Study. Proefschrift aangeboden tot verkrijging van de graad van Doctor in de Psychologische Wetenschappen door Rudi Vermote onder leiding van Professor Dr. H. Vertommen, promotor Professor Dr. J. Corveleyn, co-promotor

2

3
Touching Inner Change. Psychoanalytically Informed Hospitalization -Based Treatment of Personality Disorders. A Process-Outcome Study. Rudi Vermote Promotor: Prof. Dr. Hans Vertommen Co-promotor: Prof. Dr. Jozef Corveleyn
The RCT of Bateman & Fonagy (1999, 2001) showed the effectiveness of psychoanalytically informed hospitalization for borderline patients. Such a treatment is not symptom oriented but targets psychic changes, denoted as the psychoanalytic process.
Given the intensity and length of this treatment and the severity of the pathology, it is important to examine whether it is possible to measure these hypothesized psychic changes and their relation with outcome. Furthermore we wanted to examine whether we could discriminate different patterns of change and their relation with client variables, as this would permit to better adapt the treatment to the needs of the patients. For such a study a naturalistic design with model specific measures was most appropriate (Clarkin, 2004).However till now, there was no clear definition of the psychoanalytic process in Personality Disorders (P.D.), only few instruments to measure it exist and a link between process and outcome was shown to be weak or inexistent (Vaughan & Roose, 1995).
In the first part we define the specificity of the psychic changes in P.D. by way of a three dimensional model: a background experience of felt safety, the mental representations of self-other relationships and the mentalization of experiences.
In a second part we examined and constructed the instruments to study this process from three perspectives: an independent researcher's, a patient's and a therapist's perspective. For the independent researcher's perspective we operationalized the three dimensions to be measured on the Object Relations Inventory (Blatt & Auerbach, 2003) with a newly constructed Felt Safety Scale, with the Differentiation Relatedness Scale (Blatt & Auerbach,2003), with the Reflective Functioning Scale (Fonagy et al.,1998) and with a Bion Grid Scale. For the patient's perspective we constructed the Leuven Psychotherapy Scale with items based on statements of patients themselves and by factor analysing the answers of 492 patients in hospitalization based therapy. For the therapist's perspective we used the newly developed Psychoanalytic Process Rating Scale (Stoker and Zevalkink,2005). We studied the reliability and divergent and convergent validity of each of these scales.
In a third part we studied the outcome of 70 P.D. patients in a psychoanalytically informed hospitalization at the U.C. Kortenberg during a year treatment and a year follow-up. We found the same significant improvement as in the Bateman & Fonagy (1999, 2001) outcome study with a continuing improvement after discharge. Furthermore we could delineate outcome trajectories, which revealed an impact of personality style. This corroborates the findings of Blatt (1992). We further examined the early drop-out group in comparison with the therapy group and found differences in client variables on hostility and vindictiveness and no significant difference in symptom outcome one year after admission.
In a fourth part we studied the process and its relation with outcome from three perspectives (researcher, patient, therapist) on 47 consecutively admitted patients who stayed long enough to study the process. From the three perspectives, a peculiar but coherent change over time was revealed pointing at a regression from three to six months and an improvement of most parameters in the subsequent period. We found a significant relation between the change of most process measures and the change of symptom and personality related measures. From the three perspectives, this relation was highly significant for the basic dimension of felt safety. The relation with the dimension about the integration of the mental representations of object relations was significant as well, the dimension of mentalization was related with personality change but not with symptom change.
In a fifth part we studied patterns of psychic change and found that patients with a similar severity of borderline pathology showed a different pattern of inner change according to their personality style. We found a stable pattern of inner change with a considerable gain from the treatment as it is, starting late but continuing after treatment, to be related with an introjective, more controlled, withdrawn and narcissistic personality style. A fluctuating pattern, with an easy desintegration of mentalization and a vulnerability to separations was related with an anaclitic personality style. Clinical implications are discussed.

4

5
Dankwoord
In de eerste plaats dank ik Professor Vertommen. Hij zal als promotor altijd een
voorbeeld voor mij blijven. Zelfs in de onmogelijkste momenten was hij steeds beschikbaar
en zag hij heldere wetenschappelijke oplossingen. Met zijn ruime onderzoekservaring hielp
hij het belangrijke van het onbelangrijke te onderscheiden, legde hij de juiste klemtonen
naargelang de fase van het onderzoek en hield hij een tijdslijn in de gaten voor het
proefschrift. Een betere en toegewijdere promotor kan ik me niet dromen. Professor
Corveleyn steunde me als copromotor, hij hielp me de brug maken tussen het empirische
onderzoek en het psychoanalytische werk, met geduld en precisie doorworstelde hij mijn
sneuvelteksten. Ik ben hem ook dankbaar de drijvende kracht te zijn van het onderzoek van
psychoanalytisch georiënteerde behandelingen. Yannic Verhaest complementeerde me op
veel terreinen. Als onderzoeksmedewerkster heeft ze zich op een uitzonderlijk nauwgezette en
vastberaden manier ingezet. Ze kon de grote groep patiënten die bereid waren mee te werken
gedurende twee jaar op een fijngevoelige manier motiveren en de gegevens van de meer dan
30 medewerkers bijhouden en mee verwerken. De inzet, nauwgezetheid en de
vooruitziendheid waarmee ze dit alles deed is bewonderenswaardig. Rob Stroobants was als
wetenschappelijk medewerker onvolprezen, hij zocht van in het begin mee naar de juiste
statistische verwerking van de vele gegevens, voerde tientallen analyses uit en met een bijna
oosterse onthechting trotseerde hij gedurende jaren mijn onwetendheid. Zonder hem zou dit
werk nooit tot stand zijn gekomen. Kristof Van Steelandt bood het nuchtere perspectief van
een pur sang onderzoeker.
Mijn grootste dank gaat naar de directie van het ziekenhuis, in het bijzonder naar
Professor Peuskens zonder wiens aanmoediging en daadwerkelijke ondersteuning dit projekt

6
niet mogelijk was. Ik dank ook Professor Verhaest die de vader is van het behandelmodel en
dit tot bloei bracht.
Het voltooien van dit project en het maken van dit proefschrift waren een les in
nederigheid en vertrouwen. Meer dan dertig actieve medewerkers namen een deeltaak op
zich. Alleen al het uittikken en scoren van de 450 Object Relations Inventory, slechts een deel
van het onderzoek, nam meer dan 10.000 uur in beslag. Dank zij de onbaatzuchtige hulp,
steun en niet aflatend vertrouwen van deze groep kon dit werk voltooid worden..
Ik ben de collega's van de beide onderzoeksgroepen dankbaar in het bijzonder dr.Dirk
Smits, dr.Laurence Claes, drs.Greet Geenen en Professor Patrick Luyten voor hun steun en
medewerking.
Mijn speciale dank gaat uit naar dr. Martens, mijn collega op de dienst en de collega's
die tijdens het onderzoek in opleiding waren en die waar nodig een deel van het werk
overnamen: dr. Kathleen Michiels, dr. Claudia Devin en dr.Els De Baerdemaker, die
daarenboven alles nalas in de laatste dagen.
Aan de procesbeordeling door de therapeuten namen bijna alle stafleden deel, de
groep is te groot om iedereen op te noemen: Erica Bourgois, Ingrid Demuynck, Chris Druyts,
Ingrid Janssens, Greet Poot, Michel Probst, Ann Schiemsky, Eric Tibau, Brigitte Vanden
Berghe. Marcel Franssen en Lieve Janssens behoorden daarenboven tot de vaste groep
scoorders. Ik dank Jasper Scheir en Edith Maes voor de vervanging.
13 .studenten maakten hun thesis in het kader van dit onderzoek en hielpen de last
dragen, verschillende van hen traden op als onafhankelijke scoorder: Ellen Wos, Tine Van
Brant, Nele Sauviller, Frede Van Hemelrijck, Katy Lelièvre, Sylvia Desimone, Annelies
Hoydonckx, Jasper Scheir, Caroline Put, Evelyne Peeters, Astrid Van Raemdonck, Karen Van
Winckel, Liesbet Weverbergh. Het was een stimulerende en aangename samnwerking.
Ik dank dr. Fuhr voor de taalcorrectie.

7
Mijn grootste dank en respect voor alle patiënten die op een of andere manier
meewerkten, aan een onderzoek dat voor hen los van hun individuele behandeling stond en
die de moeite bleven doen om achteraf en op eigen kosten te blijven meewerken aan het
follow-up onderzoek.
Last but not least dank ik voor de steun, de opleiding en het vertrouwen van
toponderzoekers als Professor. P. Fonagy (University College London), Professor S. Blatt
(Yale University), Professor J. Clarkin (Cornell University) die steeds met een inspirerende
geestdrift feed-back gaven en on-line beschikbaar bleven voor adviezen. De Research
Advisory Board van de International Psychoanalytical Association verleende een substantiele
beurs aan het project in 2001, die verlengd werd in 2002 en 2003. Dit werd aangevuld door
het bestuur van het U.C. en een bijdrage van de Lotto. Door de omvang en het tijdsintensief
werk was dit een kostelijk onderzoek en ik dank dan ook uitvoerig voor deze steun en dit
vertrouwen.
Het werk draag ik op aan Dominique. Zij was er altijd om met een warm hart bij te
springen: thuis, in het ziekenhuis en als een van de vaste scoorders en corrector bij de
eindtekst. Ik draag het ook op aan Lobke, Kasper, Zoe en Samuel. Voor hen was ik de laatste
twee jaar vaak maar een schim van wie ik had moeten zijn, ik heb te weinig gegeven en zeer
veel gekregen. Ik draag het ook op aan mijn moeder, die me heeft moeten missen en aan mijn
vader die fier zou geweest zijn over de prestatie.
Ik draag het werk ook op aan de medewerkers van de KLIPP in het UC te Kortenberg,
waar het project groeide, gedragen en uitgevoerd werd vanuit de gedeelde bezorgdheid een zo
goed mogelijk behandelmodel aan te bieden.
Het is onmogelijk om alle vrienden op te noemen die me hebben gesterkt met hun
interesse en steun. Ze weten dat ik hen in mijn hart draag en dankbaar ben. Ik dank tevens

8
alle collega's, en de medeleden van de initiatieven en verenigingen waar ik deel van uit maak
om met een mededogen te verdragen dat ik niet altijd even beschikbaar was.

9
Table of contents
Preface
Part 1: Psychic Change in Personality Disorders
Chapter 1: Three Dimensions in the Treatment of Personality Disorders on Psychoanalytic
Lines: Clinical and Theoretical Background and a design for Research 23
Chapter 2: Two Sides of Mentalization: Comparing the Bion and the Fonagy Approach 51
Part 2: Instruments to Measure Psychic Change in Personality Disorders
Chapter 3: Measuring the Psychoanalytic Process in Personality Disorders on the Object
Relations Inventory: the Perspective of the Independent Researcher 73
Chapter 4: Het Meten van Psychiche Verandering met de Leuvense Psychotherapie Schaal:
het Perspectief van de Patiënt 105
Chapter 5: Measuring the Psychoanalytic Process with the Psychoanalytic Process Rating
Scale: the Perspective of the Therapist 133

10
Part 3: Outcome, Outcome Trajectories and Drop-out
Chapter 6: Outcome and Outcome Trajectories of Personality Disordered Patients During and
After a Hospitalization on Analytic Lines 151
Chapter 7: Patient Attrition in a Psychoanalytically Informed Hospitalization-Based
Treatment for Personality Disorders 183
Part 4: The Relation between Process and Outcome
Chapter 8: The Kortenberg - Leuven Process Outcome Study on Patients with Personality
Disorders in Psychoanalytically Oriented Hospitalization 203
Part 5: Patterns of Psychic Change
Chapter 9: Patterns of Inner Change and Their Relation with Patient Characteristics and
Outcome in a Psychoanalytically Informed Treatment of Patients with Personality Disorders
247
Conclusion: Pointers in Psychoanalytically Informed Hospitalization-Based Treatment of
Personality Disorders 263
References 275

11
List of Abbreviations
BDI Beck Depression Inventory
BGS Bion Grid Scale
DRS Differentiation Relatedness Scale
GPS Global Personality Score
GSS Global Symptom Score
IIP Inventory of Interpersonal Problems
LPS Leuven Psychotherapy Scale
ORI Object Relations Inventory
PPRS Psychoanalytic Process Rating Scale
RFS Reflective Functioning Scale
SCID II Structured Clinical Interview of the DSM-III-R axis II
SCL Symptom Checklist
SHI Self Harm Inventory
STAI Spielberger State and Trait Anxiety Inventory
STAXI Spielberger State and Trait Anger Inventory
TEQ Traumatic Experiences Questionnaire


13
Preface
The psychoanalytically informed hospitalization-based treatment of personality
disorders at the University Centre St.-Joseph, Kortenberg exists since more than thirty years.
The programme was set up by Prof. Verhaest. It was the first such specialized centre for
personality disorders in Belgium, with the Viersprong (J.Jongerius) in Holland and the Cassel
Hospital (T.Main) in England as its counterparts. Since, thousands of patients have been
treated and hundreds of professionals have been trained at the setting. This was before the
DSM III, axis II (APA, 1980) category of personality disorders existed. The treatment focused
on neurotic and so-called pre-oedipal pathology, mostly patients with a borderline or a
neurotic condition in the psychoanalytic sense.
The program set up by Prof. Verhaest was a unique combination of therapeutic
community principles, psychoanalytic group therapy and a range of conjoint non-verbal
therapies within a highly structured frame. The underpinning concepts were mainly
Kernberg's (1975) ideas about split transference in borderline patients and Kohut's (1971)
ideas about the vulnerable self in narcissistic patients. Patients were seen as going through a
process in which holding (Winnicott, 1960), therapeutic alliance (Meissner, 1996) and the
working through of moments of separation-individuation (Mahler, 1968) played a great role.
Over the years, the teams of the wards were stable, staff carefully trained and a therapeutic
culture became firmly established. Convinced of the role of the therapeutic environment, a
first research project took place with the Ward Atmosphere Scale (Verhaest, 1982, 1983 a, b).
Gradually the focus of the treatment shifted towards an inner process in patients, which
is catalysed when they start to deal mentally with their experiences instead of acting them out.
We called this phenomenon ' interiorization' (verinnerlijking, Verinnerliching), based on the

14
ideas of the importance of an inner mental space which is not evident in personality disorders
(Winnicott, 1971; Bion, 1955; Meltzer, 1975). In consequence, by the end of the eighties Prof.
Verhaest started to evaluate the patient's process according to three simple categories: whether
the patients were able to speak, to discuss or to reflect. As such he was a fore-runner of the
Reflective Functioning concepts (Fonagy et al, 2002), which became the base of the current
approach of personality disorders (Bateman & Fonagy, 2004). Our group further studied the
transformation of experiences into something psychic, relying on Bion's theories on
mentalization (Vermote, 1994, 1995, 1996b,c, 1997b,1998a,b,1999, 2000b,c, 2001,
2002a,b,c,e, 2003a,b, 2004a,b, Kinet & Vermote, 2005). Non-verbal therapies and
experiences at the ward play a great role in these transformations of protomental phenomena.
Mentalization is studied from two sides in Chapter 2.
In team discussions we were often impressed by the effects of these inner psychic
changes and the way they took place within the setting. Given the lengthy treatments, we
wanted to be sure that this was not wishful thinking and set up a study about the relation
between the outcome five years after therapy and the going through such a mental process.
The results showed a positive relation and were presented at the first European Conference of
Personality Disorders (Vermote & Callens, 1994). However the methodology of this study
was weak as it was a retrospective study and we were lacking a clear definition and
operationalization of the process of inner change.
Integrating experiences and theoretic concepts, a three dimensional model of this
process grew. This model became a practical tool in organising the treatment, and in the
evaluation of the patients (Vermote, 1996a, 1997a, 2000a,b, 2002 a,b; Vermote & Callens,
1994; Vermote & Van Sina, 1998; Pieters & Vermote, 2002). This model is discussed at
length in Chapter 1.

15
Motivated by the research minded director of the hospital, Prof. Peuskens, we decided to
renew our empirical approach to the treatment. We were confronted again with the
considerable lack of instruments to measure a psychoanalytic process. As Spence (1993)
stated, psychoanalysis was situated where medicine was at the turn of the twentieth century,
with a lot of clinical wisdom in the soft clinical tissue of the practitioners, described in
innumerable publications but without a thorough empirical validation of it. Discontened with
the method of measuring the effects of our treatment with clinical symptom scales, we looked
for other ways of measuring what happens in the treatment. Prof. Vertommen of the Research
Group on Psychopathology and Psychodiagnostics of the Faculty of Psychology of the
Katholieke Universiteit Leuven, offered methodological advice. Finding no appropriate
instrument to measure the specificity of our treatment, we started constructing and validating
a scale that is specific for evaluation of the process in personality disordered patients in
psychoanalytically informed hospitalisation, the Leuven Psychotherapy Scale (see Chapter 4).
In 1999 the seminal paper of Bateman & Fonagy was published about an RCT on the
outcome of a psychodynamic oriented hospitalisation for borderlines, very much like ours.
The study demonstrated the effectiveness of such a treatment and resulted in a revaluation of
the psychoanalytic method in treating personality disorders (Sanderson, Swenson & Bohus,
2002). It was a proof that it was possible to study psychoanalytic treatment with standard
methods of current research. Contacting Prof. Fonagy resulted in taking part at the UCL-IPA
research training in 2000. In our trial to study the psychoanalytic process in personality
disorders, the idea grew of operationalizing the three-dimensional model on which we
anchored our treatment in the last years. Prof. Fonagy and Prof. Target (University College
London) and other renown researchers such as Prof. Blatt (Yale University) and Prof. Clarkin
(Cornell University) supported this idea and helped with this task which we imposed upon our
services. The operationalization of the treatment model is discussed in Chapter 3.

16
Knowing that the psychoanalytic process is such a vague notion we wanted to study it
from several sides hoping that as Shapiro stated 'finally the elephant would become obvious
for the blind man....' (Bucci, 1996). We developed, translated and validated measures to score
the process from three perspectives: the independent researcher's, the patient's and the
therapist's point of view (Chapters 3, 4, 5).
Given the recent evidence of the effectiveness of psychodynamic treatments in general
(Sandell et al., 2000; Perry et al., 1999; Leichsenring and Leibing, 2003; Leuzinger-Bohleber,
2002) and of the hospitalisation based treatment on analytic lines of P.D. patients in particular
(Bateman & Fonagy, 1999, 2001), we were not so much interested in studying the
effectiveness but decided to focus on the questions of 'how it works and for whom'.
The IPA research advisory board (Chair: Prof. Wallerstein) granted the project in 2001,
2002 & 2003 which was a substantial support. The directory board of the hospital offered the
extra support necessary for research assistance by Y. Verhaest. Given the close collaboration
with the Faculty of Psychology for methodological, statistical and theoretical support, the idea
of making a doctoral dissertation about the project grew. Prof. Vertommen accepted to be the
promotor and prof. Corveleyn to be the co-promotor of this work.
We decided to study the following clinically relevant questions:
1. A formulation of psychic change in the treatment process in three-dimensional model
and its implementation in the treatment of personality disorders (Chapters 1,2).
2. The first question is of course whether it is possible to measure the inner changes
focused on in the treatment (Chapters 3,4,5).
3. Knowing that the outcome of the treatment is not homogeneous and is for a great deal
effected by pre-treatment client variables, we wanted to know whether we could discern
various outcome trajectories and whether these were indeed related to pre-treatment variables.

17
In that case it would be possible to refine the indication for the treatment and to adapt the
treatment to the specificity of the patients, instead of offering an uniform treatment for all
(Chapter 6).
4. One of the major problems of psychotherapy with personality disorders is a high
drop-out rate. This is a real problem with figures in literature up to 60 % (Chiesa, 2000).
Therefore it is important to study this phenomenon and to look for variables which predict
such early stopping of the treatment (Chapter 7).
5. The main interest of the study is the process of inner change in the patients. Are we
right in basing our treatment on this inner change, or is it an illusion?
- Firstly, we want to study the change over time of all the process parameters. Is a
coherent process revealing itself from different perspectives? If so, does the results
corroborate some theories of inner change and falsify others? Is the three-dimensional model
of inner change in PD patients, on which the treatment is based, valid? Are there implications
for the technique of the treatment?
- Second, we want to know how changes in these process parameters are related with
changes in clinical symptoms, and more important with changes in personality related ways of
coping and relating. Which parameters of the process relate with these changes? Do we
enhance these facets of the process with our treatment or do we have to accentuate other
lines? It is for instance a debate since years in the psychotherapy of P.D. patients (Akthar,
1995), whether one should confront, interpret or rather be there in a supportive way, offering
corrective experiences. How much weight should one give to one approach or to another and
at what stage of the treatment? This is studied in chapter 8.
- Third, there are strong arguments, that not all patients have the same kind of
psychotherapeutic process. Is it possible to discern different patterns of the process of inner
change? If so, is this related to client variables? Are there different mechanisms of change for

18
different groups of clients and what are the implications for treatment? This is studied in
chapter 9.
Preliminary results of the study have been presented at several conferences (Vermote,
2000e, 2002d; Vermote, Vertommen, Corveleyn & Peuskens, 2002; Vermote, Vertommen,
Corveleyn, Verhaest & Peuskens, 2003; Vermote, Vertommen, Verhaest & Peuskens, 2004;
Vermote, Vertommen, Corveleyn et al., 2004 April). To situate the method of this process
study compared with other process studies, a scheme of Hill and Lambert (2004) is used
(table 1).
Table 1
Categorisation of the Kortenberg-Leuven Process-Outcome Study on Personality Disordered
Patients in Psychoanalytically Informed Hospitalization
1.Focus of evaluation patient
2.Aspect of the process inner changes of the patient
(psychoanalytic process)
3.Theoretical basis psychodynamic
4. Perspective of evaluation - nonparticipant observer
- patient
- therapist
5. Unit studied macroprocess unit: whole treatment
6. Type of measurement interval scales and nominal categories
7. Level of inference required inferential
8. Stimulus material to make judgment participation and combination of tape and
transcript
according to Hill and Lambert (2004)

19
Guide for the reader
Each chapter is written in a way that it can be read as an independent text. This causes
overlaps in the description of study groups, instruments, procedures and analysis of data. The
reader who intends reading the whole study, can easily skip parts that overlap. A short
indication of the content of each chapter follows for the reader who is interested in a part of
the study.
Chapter 1: The clinical and theoretical background of the psychoanalytically informed
hospitalization of personality disordered patients, with formulation of the research questions
and study design.
Chapter 2: A detailed psychoanalytical approach of one of the dimensions of the
treatment: mentalization.
Chapter 3 - 4- 5: The development and the examination of the reliability and validity of
model specific scales to measure intrapsychic changes. This might be of interest for
colleagues involved in the same kind of research.
Chapter 6-7: These chapters are about the outcome, outcome trajectories and drop-out of
psychoanalytically informed hospitalisation for personality disorders. It is of interest for
psychiatrists and psychotherapists, involved in treating personality disorders.
Chapter 8: This is the capital part of the study: the study of the relation of inner psychic
changes and changes in symptoms and personality related characteristics. The findings of this
chapter may have further clinical implications.
Chapter 9: This chapter is about patterns of inner change and their relation with patient's
characteristics.
Conclusion: This chapter is about the clinical implications of the findings of this study.

20

21
PART 1
PSYCHIC CHANGE IN PERSONALITY DISORDERS

22

23
Chapter 1
Three Dimensions in the Treatment of Personality Disorders on
Psychoanalytic Lines: Clinical and Theoretical Background and a
Design for Research
The Randomised Controlled Trial Outcome study of Bateman and Fonagy (1999, 2001),
showing the effectiveness of psychotherapy for borderline personality disorders on analytic
lines on different outcome measures, was an important argument to place psychoanalytic
oriented treatment of borderline personality disorders in the APA guidelines on the treatment
of borderline personality disorders (Sanderson et al., 2002). One of the most important
questions is, of course, whether, to what degree, and how these outcome changes correspond
with the inner changes at which a psychoanalytic informed treatment aims. As there is no
direct access to these inner changes of personality disordered patients, one has to rely on a
model to apprehend them. The aim of this article is to present the model the authors
operationalized to do so. The concrete clinical anchoring of this model will be described,
before entering into its empirical operationalization for a process-outcome study1.
1 Granted by the Research Advisory Board of the International Psychoanalytical
Association in 2001, 2002, 2003.
23

24
Personality Disorders
The problematic reaction of Personality Disordered (PD) patients to the psychoanalytic
situation has been studied since the beginning of psychoanalysis. Often seemingly good
candidates for psychoanalysis during the intake sessions, these patients reveal a far more
primitive level of psychic functioning once they are in therapy. This is characterised by not
tolerating separations, having trouble with the as-if character of the transference, and an
acting out of their psychic pain instead of symbolizing it. At the worst this may be coupled
with a malignant regression and self-destructive and suicidal behaviour. Since the 30s, many
concepts and names were applied to these patients. In 1975, Kernberg hypothesized a
personality structure or organization for this group of patients, the borderline personality
structure.
Aiming for an atheoretical and descriptive vantage point, the DSM III Committee (APA,
1980), allotted a specific diagnostic axis, called Axis II, to the personality disorders (PD),
consisting of eleven categories. This categorical diagnostic approach and the psychodynamic
structural approach relate to one another in that all Axis II categories correspond to the
borderline personality structure, except for the obsessive-compulsive and depressive
categories which belong to a neurotic personality structure (Kernberg, 1996).
24

25
Three Dimensions in the Psychoanalytic Approach of Personality Disorders
When applying the wide range of psychoanalytic theories to the clinical experience of
analytic treatment of personality disorders, three dimensions emerge as being of particular
importance: a background dimension of the experience of safety, a dimension of psychic
functioning or the mentalization of affects and experiences, and a dimension of object-
relational patterns. The first two dimensions are silent and non-problematic in neurotic
structures, which is not the case in personality disorders.
The background experience of safety is seen as a basic, generalized feeling of safety. It
is described by Sandler (1960) as 'a background feeling within the ego, a feeling which can be
referred to as one of safety or security. I want to stress the positive character of this feeling
(which need not, of course, be conscious). It is a feeling which bears the same relation to
anxiety as the positive body state of satiation and contentment bears to instinctual tension.
Genetically, this feeling must be a derivative of the earliest experiences of tension and
satisfaction. It is a feeling of well-being, a sort of ego-tone. It is more than the mere absence
of anxiety, and reflects, I believe, some fundamental quality of living matter which
distinguishes it from the inanimate. It is a quality of feeling which we can oppose to the affect
of anxiety, representing in a sense its polar opposite' (p. 352).
This inner feeling of safety is frail and often lacking in personality disorder patients,
hence their great sensitivity to even small changes in the therapeutic frame. It gives rise to a
subjective feeling of self, ‘a place within oneself to retreat and relax’ (Winnicott, 1963). It is
related to many other psychoanalytic concepts such as Sroufe’s ‘felt security’ (1996), Klein’s
‘good internal mother’ (1940), Kohut’s ‘primary mother’ (1971), Doi’s ‘amae’ (1989),
Balint’s ‘primary love’(1960), and Holmes’ ‘secure base’ (2001).
25

26
Constitutional factors and environmental factors play a capital role in mediating inner
and outer stimuli to preserve this basic feeling. Winnicott (1967) and Bion (1962) talk of
seeing oneself in the eyes of the other as the basis of such self-regulating mechanisms. In PD,
this ‘felt safety’ is a necessary canvas or matrix upon which object-relational patterns can be
integrated and mentalization can take place. For PD this felt safety is not permanent, they are
constantly checking and searching for it. In doing psychotherapy and psychoanalysis, it is
important to keep the degree of Felt Safety in mind. At moments of regression one must
‘secure’ these patients; otherwise they fall back on non-mentalizing mechanisms.
The dimension of mentalization relates to the psychic transformation of affects,
perceptions, and experiences by a psychic apparatus for thinking (Bion, 1962), which may be
compared with a kind of mental processing ‘machinery’ (Damasio, 2000). More poetically,
one could term it ‘the dreamer who dreams the dream’ (Grotstein, 2000). This psychic
processing is now commonly referred to as mentalization, a term which originated in France
in the 60s. In contrast to neurotic patients, many PD patients are characterized by a failing of
this automatic psychic processing. Rather than transforming and elaborating their emotional
experiences, they tend to evacuate them from the psyche by an evocation of sensorial
experiences such as self-harm, substance use, sexual and physical excitement, or by evoking
and controlling these feelings in others by projective identification. A facet of mentalization,
the interpretation of interpersonal behaviour in terms of psychic states, is conceptualized and
operationalized for research by Fonagy and colleagues (1991, 1993, 1994, 1995, 2000, 2002)
as Reflective Functioning.
The dimension of the object relations is the most prominent dimension in the
psychoanalytic conceptualisation of PD. In fact, O. Kernberg (1975) based his paradigmatic
theory of personality organization for the greatest part on this dimension. Inner objects are
nowadays conceptualized as patterns of mental representations, originating in the experience
26

27
of very early relationships and continuing to govern relationships with oneself and others.
These self-object-relational patterns are reflected in the transference and may show many
degrees of sophistication and integration. In PD there is a pathological splitting of the inner
objects. It is the aim of the psychoanalytic therapy to transform the inner objects in the
transference-countertransference toward greater differentiation and integration. Kernberg
(1975) advocates to do so by interpreting object-relational dyads in the transference in the
here and now of a session (Clarkin et al, 1999). A Kohutian approach, rather, aims at a
regression within the transference, offering corrective experiences in an attempt to create new
holding and soothing self-objects, which the patient was missing and so giving failed
development a new chance. Most theories and therapies about personality disorders, such as
Meissner (1988), Masterson (1993), Akthar (1992), Adler (1994), and Blatt and Behrends
(1987), integrate more or less both points of view and may be put on a spectrum ranging from
the Kernberg to the Kohut pole.
27

28
The Multilevel Psychic Functioning Within the Three Dimensions
Keeping these three dimensions in mind, one may conceive in each dimension a
spectrum going form undifferentiation to differentiation, implicating that the mind
simultaneously functions at several levels or layers along this spectrum. Many psychoanalytic
theories refer to this phenomenon. Ferenczi’s ‘confusion of tongues’ (1949) deals with the
problem of understanding someone who is functioning at a developmentally earlier level,
while one is listening at a more developed level. Balint’s ‘basic fault’ (1968) describes a layer
in primitive mental disorders which is characterised by a two-person relationship with an
intense preoccupation with primary love and difficulty grasping conventional language and
meaning. Bion’s (1962) conceptualisation of a psychotic and non-psychotic part in each
personality reflects the same multilevel functioning. The same approach is found in Matte-
Blanco’s ‘unconscious as infinite sets’ (1988), with levels going form pure undifferentiation
characterized by an interchangeability of elements and symmetry, as in mathematical infinity,
and levels of a much greater differentiation. The equivalent and pretend mode of psychic
functioning as conceptualized by Target and Fonagy (1996) are referring to the same
phenomenon of multilevel mental functioning.
Neuroscientific models of the mind point in the same direction, demonstrating that the
mind functions at several levels without there being necessarily an interaction between these
various levels. Emotions are, for instance, dealt with simultaneously by cortical cognitive
processing systems and amygdala-thalamus systems (Damasio, 2000; Kandel, 1999; Ledoux,
1998). The same was noted with regard to perceptions (Ramachandran, 1998) as well as
procedural and explicit memory systems (Ledoux, 1998). There are only partial interactions
between the levels, and they are marked by a different set of characteristics. At the lower
levels, for instance, there is no naming, no differentiation, a determination and an immediacy.
28

29
At higher levels the pace is slower, and a linking with verbal and autobiographical memory
elements takes place.
It is obviously too early and too complicated to link psychoanalytic concepts to such
neuroscientific findings, but we should at least question the widespread one-way
psychoanalytic model according to which somatic experiences and emotions are being
processed to still higher forms of symbolization. In any case, a multilevel approach makes it
easier to understand the clinical symptoms and the features of the inner psychic process in
PD. In treating PD, a permanent monitoring of the level of psychic functioning in the here and
now is fundamental to be sure to reach the patients at their level. It determines whether one
should intervene with securing, clarifying, containing, or with interpretation of meaning.
An Integrative Three Dimensional Model of Personality Disorders
In applying this multilevel approach to the three dimensions discussed, one may
distinguish a gradient from undifferentiation to differentiation in each dimension. At the basic
level, each of the three dimensions are undifferentiated and therefore also undifferentiated
from each other. This is what we see in regressive states and in primitive pathology:
borderline patients act in the here and now, without making a differentiation in the analytic
situation between past and present, between self and other, or between a real and a
transferential as-if relationship. Moreover, these patients have difficulty in naming and
differentiating affects, which are often experienced as a generalized and threatening feeling.
In the same vein, interventions at a symbolic level are experienced as real and concrete,
leading to confusion.
At a more differentiated level, the level at which people with neurotic structures are
typically functioning, the three dimensions are clearly separated. At this level, the dimension
29

30
of felt safety corresponds to a working alliance, and the object-relational dimension is
characterized by an as-if transference in which inner conflicts are repeated and may be
interpreted. In the mentalization dimension, at the same time, affects are mentalized and their
meaning may be analysed at a symbolic level and the model of the mind of self and others is
taken into account.
While the basic undifferentiated layers are prominent in PD, this does not preclude that
PD patients could function to a certain degree at the more differentiated levels as well. This
frail upper layer may cause these patients to be misleadingly perceived as neurotic in a first
encounter. However, after starting psychoanalytic therapy and regression occuring, they easily
shift to a functioning at the basic, undifferentiated level.
The Specificity of the Therapeutic Process in Personality Disorders
In PD it is not only important to discern different levels of psychic functioning in the
present state but to differentiate as well the levels at which the therapeutic process takes place
(Balint, 1968; Godfrind, 1993; Vermote, Vertommen, Corveleyn & Peuskens, 2002).
At the more differentiated levels, whole-object relationships result in a working
alliance and in an as-if transference in the therapy, where intrapsychic conflicts may be
repeated, interpreted and worked through. This corresponds to what can be called the classic
psychoanalytic process (Vaughan & Roose, 1995).
At the undifferentiated level there is a confluence of the three dimensions. At this level
the therapy progresses in a total situation in the here and now. The boundaries between patient
and therapist are fragile due to projective identification. The therapist is seen as a real figure,
and actions speak louder than words.
30

31
Figure 1. A three-dimensional Model of the Psychoanalytic Process in Personality Disordered
Patients.
1.Felt safety
working alliance
total situationfeeling of safety through
therapeutic frame and therapeutic relationshipin the here and now
3.Objectrelations
whole objectsas-if transference
part -objectsprojective identication
2.Mentalisation
symbolisationpretend mode
acting out ,evacuation
psychic equivalent mode
Mentalization of what is experienced is poor, resulting in a constant danger of acting
out. At this so-called basic level of the psychoanalytic process, the main therapeutic tools are
securing, support, containment of the intense emotions, and giving form to and naming the
undifferentiated mental experiences.
These two levels happen together and influence each other. Interpretations at the level of
the classic process may result in a greater anxiety and put the basic process again to the fore.
Then containment, support, and structure are vital before it is possible again to interpret at the
content level, or the level of differentiated transference phenomena.
Merging of the three dimensions with part-object relating in a total situation: Basic PA Process: growing feeling of safety, decrease of splitting and part-object relating and increase of mentalization
Classic PA Process: repetition, regression, working through of inner conflicts
diff
eren
tiatio
n
31

32
Clinical vignette
The following example illustrates how the installation of a background feeling of safety
by being contained was basic to the integration of the excessive splitting of a patient's inner
representations and her capacity to deal with her inner world at a more differentiated level.
Jane is an 18-year-old girl. Her parents got divorced when she was 14. One sister
makes her way in life, but her brother is a psychiatric patient and on drugs. Jane
displayed behavioural problems such as physical and verbal aggression, suicide
attempts and parasuicidal behaviour, drug abuse, and promiscuity. There was no
background of safety, and an undifferentiation of feelings, which were not mentalized.
Her mother was terminally ill at the time of admission to our setting.
The same picture was manifest in the beginning of her therapy. She was very
provocative, shouting at fellow patients and staff, and was absent for a great deal of the
time during the therapies. In the pre-therapy group she told about her intense feelings of
hate towards her mother, whom she did not want to see. This showed how she was
overwhelmed by her feelings, could not differentiate them nor accept ambivalence. Her
mental representations of significant others were diffuse and split. Her only way to deal
with these intense affects was to run away a few times from the therapeutic setting and
finally to commit another suicide attempt, for which we had to transfer her to a secluded
ward. There she was visited and comforted by several staff members.
The staff decided to give her another chance and to admit her again to the clinical
psychotherapy ward. This was not self-evident, and she was very grateful for this. Her
confidence in the staff and her felt safety gradually increased, and she began to show
some of her psychic pain and talk about it. Though she needed a great deal of external
structure, she relinquished her disturbing behaviour. A transformational moment
occurred when a team member accompanied her on a visit to her severely ill mother for
32

33
the first time after a long lapse. This was a new experience to her. The colleague helped
the mother and daughter in containing the very painful emotions, and contacts were
renewed before the mother died.
Gradually the patient got attached to the ward. She became reflective about her
inner world, developed a kind of empathy towards others, and started to talk and reflect
instead of getting lost in the acting-out and evacuation of her feelings. Her habitual way
of dressing with stiletto heels and transparent blouses became less provocative as she
became aware of the defensive side of her attitude and the effect it had on other patients.
In the non-verbal therapies she discovered an inner richness of which she had not been
aware.
At a more differentiated level she started to work through separation and loss. It
was hard for her to realize what she had been missing in her life so far. She could
integrate different aspects of the representations of her mother and mourn for her. She
was able to give in to her own vulnerability. She could relate this to her feelings towards
the therapist. Thinking about her transference towards the other members of the group
and their reactions towards her made her conscious about her identity as a young
woman. Sexuality had another meaning now than the evacuation of tension. At the end
of her therapy she behaved as a kind of senior patient, helping new patients in finding
out what the therapy was about. After dismissal she resumed her studies.
Structural Diagnosis and the Three Dimensional Approach of Psychic Functioning in
Personality Disorders
Personality disorders can be classified within this three-dimensional spectrum. From an
object-relational point of view, Kernberg (1996) made a distinction between high-level and
33

34
low-level borderline personality organizations. In low-level borderline cases there is more
splitting and a greater infusion of aggression in their object-relations than in high-level
borderline cases. Similar relations between categorical and structural diagnoses have been
made by Gabbard (1994) and Akhtar (1992). This comparative approach from an object-
relational point of view can be complemented with the two other discussed dimensions: felt
safety and mentalization. It enables us to discern different structures behind identical
symptom clusters, each having different prognoses and therapeutic needs, a differentiation
which is lacking in the descriptive categorical DSM approach, as is evidenced by three
examples.
The symptom cluster of attention-seeking, dramatic behaviour, dependency, and sexual
dysfunction may point towards a neurotic (hysterical) or a histrionic (borderline) structure.
The hysterical PD has a neurotic structure with a well-established background feeling of
safety and a capacity of tolerating separations, which is not the case for the borderline
histrionic structure. As mentalization is functional in the hysterical neurotic structure,
symptoms may have a symbolic meaning, reflecting intrapsychic conflicts often sexual in
nature. In contrast, sexuality in histrionic borderline cases rather serves as a concrete
evacuation of tension. The object-relations in the hysterical neurotic structure most often
reflect a typical triangular nature, while histrionic patients are characterized by dyadic part-
object-relations with many projective identifications and a lack of boundaries.
Behind the symptom cluster of social detachment and lack of interpersonal and intimate
relationships, one can discern either an avoidant PD with a high-level borderline personality
structure or a schizoid PD with a low-level borderline personality structure. The first group
lacks felt safety and scans the environment for signs of safety but does, nevertheless, succeed
in mentalizing their feelings and perceptions. Also, their object-relations are basically of a
whole-object nature – tolerating ambivalence. The low-level schizoid PD is different. There is
34

35
a basic distrust, a failing mentalization with a danger of micropsychosis, and an inner-
borderline world of split self-object-relations which is hidden behind the schizoid, detached
appearance.
The structural difference between the two kinds of narcissism is well known. There is a
difference between borderline narcissistic patients as described by Kernberg and narcissists as
described by Kohut (1966). This differentiation corresponds more or less to Rosenfeld’s
(1987) and Bateman’s (1998) distinction between thick- versus thin-skinned narcissists,
Masterson’s (1993) exhibitionistic versus closet narcissistic type, Gabbard’s (1989) oblivious
versus hypervigilant type, or Akthar’s (1992) overt versus covert type of narcissism. In all
these distinctions we again see a difference in the three dimensions. The second category of
narcissists has a high-level borderline personality structure and shows a lack of felt safety
which they try to repair by constantly looking for approval from others. They tend to have
whole-object-relational patterns and only slight problems in mentalization. In contrast, the
first group of narcissistic patients displays a false sense of safety in hiding behind a grandiose
self, with their object-relations being split and split parts being projected unto others with an
omnipotent control of others. Their failing mentalization easily results in narcissistic rage.
Malignant narcissism (Kernberg, 1984) is an even lower level of narcissistic borderline
organization close to antisocial personality disorders. In malignant narcissism, the background
feeling of safety is missing, resulting in the feeling of living in a dog-eat-dog world, with a
failing mentalization and an eager and dangerous acting-out of affects, rather than a
mentalization of these affects. The internal objects are split, resulting in an infusion of
aggression and a typical idealization of bad internal split objects – the internal mafia gang of
Rosenfeld (1987). When there is a failing Super-Ego as well, this malignant narcissism results
in an antisocial personality disorder.
35

36
Clinical vignette
Jane is an obese young woman of 20. She is the only daughter from a poor family
and ashamed about this. She is very socially withdrawn and was referred to the
secluded ward because of repetitious and severe suicide attempts with strangulation and
self-mutilation. As a child she was often ridiculed in school. But being intelligent and
having good marks at school, she was idealised by both her parents. The referring
psychiatrist suspected an Asperger syndrome, given Jane’s severe problems in contact,
her difficulty to talk about emotions, and her frequent use of stopgaps in conversation.
In the beginning of treatment the suicidal thread remained, resulting in one
attempt with a phone cable. Intrapsychic dynamics became clearer by her works at art
therapy, which were full of aggressive content. Music therapy showed her difficulty with
rhythm and with letting herself go in improvisation. For a long time she tended to be
very withdrawn during group psychoanalytic sessions but finally became less anxious,
and her felt safety increased. She started to talk about her inner world, which was
characterized by fantasies about grandiosity. The group members dealt with these issues
in a delicate way. She revealed herself as a covert, hypervigilant narcissist with
avoidant traits corresponding to a high-level borderline personality structure.
Instead of acting out undifferentiated feelings of distress by self-destructive
behavior, she could gradually experience and differentiate her inner feelings and reflect
upon them. For the first time in his life she experienced an intimate relationship with a
man. During the group sessions she allowed herself to experience feelings of rivalry
with a female group member. All this was new to her. It may have corresponded to new
internal objects. Typical oedipal transferential themes became evident in the group
therapy. She started training in the running and fitness group. The suicidal thoughts
disappeared, she felt very attached to the setting and several patients, felt stronger, and
36

37
decided to take up her studies again. She was very grateful for the changes she
experienced.
The Organization of an Inpatient Psychotherapy Setting of Personality Disorders Based on the
Three Dimensional Model of Intrapsychic Change in Personality Disorders
This three-dimensional model served as a mould in organizing the inpatient treatment of
personality disorders at the University Centre Kortenberg, Belgium (Vermote & Van Sina,
1998; Pieters & Vermote, 2002). The setting comprises two wards. One ward is organized as
a day hospital for two groups of eight patients, while the other offers inpatient psychotherapy
for four groups of eight patients. The stay is open-ended with one year as a maximum limit.
The two wards have a similar therapy programme consisting of one and a half hours of
psychoanalytic psychotherapy three times a week, three art therapy sessions a week, two
music therapy sessions a week, and three psychomotor therapy sessions a week. Moreover,
the wards offer sociotherapy, with community meetings and small group meetings, and
activities such as cooking and cleaning as well. On the whole, the setting does not differ all
that much from many others, except for the fact that our approach was tailored to the
psychodynamic three-dimensional model.
The Dimension of Felt Safety
The background experience of safety corresponds to the way patients ‘use’ (Winnicott,
1969) the security and the predictability of the therapeutic setting to feel secure. Ideally this
felt safety should be interiorized during therapy or enhance an already existing kernel of inner
37

38
safety. The feeling of safety is mainly guaranteed by a strong psychoanalytic frame. All
therapies take place within a frame of fixed appointments, and absences are interpreted
meticulously. Additionally, there is a well-considered frame of rules and agreements as far as
the goings-on at the ward are concerned. The rules are not open to discussion and relate to
matters such as physical violence, sexual contact with fellow patients, substance use, self-
harm, and parasuicidality. The information about these rules and the making of a contract on
these matters are part of the regular intake procedure prior to the treatment. When there is a
tendency to offend the rules, mostly due to an evacuation of intolerable feelings by action
instead of mentalizing them, patients are offered alternatives and support mainly by looking
together for the meaning of this behaviour. It is very seldom that people have to leave the
setting because of breaking the rules. Apart from these set rules, there are a number of
agreements which remain open to discussion and can be adjusted after consideration at the
patient-staff meeting. They concern matters such as entrance times, the location of the
smoking area, changes in the therapeutic programme, and so forth.
Within this predictable system, there is a 24-hour staff coverage for both the partially
and fully hospitalised patients. A number of staff members are available at all times in case
the patients should feel the need to talk, while a need for protection can be mediated by a
graded system of guidance (phone calls, being accompanied by a fellow patient, family
member or nurse).
The therapeutic frame is seen as a dynamic structure, which is always under attack,
given the psychodynamics of patients with PD who are mostly borderline patients. These
attacks are not dealt with in a repressive way. This strict frame is used, rather, in a transitional
way like Winnicott’s (1941) use of the spatula in children, meaning that the frame stands the
same for everyone, therapists and patients alike, but the very divergent reactions which this
evokes in the individual patients are used as food for therapeutic reflection. One could say
38

39
that safeguarding the therapeutic frame is fifty percent of the job of the staff. The transitional
attitude aims at keeping open the mental space and avoiding action-reaction sequences.
Giving the constant current of destructiveness, parasuicidality and suicidality in low-
level borderline patients, regular briefing of staff members is a prerequisite. Such briefings are
exercises in equilibrium, maintaining the balance between being protective on the one hand,
and guaranteeing an open, free and therapeutic atmosphere on the other. This is the primary
goal of the weekly patient-staff meeting. Topics under discussion at these meetings are the
dynamics and behaviour which undermine the frame, such as alcohol abuse, the formation of
small groups, sexual acting-out, splitting, malignant regression, gossiping, violating the group
therapy confidentiality, and absenteeism. The aim of these discussions is to interpret these
phenomena at the group level. The meetings focus on the therapeutic culture in an attempt to
safeguard a ‘good enough’ atmosphere at the ward. Working intensively with PD patients in
an inpatient treatment has the advantage that patients see their problems reflected in others but
the down side is that this causes a continuous need to prevent this milieu from becoming
toxic. This is the main reason why antisocial personality structures are contra-indicated for
admission. Their tendency to exploit the openness and vulnerability of other patients toxifies
the therapeutic atmosphere before the antisocial patient can even begin to regress, become
dependent, and gain from the treatment.
Another way to secure the feeling of safety consists in additional regular meetings
where non-personal and matter-of-fact scientific answers are supplied to problems or
questions submitted in writing and deposited in a box during a set of regular informative
meetings. The questions mostly relate to diagnoses, outcome, etiology, or genetics. These
open discussions are experienced as securing, as they make the therapeutic approach
predictable. The therapists do not partake in them, in order not to interfere with transference.
Moreover, the patient is informed individually about the extensive bimonthly evaluation of his
39

40
or her process, not by the therapist, but by the psychiatrist of the ward. Depending on the
stage and the needs of the patients, such feedback may be either more confronting or
supportive. We always make sure, however, that the feedback is given in an open, unsaturated
way so as to ensure that it does not preclude the on-going spontaneous process of the patient.
Family members are seen and given information by staff members other than the therapists.
As the danger of such a strong frame is a malignant regression, we discourage that
patients continue to discuss group dynamics and inner feelings with each other outside the
therapy sessions. Besides the regular therapy programme with an analytic frame in all
therapies, we offer an elective programme of ateliers or workshops. These workshops, such as
running, poetry, assertiveness training, philosophy, gardening, and drama, are optional and
consist of a limited number of sessions. In contrast to the regular therapy programme, there is
in these workshops no focus on the inner process, and the activities are for the larger part in
the hands of the patients themselves. These occupations are seen as a way of stimulating non-
problematic and active parts of the patients. Furthermore, a high degree of freedom within the
setting and a free circulation with the outpatient milieu greatly contribute to the feeling of
safety. An important step in this respect was the decision to eliminate the ‘isolation room’.
The frequent use of it at the time may be seen as an enactment of internal cruel objects, a
short-circuiting of a mental processing, a collusion with an archaic need for being held in the
literal sense, and the evocation of intense counter-transference feelings of insecurity in the
staff. In doing this we noted a very strong decline of malignant regression, and the need for
this isolation room totally disappeared from the therapeutic culture at the ward.
40

41
The Object-relational Dimension
The object-relational patterns are manifested in a transferential way in the group
psychoanalytic psychotherapy, group psychomotor psychotherapy, and group improvisation
during music therapy. The large and the small groups at the ward are the theatre of the
internal object-relations. At the differentiated or neurotic level, the intrapsychic conflicts
represented in the transferences may be linked with the past. A working through at this level,
however, takes place mostly at a later stage of the inpatient therapy or in the classic
psychoanalysis or individual psychoanalytic psychotherapy, in which 80% of the patients
engage after their inpatient therapy. During the inpatient therapy, however, most patients
function prominently at more undifferentiated levels, and dealing with the transference
consists in the first place of containing the multiple transferences, making patients aware of
them and giving form to them. In addition, many PD patients have to experience something in
reality first, before they can start to think about it. They have to put it on the stage so to speak.
What may be seen as an acting out of transference in patients functioning at a differentiated
level is at this undifferentiated level an ‘acting through‘ (Van Sina, 1993).
In the same vein the setting also offers new experiences in a relatively safe environment
which may give rise to new inner objects. Defences such as splitting groups and teams, denial,
and different relational patterns reflecting rivalry, envy, authority conflict, symbiosis, lack of
structure and boundaries, and repeating abuse, all become manifest during the stay. The
resulting counter-transference is again a major therapeutic tool in recognizing and handling
this. Intervision and supervision groups help to handle the transference-countertransference
matrix. Such groups exist at staff level but also at the level of the analysts and analytic
therapists and at the level of the nurses.
41

42
The Mentalization Dimension
The mentalization dimension is a major concern in the treatment. Patients functioning at
an undifferentiated level are not in contact with their inner world and have no, or poor,
introspective capacities at the onset. From a psychoanalytic point of view, these patients
function for a great deal in a so-called operative (Marty, 1963) or two-dimensional modus
(Meltzer, 1967) or psychotic part (Bion, 1962), meaning that these experiences are evacuated
by creating sensorial excitations instead of being elaborated internally.
Such short-circuiting of psychic processing may consist in alcohol or drug abuse, eating
disorders, excitation by sexual behaviour, and self-mutilation with a view to instant-tension
reduction. Patients are often hard to reach when caught in this modus, as they find themselves
on another wavelength of communication. In psychoanalysis this is conceptualized as the so-
called Bionian beta-elements (Bion, 1962). As this condition is often marked by bodily
excitations, many of these behaviours can be seen in the context of Tustin’s autistic objects
(1986).
For many patients, finding a way out of this mode of psychic functioning is the primary
goal at the beginning of the treatment. We therefore instituted a pre-therapy group where
these types of short-circuiting in the psychic processing are addressed. In these meetings we
are concerned with what happens between the moment of an act and the emotions prior to the
act, in such a way as to attempt to open a psychic space. In this group, the anxiety level is kept
low by avoiding transference interpretations and maintaining a supportive stance. When
patients are in touch with their inner world and get used to talking and associating about their
feelings and thoughts without relying on acting-out behaviour to reduce psychic tension, they
are ready to commence the proper psychoanalytic group therapy.
42

43
In our experience, the non-verbal therapies play a capital role in catalysing the process
of mentalization. It is more difficult to block inner feelings during the psychomotor, music
and art therapies than in the verbal therapies. Patients are confronted with, and often surprised
by, what they find themselves doing and feeling. The non-verbal therapists are very attentive
to these expressions of the patients’ inner world and try to sensitize the patients to this aspect
of their process by giving authentic feedback in the language of their therapeutic medium and
by avoiding at all costs rationalizing via pseudo-analytic interpretations. Instead, they search
and help to create psychic moulds which may shape and catalyse psychic processing. These
moulds can be tangible objects, words, or even a melody. Patients bring these inner
experiences of themselves or other group members to the group psychoanalysis, where the
inner dynamics are further elaborated. At a staff level the different aspects of the inner
process are brought together in regular patient evaluations, where there is room to discuss
staff counter-transferences as well as the patient’s socio-familial dimension.
In addition to fostering the psychic processing of affects and perceptions which is
predominantly based on Bion’s vision of mentalization, special attention is given to another
side of mentalization as well. This latter interpretation of mentalization is conceptualized by
Fonagy and colleagues (Fonagy & Target, 1996; Fonagy, Gergely, Jurist & Target, 2002) as
the way patients consciously and actively reflect on their behaviour and relationships. Such a
therapeutic approach can be done by nurses, trained in this Reflective Functioning approach.
It is a guideline in the brief and supportive talks with the patients, as well as during the small-
group meetings. The very concrete and shared experiences at the ward actually offer good
occasion to confront different ways of reflecting. This is an essential part of the
psychoanalytic group therapy as well, where patients are systematically put in touch with
what happens in the minds of others and with different ways of interpreting shared
experiences.
43

44
Clinical vignette
Olga is 19 years old and was holding her father’s hands during the intake. She
hardly spoke at all and was trembling heavily. The father indicated that his daughter
was suffering from a neurological condition, recently aggravated by symptoms of
paralysis, for which she needed a wheel chair from time to time. Neurological exams
were, however, near to normal. The reasonable hypothesis of the father was that his
daughter was functioning far above her intellectual capacities and ultimately suffered a
mental breakdown. She herself expressed in a hardly audible way that she felt very
anxious and empty, and had suicidal ideation but never went as far as an attempt.
Although she was described by her father as a mild mentally retarded young woman and
she behaved in ways congruent with that description, there were reasons to see her as
an immature girl with histrionic personality traits.
In the setting she relied much on the support and took profit from numerous
individual sessions with the group nurse. This made her feel safe and less anxious.
During psychomotor therapy she experimented with her body image, which led to her
feeling more relaxed. The psychoanalytic group sessions centred on object relational
issues. Gradually she succeeded in differentiating herself from other group members
and in taking a more autonomous stance towards her parents. Her somatic complaints
about paresis disappeared after some months, probably by the fact that she could deal
now with her inner experiences at a mental level.
At another level, sexual themes surfaced in therapies such as the group therapy
and the art therapy, and she was able to talk about and reflect upon these issues. Finally
she engaged in a relationship with a man outside the setting. She left the setting at eight
months to take up her studies again, but continued to attend ambulant psychotherapy
44

45
where oedipal issues and the complex relationship towards her father could be taken up.
The relationship with a man, years older than she is, can be seen as an acting out of
these conflicts but now at a more neurotic level.
Research on the Psychoanalytically-Informed Treatment of Personality Disordered Patients
Two Questions
In an RCT Bateman and Fonagy (1999, 2001) showed the effectiveness of
psychoanalytically informed hospitalization-based treatment in a selected group of borderline
patients. Two further questions are of particular clinical interest for the treatment under
discussion. For whom of a larger group of personality disorders with a borderline personality
organisation does this kind of treatment work and how does it work? The answer to the first
question is an outcome study with a naturalistic design, the answer to the second question is a
process-outcome study.
The Development of Model Specific Instruments to Measure the Process of Psychic Change in
Psychoanalytically Informed Hospitalization Based Treatment of P.D.
The process of psychic change in the psychoanalytically informed treatment of
persoality disorder being defined in this chapter by a three dimensional model of change, we
further developed and validated model specific measures to assess it. As the process is only
measurable in an indirect way, it is indicated to assess it from different perspectives.
45

46
A first perspective is from a independent researcher's point of view and based on the
three-dimensional theoretical model. To measure the dimension of felt safety, we constructed
a Felt Safety Scale, for the dimension of the object relations we translated and the
Differentiation Relatedness Scale (Blatt et al., 1996), and for the dimension of mentalization
we translated the Reflective Functioning Scale (Fonagy et al., 1998) about the capacity to
perceive ones own and others' actions, thoughts and feelings in terms of mental states and we
constructed a Bion Grid Scale to measure the degree of creative exploration and mental
transformation of experiences. These scales were manualised to be assessed on the Object
Relation Inventory (Blatt, Stayner, Auerbach& Behrends, 1996). The scales are presented and
the reliability and the validity of the scales is studied and discussed in Chapter 3.
A second perspective is that from the patients. We found no instrument that was
appropriate to measure the inner changes during a psychoanalytically informed hospitalization
based treatment for personality disordered patients, therefore we decided to construct a scale
that was not theory driven, but derived from patient statements about their experience of inner
change during such a treatment. The Leuven Psychotherapy Scale is presented and its
reliabilty and validity is discussed in Chapter 4.
A third perspective is that from the therapists, we use of a newly developed scale by the
Dutch Institute of Psychoanalysis: the Psychoanalytic Process Rating Scale (Beenen &
Stoker, 2001; Stoker & Zevalkink, 2005 ). This scale and the examination of its realibility and
validity is discussed in Chapter 5.
46

47
The Outcome and Process-Outcome Study
Participants.
We asked all patients admitted in the period between 23 June 2001 and 15 July 2002 to
the setting for psychoanalytically informed hospitalisation based treatment ( inpatient and day
hospital) at the University Centre, Kortenberg to take part at the study. We studied the
outcome trajectories of this large group. We further delineated a group that stayed at least
nine moths to have enough moments to study the process, this group could also be compared
with the treatment group of the Bateman and Fonagy 1999 RCT.
Instruments
Outcome measures.
To measure the outcome we used the following clinical symptom outcome measures: the
Symptom Checklist 90 (SCL-90) (Derogatis, 1977), translated by Arrindell & Ettema (1993),
the Spielberger State-Trait Anxiety Inventory (Spielberger et al., 1970) translated by Van der
Ploeg et al. (1980), the Spielberger State-Trait Anger Inventory (Spielberger et al., 1980)
translated by Van der Ploeg et al. (1982), the Beck Depression Inventory (1961) translated by
Bouman et al. (1985). As self-harm and suicidality and para-suicidality are important
symptoms of patients with a borderline personality organisation, we used the Sansone Self
Harm Inventory ( Sansone et al., 1998).
47

48
As we expect the treatment to influence less volatile personality characteristics, we
assessed personality related measures as well. We used the questionnaire of the Structured
Clinical Interview for DSM-III-R Axis II disorders SCID-II (Spitzer & Williams,1985), which
has been translated in Dutch and adapted for DSM-IV, by Weertman, Arntz & Kerkhofs
(2000).
As this gives no indication of the severity of the personality disorder, we used the
Inventory of Personality Organization (Lenzenweger, Kernberg, Clarkin, Foelsch, 2001;
Normandin et al., 2002). The IPO is a self-report questionnaire used to assess a structural
diagnosis according to the model of Kernberg on borderline personality organization. The
reliability and validity of the IPO translated in Dutch was demonstrated by Vermote, Maes,
Vertommen et al. (2003) and Vermote, Smits, Claes & Vertommen (2005). We further used
the Inventory of Interpersonal Problems (IIP), circumflex version (Alden et al., 1990) which
measures interpersonal functioning. As traumatic experiences are common in personality
disorders, we employed the Traumatic Experience Checklist (TEC, Nijenhuis, Van der Hart &
Kruger, 2002) to measure this aspect of the patient characteristics. All these measures have a
well established reliability and validity.
Process measures.
See measures discussed in 'The development of model specific instruments to measure
the process of psychic change in psychoanalytically informed hospitalization based treatment
of P.D.'
48

49
Procedure
We assessed the process measures and the clinical syptom outcome measures, every
three months. The personality related measures were measured on admission and discharge.
In follow-up, all measures were assessed three months and one year after discharge.
Answer of the questions
Whether it works : The change over time of the outcome measures is studied for the
therapy group with analysis of variance, repeated measures and compared with the therapy
group of the Bateman-Fonagy 1999 RCT study (Chapter 6).
For whom it works: We conducted a Trajectory analysis with Traj Proc on the large
group of all admitted patients and examined whether the trajectory groups are related with
patient varibales by an analysis of variance and subsequent post-hoc tests (Chapter 6). We
further studied the characteristics of the drop-out group of this large group by comparing
their patient variables with those of the therapy group with a t-test and by studying differences
in early process and in outcome change with a two-way analysis of variance (Chapter 7).
How it works:
1. We studied the process of psychic change measured from three perspectives: by an
independent resarcher (with the four ORI measures), by a patient self-report (the Leuven
Psychotherapy Scale) , by the therapists (the Psychoanalytic Process Rating Scale). Analysis
of variance, repeated measures (Chapter 8).
49

50
2. We study the relation between outcome changes and the process as measured form
three perspectives (ORI-measures, LPS, PPRS). First the outcome scores were reduced in a
Global Symptom Score and a Global Personality Score by Principal Component Analysis and
then the change over time of these global scores were related to the change of the various
process easures by an analysis of variance (Chapter 8).
3. Patterns of the psychoanalytic process were exmined with a cluster analysis and the
relation of the the clusters with patient characteristics were examined with a t-test and
analysis of variance (Chapter 9).
Conclusion
The authors present an integration of the major psychodynamic theories on personality
disorders in a three-dimensional model. The clinical anchoring of this model is discussed and
it is shown how a setting for treatment of personality disorders on analytic lines was organised
on the basis of this model. The effectiveness of such a setting having been shown in RCT
studies, the authors wanted to study the intrapsychic changes happening during the
psychoanalytic-informed treatment of personality disorders and the relationships among those
changes and clinical and personality outcomes. The advantage of an inpatient treatment is that
it was possible to assess process and outcome without being disruptive to the treatment and to
collect data on a larger scale, making it possible to analyse these data with parametric
statistical methods. The authors discuss the operationalisation of the three-dimensional model
of psychic change in personality disorders and the research design of an outcome and process-
outcome study on a psychoanalytically informed hospitalization-based treatment of patients
with personality disorders.
50

51
Chapter 2
Two Sides of Mentalization: Comparing the Bion and the Fonagy
Approach
Mentalization is a powerful concept. One may even see psychoanalysis as ‘mentalizing
in the presence of another’ (Green, 2001). It can be defined as the psychic processing of
experiences from the outer and the inner world, such as drives, perceptions, impressions,
sensations, feelings, emotions, fantasies, thoughts, acts. These transformations may occur at
an unconscious, a preconscious and a conscious level.
The term mentalization is of French origin and linked to the work of Marty and De
M’Uzan (1963) with psychosomatic patients, in whom they supposed a lack in the symbolic
elaboration of affects. These ideas were integrated with Bion’s concepts on thinking by Green
(1973) and further extended by Marty (1991) and Luquet (1981, 2002). The French approach
of mentalization is very well elaborated in the study of Lecours and Bouchard (1997).
In the present text, we focus on the two main Anglo-Saxon approaches of mentalization
First we concentrate on the evolution of Bion’s theory of thinking and how this is reflected in
his Grid, in which he categorizes the psychic elements according to their use (x-axis) and
genesis (y-axis). Then we will compare this approach with Fonagy’s and colleagues approach
of mentalization.
51

52
The Various Stages in Bion’s Theory of Thinking
Bion’s Basic Model of Thinking
Freud (1911) based his psychoanalytic theory of thinking on the existence of a reality
principle which implies a latency period between impulse and action. In this latency period he
distinguishes notation, attention, inquiry and action. Bion replicated these categories in the x-
axis of his GRID. The y–axis of this Grid shows the genesis of thoughts according to Bion
(1957,1962a). Bion elaborated Klein’s (1932) concept of phantasy, an unconscious and
continuous ‘mental processing’, in a way which is close to English philosophical tradition.
Sensorial data (beta-elements) are transformed into impressions with a budding psychic nature
(alpha elements). These alpha-elements correspond to Locke’s (1698/1975) sensations.
Proust’s ‘le goût des madeleines’2 for instance, is such a sensation or alpha-element, and is
clearly more than a mere sensory perception (Anzieu, 1993). These alpha-elements are linked
in a spontaneous, associative process just as they are in Hume’s vision of thinking
(1739/1985, p.153): ‘thinking is a custom before reflection’. Alpha-elements are contained in
pre-existing patterns or pre-conceptions, similar to Berkeley’s pre-existing patterns
(1710/1993). The spontaneous, associative process results in more complex and abstract
elements.
However, this automatic, creative process is not yet thinking, and at this point this
theory shifts to the psychoanalytical field. Thinking emerges when these elements are used to
represent something, which is not present. A thought for a no-thing as Bion (1962a) puts it.
This thinking process needs an environment to help bearing the affects and to provide
thoughts to contain them (Bion (1962a). Thoughts are there before thinking. Bion’s
2 Madeleines are an old fashioned type of French cookies, often served with champaign.
52

53
containment model of thinking is widespread and became one of the watersheds of
contemporary psychoanalysis.
Bion’s Second Model of Thinking: Thinking and Transformation in Knowledge
In 1963 Bion added a new dimension to his model of thinking by presenting it as a
process of transformation of elements. A spontaneous dialectical movement between a mental
state in which the inner and the outer reality are perceived as phenomena without cohesion
(Ps) shifts into a mental state in which some order and relatedness between phenomena
emerge (D). The shift from PS to D is occasioned by the emergence of a ‘selected fact’, which
creates order in the complexity (Bion, 1962b, p.72). There is a constant oscillation between
PS and D, generating new selected facts. Bion’s model of creativity and thinking is rooted in
this PS-D oscillation and not in the depressive position as in Klein’s and Segal’s models. To
Bion (1962b, 1963a) this Ps-D shift is not a cognitive but an emotional experience.
In this second model of thinking he relies more on Plato and on Kant than on the
empiricist philosophers as in his first model. He sees the elements of a session as related to an
unknowable psychoanalytic object, a Kantian ‘ Thing in itself’ (Kant, 1781/1929) or O, an
invariant which expresses itself in a flow of ever changing manifest psychic elements. Bion
(1965) suggests that the analyst ought to desaturate his thinking in the sessions as much as
possible trying to ‘see, feel, intuit’ this O. He thus propagates an attitude marked by a formal
regression of thought (De M’Uzan, 1989 and Botella e.a., 2001), a radical free-floating
attention to enhance transformations in K of O.
53

54
Bion’s Third Model of Thinking: Thinking and Transformation in O
Bion became increasingly convinced of the impossibility of getting in touch with O even
when K is desaturated as much as possible This is due to the fact that thinking is originally
too utilitarian and sensorial to get through to O. Bion (1965, p.144) saw thinking as a curtain
of illusions, a protective coat of lies (Bion, 1965, p.129). In ‘ Transformations’ Bion (1965)
tried to transcend the limitations of the senses with a mathematical and formulaic approach,
but did not succeed. Therefore he decided to concentrate on O itself, instead of focusing on
the continuously changing phenomena which are expressions of the transformations of O.
This contact with the O of a session must be found in the void between two thoughts, when
imagination fails, when there is no thought to fit it. In Bion’s terms one has to become O
instead of trying to know it. He saw psychic growth as being achieved through
transformations in O while knowing about growth happens through transformations in K
(Bion, 1965, p.156).
An example: The patient suspects that she was sexually abused as a child, but does
not remember it. She has been in analysis for nearly two years, without much
regression. In line with Bion’s recommendations the analyst refrains to concentrate on
his free associations and logical thinking, what Bion calls an act of faith. A period
comes about when the patient is scared during the sessions and experiences her body as
if it were poisoned. She develops some kind of asthmatic reaction and needs to consult a
doctor. Then she introduces the image of a ramshackle, flaking wall with electric wires,
the insulation of which has been pulverized, lying beneath it. The image happens within
her, not in the analyst. The catastrophic feelings present in the sessions for patient and
analyst, were transformed in an image, but this transformation in O happens in her. It is
54

55
an emotional experience, a becoming. This is different from the usual technique where
the analyst understands (a transformation in K) and communicates this as an
interpretation.
Applications of Bion’s Theory of Thinking, as Expressed in His GRID
Bion’s GRID as a Map
Bion conceived his Grid as a device for observation, to be open to psychic elements to
come or yet unknown. We present this Grid as a map to explore the landscape of
mentalization, this way we are able to make a link between zones of mentalizing and
psychopathology. Three zones of psychic functioning can be discerned: a zone of unmediated
elements, a closed zone and an open zone. As thinking is a multilayered phenomenon, it spans
several zones of the GRID simultaneously. For reasons of clarity, the three zones will be
discussed separately.
The first zone or the A-row consists of unmediated beta-elements. These elements are
unknowable and fit for evacuation or A6 (Bion, 1963a), which makes them manifest. Such an
evacuation can take place by some forms of projective identification or by generating surface
experiences such as self-cutting, excessive eating or dieting, vomiting, some kinds of physical
exercise and sexual activity, drugs, alcohol. These actions (A6) are a substitute for thought.
They have to be repeated over and over to evade intolerable mental states. Such A6 elements
are excessively present in borderline pathology.
55

56
Figure 1. Three zones in the GRID: the zone of the unmentalized elements (black), the closed
zone (dark grey) and the open zone (light grey).
Definitory
hypotheses
1
ψ
2
notation
3
attention
4
inquiry
5
action
6
...n
A βelements
A1
A2
A6
B αelements
B1
B2
B3
B4
B5
B6
...Bn
C Dream
thoughts...
C1
C2
C3
C4
C5
C6
...Cn
D pre-concep
D1
D2
D3
D4
D5
D6
...Dn
E conception
E1
E2
E3
E4
E5
E6
...En
F concept
F1
F2
F3
F4
F5
F6
...Fn
G Scientific
deductive
G2
H algebraic
calculus
In contrast, A1-A2 unmediated elements remain in the darkness of the unthought reality,
for instance after trauma. These non-mediated elements can be seen as deposited in crypts
within the personality (Wilgowicz, 1991). It is often ‘après-coup’, catalysed by a sensorial
experience that a patient may bump up against them and have a flashback.
Some patients with a borderline organization, are characterised by unmentalized A1 and
A2 elements rather than by A6 elements. They have less symptoms and are more quiet. They
56

57
function to a large extent in two-dimensional way (Meltzer, 1975), displaying autistiform
traits. These patients seem to suffer from a developmental lack in mentalization, which may
be compensated by intellectual skills.
In psychosis not only psychic experiences but also fragmented parts of the ego and ego-
functions such as seeing, hearing are evacuated by projective identification (A6), forming A-
row elements which Bion named ‘ bizarre objects’, resulting in hallucinations (Bion, 1962a,
p.25). Delusions are formed by an agglomeration of beta-elements and debris of fragmented
thoughts and precursors of thoughts. This may function as a beta-screen (A2), which
precludes a further processing of emotional experiences.
An example of the presence of such a beta-screen in psychosis is a patient giving the
same stereotypical delusional explanation of his world, session after session for years. This
explanation has turned into an artificial, sterile thing to ward off the psychic processing of
emotional experiences
A-row or beta-elements are not as excessively and disturbingly evacuated in neurotic
patients as in borderline patients, but unthought experiences in neurotic patients are a point of
attention in contemporary psychoanalysis (Bollas, 1987).
In all other rows than the A- row of the Grid, the elements have a psychic quality.
Column 1 & 2 is called the closed zone, because these elements are used to restrain the
thinking process and avoid psychic pain by fixating the experience in a definition (column 1)
or by denying it (column 2). This is characteristic for narcissistic patients, the price to pay
being an emotional poverty and a lack of creative functioning. In sociopathy there is often a
combination of these elements with A6 elements.
For instance, a sociopath hiding behind the fantasy of standing at the helm of a ship that
has to be kept in balance during a storm. Every emotional situation, which this patient has to
57

58
deal with, is closed down and reduced to this static fantasy (C2-E2). However when he is
frustrated too much, he becomes violent (A6).
The third zone consists of columns 3, 4, 5, 6 except A6. These are the elements with a
psychic nature which emerge spontaneously and go with the experience of the opening up of a
mental space. This open psychic processing is a kind of creative preconscious process, with
an automatic, spontaneous generation of psychic elements. Psychopathology in the open zone
is characterised by a diminished presence of elements of columns 3, 4, 5 and a diminished
fluidity of the shifts between elements in this zone. This is similar to what Green (2000)
described as the ‘central phobic position'. Patients in whom the problem merely centres on the
fluidity, which is the case in high-level borderlines and neurotic patients, are good candidates
for psychoanalytic work (Van Lysebeth, 2001).
According to Bion (1957) patients with personality disorders and psychosis feel more
pain when entering the open zone and have a tendency to withdraw into the closed zone or to
attack emerging and existing links and to lapse into the A-row level. This is the therapeutic
dilemma in low-level borderline patients and psychotics (Vermote, 2002a).
Some Implications for Treatment
Layers of the psychoanalytic process.
The psychoanalytic process differs according to the zones in which the transformations
of the elements take place. In the unmentalised zone (the A-row level), the elements are
distorted by projective transformation and by transformation in hallucinosis (Bion, 1965) and
remain concrete, unmentalized experiences in the here and now. Mental space is lacking,
actions speak louder than words and boundaries fuse. Containment, basic safety and giving
58

59
meaning are fundamental at this level – the so-called ‘basic psychoanalytic process’ (Balint,
1968, Godfrind, 1993) which is always part of the psychoanalytic process, but especially
important in personality disorders and psychotics.
In the open-zone the elements reflect the internal objects with hardly any deformation in
the transference, hence Bion’s term ‘Rigid motion transformations’ (Bion, 1965,p.19). The
resulting as-if character of the transference makes it suitable for interpretations. This is the
level of the ‘classic psychoanalytic process’.
Elements at the level of the closed zone (columns 1 and 2) are fixed in protective
organizations, leading to an impasse in the process as Steiner (1993) described.
Wavelengths of communication and action.
Communications can take place at different wavelengths (Lucas, 1993), according to the
zone of psychic functioning. The same flag may have different ‘charges’. At the A-row level
for instance, communication does not serve to transfer meaning, but to do something, to evoke
something in someone else, to get rid of something. This kind of communication is for the
greater part exempt of symbolic meaning. Patients at this level tend to listen less to what the
therapist is communicating than to his voice, to the music of the words, scanning the
environment for safety.
The actions may reflect these different wavelengths: actions in A6 are substitutes for
thought, actions in B6 are a prelude to thought (Bion, 1977, p.7) and acting-out is in relation
to a repressed unconscious conflict (C6 for instance). Bion’s Grid makes it possible to
differentiate the level at which an act takes place.
59

60
Kinds of repetition-compulsion.
Repetition-compulsion has a different character according to the zone to which the
elements belong. Unconscious elements from the open zone may then be repeated till they
become conscious (for instance mental representations repeated in transference). This is in
contrast with elements from the unmentalized zone, which may be repeated until a form is
found to contain them, thus allowing them to be mentalized. This repetition may take place by
evacuating and controlling them in someone else through projective identification, a
characteristic of the relationships and transference of many borderline patients.
Fonagy’s Various Concepts of Mentalization
Reflective Functioning
Although both Fonagy’s Reflective Functioning and Bion’s Thinking go by the name of
mentalization, both concepts have different origins. The concept of ‘Reflective Functioning’
is grounded in attachment research, and (Fonagy, Steele et al.,1991, 1993; Fonagy, Steele,
Steele et al., 1994, 1995; Fonagy & Target, 2000; Fonagy, Gergely et al., 2002). Fonagy and
colleagues' (1991) RF is an elaboration of Main’s concept of ‘metacognitive monitoring’. It is
the capacity to perceive ones own and others’ actions in terms of mental states such as
feelings, beliefs, intentions, wishes (Fonagy & Target, 1996).
Fonagy and colleagues developed a scale, to assess the RF on the AAI and found that
the RF of future parents was predictive of the attachment of the children to their parents
(Fonagy, Steele et al., 1991,1993). Attachment patterns are not copied mechanically by
children, but a child assimilates its parents’ psychical dealing with relationships (Fonagy &
60

61
Target, 1996b). The link between the attachment patterns and RF is merely indirect, both
being rooted in a common base, namely security (Fonagy et al., 2002).
Target and Fonagy (1996) elaborated this concept of RF from the point of view of
developmental psychology. A child initially functions in the equivalent mode: it makes no
difference between its experience of the world and how the world really is. In the pretend
mode the inner and outer world are gradually differentiated but the inner world is experienced
as having no connection with the outer world. At the age of 3-4 the child integrates the two
modes in the reflective mode. This process seems to be impeded when the environment fails
to see the child as an intentional being. This is seen in borderline patients who remain stuck in
their inner turmoil and remain functioning in the equivalent mode, as they do not have a RF to
develop an object and self-constancy.
Empirical research has further shown that borderlines and patients with antisocial
personality disorders have a significantly lower RF than normal people (Fonagy et al.,
1995,1996a), that RF was protective against post-traumatic pathology (Fonagy et al., 1996a),
and that there are good reasons to see it as a factor of resilience in people with constitutional
or environmental vulnerabilities (Fonagy et al., 1994).
Psychoanalysis and psychoanalytic therapy offer an opportunity for patients to develop
their capacity for reflective functioning, reflections on their mental states being worded by
patient and analyst and differences between their own reflections and the one’s of the analyst
becoming clear through interpretations (Fonagy & Target, 1996). This is a necessary
developmental help in severe psychopathology (Bateman & Fonagy, 2001) and in child
psychotherapy (Fonagy & Target, 1998b).
The major point of criticism on RF (Bram & Gabbard, 2001) is that, although it is not
conceptualised as a cognitive function, its operationalization in the RF scale is mainly
cognitive.
61

62
Mentalization and Interpretive Interpersonal Function
Fonagy, Gergely et al.(2002) further expanded their theory of mentalization to the
concept of Interpretive Interpersonal Function (IIF). It is a function of an overarching neural
system in the related domains of affect-regulation, psychic processing and reflective
functioning. RF or the interpretation of relational experiences in terms of mental states is now
seen as part of this larger interpretive function. The other parts are the psychic processing by
the creation of second-order representations which are the base of affect-regulation. Parental
mirroring is crucial to this function. Although based on developmental and biological theories
it is close to Bion’s approach of mentalization: the psychic processing of affects and
experiences into psychic elements with the parental containment as a base.
Moreover, IIF is seen as a dual function: IIF-affective, which is closer to empathy, and
IIF-cognitive, which is closer to RF. This dichotomy overcomes the former critique that the
mentalization concept of Fonagy and colleagues is too cognitive in nature.
The psychopathological hypotheses stemming from this new model are somewhat different
from the ones linked with the original RF theory. Borderline pathology was explained
originally as not having a model of what happens in the mind of another, which leads to a lack
of object-constancy and having to rely on other mechanisms to release tension (Fonagy,
Steele et al., 1994), and on the persistence of the dual (equivalent-pretend) mode of
functioning (Fonagy & Target, 2000). In the recent model (Fonagy, Gergely et al, 2002)
borderline pathology is linked with a lack of parental affect regulation, causing a failure in the
creation of second-order representations, which is close to Bion’s point of view. Narcissistic
pathology is explained as a special failure of parental affect regulation. When the infant’s
emotion is misperceived by the caregiver, the mirrored state is incongruent with the infant’s
62

63
feelings. This gives raise to an alien state, which is the accumulation of un-mirrored internal
(self) states that have acquired a representation in terms of the other rather than the self. Yet
they are experienced as part of and within the self.
The Bion and the Fonagy Approach of Mentalization Compared
Empirically
To study whether changes in the mentalization according to Bion are a significant
dimension of the therapeutic process, we constructed a scale to rate changes in the Bion
mentalization level (B-Grid scale). Bion made an open Grid of at least 34 categories to
categorize the level and the use of psychic elements in the automatic, preconscious
mentalization process. The 6 point scale is based on the presence of these categories in the
answers of the ORI 1. means no psychic elaboration in the responses, 2. some psychic
markers, 3. clichéd, existing forms of containing an affective experience or mentally warding
off affective experiences for instance by rationalisation, 4. spontaneous and creative use of
metaphors, dream images, thoughts, 5. the use of more than one category in mentally dealing
with an emotional experience evoked by a question in the ORI, 6. a flexible use of a wide
range of categories. The B-Grid scale is assessed on the Object Relation Inventory of Blatt
(1996). Preliminary results about inter-rater reliability and validity are good (Vermote,
Vertommen, Verhaest et al., 2004).
Fonagy and his colleagues’ Reflective Functioning, is operationalised in the Reflective
Functioning Scale (Fonagy et al., 1998a). This is a 7 point scale with intermediate levels
about the capacity to perceive ones own and others actions in terms of mental states (Fonagy
et al, 1998):1. means Lacking RF, 3. Questionnable or low RF, 5. Ordinary RF, 7. Marked
63

64
RF. The RFS is assessed on the Adult Attachment Interview. The RFS has an established
reliability and validity (Fonagy et al., 1998a).
The main difference between the two scales is that RF as expressed in the RFS relates to
the way people reflect (cognitively and consciously) on self-other relationships, while Bion’s
thinking as reflected in the B-GRID scale concerns the way people process their perceptions
and affects at a preconscious level.
Both approaches do not necessarily coincide, there are patients who have a high capacity
for interpreting others’ behaviour without being in contact with their own affects. On the other
hand, there are also patients with highly creative, associative psychic functioning who are in
contact with their emotions, but who lack the capacity for interpreting their own behaviour
and that of others. This contrasts with the view of Lecours and Bouchard (1997), who
hypothesize the Reflective Functioning of Fonagy and colleagues being on top of the
mentalization as formulated by Bion, they suggest that the psychic processing as described by
Bion is the basis of Reflective Functioning.
In a pilot study we examined the relation between the Reflective Functioning of Fonagy
and colleagues and the mentalization according to Bion. We further examined whether the
DSM IV axis II borderline group showed a significant difference with the non-borderline
group in RF and in Bion Grid measures. A significant lower RF was expected in the
borderline group (Fonagy, 1996a).
The subjects of this pilot study were 14 personality disordered patients, 5 M, 9 F,
admitted to the setting for psychoanalytically informed hospitalization at the University
Centre St.-Joseph, Kortenberg, Belgium. Their DSM IV axis II diagnoses were: 8 borderline,
3 narcissistic, 1 schizoid-narcissistic, 1 dependent, 1 not otherwise specified.
An ORI and an AAI were assessed wihin the same week by two separate psychologists,
one trained in assessing the ORI and one trained in assessing the AAI. The anonymised ORI
64

65
transcripts were rated with the Bion Grid Scale by a third psychologist, who was blind to any
information about the patients and who was using a manual to rate this scale. The anonymised
transcripts of the AAI were rated with the RFS (Fonagy et al., 1998) by a fourth psychologist,
trained in rating RF on the AAI.
The correlation between the Bion Grid scores and RF scores was calculated with the
Pearson Correlation Coëfficient. The difference of RF measures of the borderline versus the
non-borderlinegroup was calculated with a t-test. The same was done for the Bion GRID
measure. Given a lacking equality of variances in this small group, non parametric
alternatives such as the Wilcoxon Rank Sums, the Chi-square and Exact tests were calculated
as well.
As a result we found no significant correlation (r = .37, p = .19) between RF values (M
= 5.43, SD = 1.91) and Bion Grid values (M = 5.57, SD = 1.22). The mean RF of the
borderline group was 4.13 (SD = 1.36) and that of the non-borderline group was 7.17 (SD =
0.74). The mean Bion GRID scale of the borderline group was 5.00 (SD = 0.75) and that of
the non-borderline group 4.67 (SD = 0.82). As expected from Fonagy et al. (1996a), the RF
differs significantly as a function of the DSM IV axis II diagnosis of BPS (t (12) = – 4.92, p =
.0004), while this is not the case for the Bion Grid scale (t (12) = –0.70, p = .49). Non
parametric tests yielded similar results.
We may conclude that the poor correlation between RFS and Bion Grid scale suggests
that the RFS and B-Grid scale measure different aspects of mentalization, rather than RF
being on top of the Bionian mentalization, as Lecours and Bouchard (1997) hypothesized. The
difference of the mean RFS for the borderline versus the non- borderline group was expected
from Fonagy et al. (1996a). The fact that this is not the case for the Bion Grid Scale is another
argument that it are two different aspects of mentalization.
65

66
Clinically
The difference and relationship between the two concepts of mentalization is illustrated
in the following sequence of sessions of a psychoanalysis.
Session 1. The patient associates about eagles with powerful claws, which reminds
her of her father.’ My family says that, as a child, I had a tendency of staring out in
front of me at the world.’ The staring eyes evoke in her the image of an owl. She
imagines a large owl behind her, who is covering her with his wings. She contrasts the
owl with the eagle. (The elements in this session have the characteristics of what we
called the GRID open zone, emotional experiences evoking dreams, images, thoughts.
The communication between analyst and analysand is on the same wavelength. It is
clear that this owl refers to me (one of the authors which was the analyst) as a good,
backing object).
Session 2. The analysand expresses her anxiety over some serious problems with
one of her adolescent children, which she interprets from her own feelings of guilt. She
is not in a mood for associative work and wants to act and take immediate decisions
about a hospitalization of her daughter. (It is striking that despite her usual rich
mentalizing in the Bionian sense, she shows a low RF capacity in the Fonagy and
Target sense. She has no idea at all of what is going on in the mind of her daughter. She
is in the equivalent mode, making hardly any distinction between her inner psychic
reality and the reality outside.) After giving it a thought, I decide to ask her how she
thinks her child is experiencing what is happening. Through reflection she develops a
number of points of view, which her daughter might be holding, and then decides to find
out what happens in the mind of her daughter instead of taking an immediate decision.
66

67
Session 3. In the next session, the analysand says how much the previous session
has helped her. Her daughter took a good solution herself for her problem. The patient
has the feeling that she understands her daughter better. (Nevertheless, I felt very
ambivalent. On the one hand, I had the feeling of having helped her, but on the other
hand I might have broken the spontaneous mentalization process by asking her
questions, stimulating a cognitive stance). Then she mentions a dream she had the other
night, a dream about the splitting of a Siamese twin. (The image reflects how she
experiences her daughter more as an individual on her own. The dream also shows that
while she felt helped at a conscious level, at a preconscious level, she felt put at a
distance by my asking her questions and taking a cognitive stance, which is not my usual
attitude. It is switching from the open to the closed zone in the Bion Grid.
Preconsciously she perceives this switch, she feels put at a distance by the cognitive
stance and elaborates this creatively in the dream of the splitting of the Siamese twin.
She repairs at a preconscious level what could have been a rupture of the mentalization
process according to Bion. Her dream brings the mental functioning and the interaction
between analyst and analysand, again in the so-called open zone of Bion’s Grid.
Theoretically
We argued how the recent concept of Interpersonal Interpretive Function (IIF), is closer
to the Bion model of mentalization, than is the case with RF alone. However it is difficult to
compare the two models. The IIF concept integrates genetic, biological, affective, and
learning theory, developmental psychology, as well as several empirical and conceptual
approaches to psychoanalysis, while Bion’s model is rooted in his experiences as an analyst.
A clinical meeting ground for the two approaches may be found with the Mentalized
67

68
Affectivity (M.A.) concept (Fonagy, Gergely et al., 2002, p.436) which is defined as a
sophisticated kind of affect regulation that denotes how affects are experienced by a person
which is self reflective, while he/she remains within or recaptures the affective state. This
concept may be particularly helpful in the treatment of borderline patients, whose capacity for
reflective functioning needs to be increased.
Fonagy discriminates three dimensions in dealing with affects in M.A, which he calls
elements. This is different from the meaning of the word element in the Bion where it denotes
a manifestation of psychic functioning which can be placed in a single Grid category (Bion,
1963a). The three M.A. elements are the identifying, the modulating and the expression of
affects. The ‘identifying of affects’ dimension is in its basic form the naming of affects and in
its complex form the discerning of their relation. The ‘modulating of affects’ dimension
consists basically of a modification of affects in intensity and duration, in the complex form it
is about the making of coherent links with one’s history. The ’expression of affects’
dimension is in its basic form the restraining or expressing of affects, outwardly or inwardly.
The inward mode is especially consistent with self-reflexivity. The complex form indicates a
communication with the expectation of how it will be received by the others.
Fonagy’s M.A. scheme and Bion’s Grid both offer a good reference in dealing with
personality disorders, to know at which level they are functioning at a given moment and to
know at which level one has to respond. A patient in an emotional storm for instance, with
undifferentiated feelings and an intense outward expression is not helped with sophisticated
interpretations but with a restoration of his feeling of safety. The M.A. offers a good frame of
reference to detect at which level the patient deals self-reflexive with his affects so that the
therapist can match his intervention.
This way, it is similar with Bion’s Grid, which is as well an instrument to increase the
awareness of the level at which the patient is functioning (Chapter 1). The Grid is finer in this
68

69
respect in showing also the preconscious and unconscious associative-creative dealing with
affects, in having a larger scope of notation (preconcept, conception, dream-images, thoughts,
abstract ideas) and that the use of the notation is also taken into account (definition- denial-
exploration-action). In addition, the B-Grid offers in addition an elaboration of the dealing
with affects which are warded off by the psyche, the elements of the A row and the second
column.
The element of M.A, denoted as ‘ expression of affects’ is divided in an inward
expression which corresponds to the first five columns of the B-Grid, and an outward
expression which corresponds with the last column of the B-Grid. In comparing M.A. and the
B-Grid, the Bion Grid proves to be more complicated and serves to elaborate a session after
the session, while M.A. may be used as a compass during a session which is useful especially
in intense sessions with personality disordered patients. Furthermore, the Bion Grid lacks the
relational component of the M.A. approach, as it focuses only on intrapsychic processing.
Both approaches are overlapping in some aspects and complementary in others, but at
large the dichotomy between the two approaches of mentalization remains. The Fonagy
approach is more on the ego-psychological side (a thinking agent) while the approach of Bion
is more kleinian and stresses the unconscious and preconscious side of psychic functioning
where feeling and thinking are interwoven in alpha-elements, phantasies, preconcepts, to
more conscious thoughts and theories.
Conclusion
Mentalization is a many-sided concept. The RF concept was developed of research
findings and proved to be predictive of the transmission of psychopathology, and is lowered
in BPS and antisocial disorders. As it is rooted in research, it is a simple and well-defined
69

70
concept, which offers a model for understanding psychopathology. The concept of RF is
concerned with the conscious dealing with inter-psychic issues, a register which received less
attention in adult psychoanalysis because of the limits of the psychoanalytic frame based on
free association. In contrast, the bionian concept of thinking deals with the way preconscious-
unconscious psychic processing occurs in the sessions. Bion elaborated his concept over a
period of twenty years into a kind of epistemology with consequent implications for practising
analysts. By using Bion’s Grid as a map, we could explore psychopathology with his model
on psychic functioning and explore the psychoanalytic process. Preliminary results of
operationalization in a GRID scale indicate that Fonagy’s RF and the bionian thinking are two
sides of mentalization.
Fonagy and his colleagues’ new IIF model of mentalization integrates RF but also pays
attention to the aspect of psychic processing and affect regulating. There is no empirical
operationalization of this global IIM model yet. The notion of Mentalized Affectivity is a
translation of this model in the practice of psychotherapy and psychoanalysis. This makes it
possible to hold the Fonagy and Bion models against the light of clinical experience. In so
doing, the difference between an ego-psychological and a kleinian watershed becomes clear,
as two different, largely complementary sides of a complex phenomenon.
70

71
PART 2
INSTRUMENTS TO MEASURE PSYCHIC CHANGE IN A
PSYCHOANALYTICALLY INFORMED HOSPITALIZATION
BASED TREATMENT OF PERSONALITY DISORDERS

72

73
Chapter 3
Measuring the Psychoanalytic Process in Personality Disorders on the
Object Relations Inventory: the Independent Researcher's Perspective.
Chiesa (2000) reported overall improvement with a psychoanalytically informed
hospitalization, followed by outreached therapy. The Bateman and Fonagy RCT study (1999,
2001) demonstrates significant clinical improvement in an 18-month psychoanalytically
oriented partial hospitalisation program and found that the improvement continued in the 18
months after the treatment. This study caused a reappraisal of a psychoanalytic approach for
personality disorders as expressed in the guidelines of the APA (Sanderson et al., 2002).
The therapy being effective, the underlying mechanism of change remains unclear.
Psychoanalysts traditionally link this outcome with the psychoanalytic process they focus on.
As we described in Chapter 1, the focus on inner change entails other techniques, evaluations
and length of stay than symptom oriented therapy. It is therefore important to examine these
inner changes with psychoanalytic model specific measures and to study their relation with
changes in symptoms and ways of coping and relating. The problem is however that there is
no clear definition of the psychoanalytic process (Vaughan, Spitzer, Davies & Roose, 1997).
There exist nevertheless several scales for rating it by independent researchers. Most of these
scales have a large number of items which may be related to a process of inner change. The
Psychotherapy process Q-set (Jones, 2000), is such a scale with 100 items describing patient's
and therapist's actions, attitudes and experiences and their interaction. The scale does not rely
on a definition of the analytic process but categorises what happens in a session, which is then
considered to be related to the analytic process. Clinical judges watch a videotape or read the
transcripts of a therapy session and then sort the 100-items on a continuum from least to most
73

74
characteristic. Although this instrument has high levels of reliability and validity, it was
problematic for us to use this time consuming instrument, in a prospective process-outcome
study on a larger group of hospitalised patients. Furthermore it is difficult to use it in an
integrated therapy programme, based on group therapy.
Another instrument is the Columbia Analytic Process Scale (CAPS) (Vaughan et al.,
1997) which is based on a definition of the psychoanalytic process as existing of free
association, interpretation and working through. The CAPS measures these facets of an
analytic process. We have discussed in Chapter 1 why, in our opinion, this definition of the
PA process is not appropriate for PD. Moreover, the scale did not reach construct validity
(Vaughan & Roose, 1995).
Another overall scale which may be used to measure the psychoanalytic process is the
Scale of the Psychological Capacities (SPC, Wallerstein, 1994), describing 17 psychological
capacities which are measured on an audio-or videotaped clinical interview. This instrument
with an ego-psychological base has a good reliability and validity, but it is a measure which
seems more apt to measure outcome than the mechanism of inner change.
Most other psychoanalytic model specific process scales for independent researchers
measure components of the psychoanalytic process. The Core Conflictual Relationship Theme
(CCRT, Luborgsky & Crits-Christoph, 1999) is a reliable method to capture the 'central
relationship pattern' on transcripts of sessions. Three components are rated: a subject's wish,
need, or intention; a response from another person to the subject's wish; and the response of
the subject to the response of the other. Another scale about the specific interaction in
psychoanalytic sessions is the Transference Countertransference Analysis (TCA, Normandin
& Foelsch, 1999), which is applied to videotapes. It has good inter-rater reliability but there
are no data available about validity.
74

75
Other tests study facets of the process in detail, like the Referential Activity Scale
(RA, Mergenthaler & Bucci, 1999) which is a measure about the linking of verbal and non-
verbal representations. This scale measures the concreteness, imagery, specificity and clarity
of speech. A detailed examination of the interaction between therapist and patient can be
studied by using Facial Action Coding System (FACS, Ekman & Rosenberg, 1997). The scale
makes it possible to register descriptively on videotapes, every facial movement on patient
and analyst that is anatomically feasible. It is evident that these mesaures are difficult to use in
a study of a large number of personality disordered patients in a psychoanalytically informed
hospitalisation with a integrated programme existing of several kinds of therapies.
It was J. Clarkin who suggested to try to operationalize the three dimensional model of
change discussed in Chapter 1, in trying to measure the psychoanalytic process of P.D.
patients. It was a major advantage that there existed already two well established measures
about inner change in patients, the Differentiation Relatedness Scale (Blatt et al., 1996) and
the Reflective Functioning Scale (Fonagy et al., 1998) which could be related to this model.
This project which was supported by the authors of these scales (P. Fonagy and S. Blatt).
The three dimensional model of inner changes in personality disorders (Chapter 1) is based on
the common assumption (Balint, 1968; Bion, 1962; Fonagy, 1999; Kernberg, 1980 and
Winnicott, 1965) that psychoanalytic change is not about the interpretation of unconscious
meaning or about searching for a historical truth. We see intrapsychic change in personality
disorders as being related to changes in three inter-related dimensions. The first dimension is
about the subjective experience of security or safety and is related to a range of concepts. The
second dimension is about mentalization of which we discussed two facets. The first facet
being the psychic processing of intrapsychic and interpersonal experiences, a preconscious,
automatic, spontaneous process which is studied in detail by Bion (1962, 1963, 1965) in his
theory on thinking. The second facet being the intentional stance toward interpersonal
75

76
experiences in taking a step back and trying to understand one’s own and others' behavior in a
psychic way, which is called Reflective Functioning by Fonagy and colleagues (Fonagy,
Steele, Moran, Steele & Higgitt, 1991; Fonagy, Moran, Edgcumbe, Kennedy & Target, 1993;
Fonagy, Leigh, Steele, Steele, Kennedy, Mattoon et al., 1996; Fonagy &Target, 2000;
Fonagy, Gergely, Jurist & Target, 2002). The third dimension is the transformation of mental
representations of self and others, into more differentiated and stable ones with an increasing
tolerance of ambivalence. Mental representations are active schemata, which may be
conceived as linked to autobiographical as well as to procedural memory (Blatt & Auerbach,
2001; Blatt, Stayner, Auerbach, & Behrends, 1996). They structure and modulate how one
feels and thinks about others and oneself.
The aim of this study is to examine whether it is possible for external researchers to assess
and measure the psychoanalytic process in personality disorders according to these three
hypothesised dimensions of intrapsychic change.
76

77
Measuring the Three Dimensions of Intrapsychic Change on the Object Relations Inventory
(ORI).
To measure the first dimension or the feeling of safety, we considered at first to rely
on existing scales such as the Barett-Lennard (1966) or existing attachment scales. However
they are more about the therapeutic relationship and take the therapist into account as well,
while we are interested in measuring psychic changes in the patient, more specific his psychic
experience of safety. Therefore we decided to construct our own scale. It is a 5-point scale
about felt safety (FS), which reflects the security a patient is experiencing in the setting at a
given moment (Figure 1).
Figure 1. Felt Safety Scale (Vermote & Vertommen, 2004).
Felt Safety Scale
1. No felt safety in therapy
2 Predominance of feelings of unsafety, minimal expression of safety
3. Fragile, changing feeling of safety
4. Predominant feeling of safety
5. Strong feeling of safety
To assess the second dimension or mentalization we used two scales. For the reflective
functioning aspect of mentalization, we used the Reflective Functioning Scale (RFS; Fonagy
et al., 1998, Figure 2). This is an 11-point scale designed to be assessed on the Adult
Attachment Interview (AAI; Main et al., 1991), ranging from anti-reflective responses to
responses with an exceptional sophistication. The RFS as measured on the AAI has
established reliability and validity (Fonagy et al., 1998).
77

78
To assess the psychic processing aspect of mentalization, the generation and
transformation of mental representations, we rely on Bion’s theory (1963) of the elements of
thought and his theory of the transformations of these elements (Bion, 1965; Vermote, 1998).
Bion categorizes the levels and the use of these elements in a Grid (Chapter 2). A translation
of Bion’s Grid into a scale was achieved by discerning four zones in his Grid (Figure 3).
These zones could be placed in a ranked order. Taking the flexibility of switching within and
between zones into account as well led to 6 distinguished categories (Figure 4).
To assess the third dimension or the mental representations of self and others, we use
the Differentiation Relatedness Scale (DRS) (Diamond, Blatt, Stayner, & Kaslow, 1993; Blatt
& Auerbach, 2003). The DRS (Figure 5) integrates the conceptual level of the representations
with the level of self-other differentiation which they reflect, both seen from an interpersonal
perspective. The DRS is based on the assumption that development moves toward (a) a
consolidated, integrated, and individual sense of self-definition and (b) an empathically
attuned mutual relationship relatedness with significant others (Diamond, Blatt, Stayner, &
Kaslow, 1993). Differentiation and relatedness are seen as interactive dimensions that unfold
throughout development (Blatt & Auerbach, 2001). The DRS is an ordinal scale with ten
levels
78

79
Figure 2. Reflective Functioning Scale (Fonagy et al., 1998).
Level/ Scale Point Description
-1 Negative reflective
functioning
Responses that are anti-reflective, hostile, bizarre, or inappropriate
in the context of the interview
1. Absent Reflective
functioning
Responses either totally or almost totally lacking in reflective
functioning, with little evidence that the person thinks about mental
states. Accounts are barren, lacking in detail pertaining to mental
states, or are egocentric and self-serving.
3. Questionable Reflective
Functioning
Responses contain some evidence of consideration of mental states,
but most references are not made explicit. The person’s
understanding of mental states is either banal and cliched or diffuse
and unintegrated.
5. Ordinary Reflective
Functioning
The subject displays an ordinary capacity to make sense of
experiences in terms of thoughts and feelings and has a consistent
model of mental states that requires little or no inference from the
rater. The subject’s model is limited and does not include
understanding of conflict and ambivalence
7. Marked Reflective
Functioning
Responses contain numerous instances of full reflective functioning
suggestive of a stable psychological model of the mind. Much detail
about thoughts and feelings is present and implications of mental
states are explicitly spelled out. The subject is usually able to
maintain a developmental interactional) perspective and to arrive at
original reintegrations of states of mind
9. Exceptional Reflective
Functioning
Responses show exceptional sophistication. They are commonly
surprising in their insights, are quite complex or elaborate, and
consistently manifest causal reasoning with regard to mental states.
The subject displays a consistent reflective stance across all
contexts.
79

80
Figure 3. Zones in Bion's Grid
Definitory
hypothesis
psi notation attention inquiry action …n
A
Beta-elements
A1 A2 A6
B
Alfa-elements
B1 B2 B3 B4 B5 B6 ….Bn
C
dreams, myth
C1 C2 C3 C4 C5 C6 ….Cn
D
Pre-conception
D1 D2 D3 D4 D5 D6 ….Dn
E
Conception
E1 E2 E3 E4 E5 E6 ….En
F
Concept
F1 F2 F3 F4 F5 F6 ….Fn
G
Wet. syst
G2
H
Algebr.calculus
Bion distinguishes eight rows from A to H in the vertical axis of the GRID, which represent the genesis of thinking and levels of psychic functioning. -A: Beta-elements are sense data, which have not yet been processed. and are unfit for
linking-articulating. Therefore they can only be placed in columns 1, 2 and 6. -B: Alpha-elements, bear a trace of something psychic and are the building bricks of dream
thought. -C: C-elements are the products of dream thought: dreams, myths, phantasms; usually visual
in nature, but olfactory, tactile and auditive images exist as well.
80

81
-D: Pre-conceptions are empty, existing moulds (congenital, or as part of culture and language). They are open to a limited range of experiences, have the capacity of containing sensory and affective perceptions.
-E: Conceptions emerge when a perception or feeling meets an appropriate form or mould, thus resulting in a realization.
-F: Concepts are conceptions which have freed themselves to certain extent from concrete experiences or perceptions, and are therefore more generally applicable.
-G: Scientific Deductive System: is a linking of various concepts resulting in a predictable theory. Bion (1963) believes that psychoanalysis does not reach this level, G can only be found in the second column.
-H: Algebraic calculus: a scientific system in algebraic terms, to make predictions, free from actual reality (e.g. physics). H does not yet exist for psychoanalysis.
The horizontal axis shows how these elements are put to use. 1. Definition: elements used as a statement which closes or anchors something (e.g. you are
depressed). 2. Ψ: elements used to dissipate something (denial: e.g. a handshake to dispel anger), or to lie.
According to Bléandonu (1994), the term “Ψ” is derived from “προτον ψευδοσ”, the first lie as described by Freud in is work on Hysteria.
3. Notation: taking note of (e.g. writing something down, transposing something into images, stories, etc.).
4. Attention: the rêverie, the free-floating attention which allows for the emergence of a “selected fact”. Bion held on to this state of non-fixation and labelled the category D4 as the locus of the love of the analyst for his patient (Bion, 1965).
5. Inquiry: a more goal-oriented scrutiny; marked by obstinacy and curiosity. Therefore labelled first the “Oedipus” column (Pontes de Miranda Ferreira, 1997).
6. Action: doing something with the elements (e.g. evacuation of a beta-element, giving an interpretation, or making a decision).
Figure 4. The Bion – Grid Scale. 1 No psychic quality, not articulated A1,A2, A6 2 Psychic impression, trace, marker, barely psychic B1,B2,B3,B4,B5,B6 3A Elements used for closing the psychic elaboration of an experience
C1,C2,D1,D2,E1,E2,F1,F2,G3 3B Searching for an existing form to contain an impression, an experience D3,D4,D5,D6 4 Emerging dreams, myths, images C3,C4,C5,C6 / Spontaneous finding of a psychic form
to contain a concrete experience E3,E4,E5,E6/ Open elaboration by concepts (less concrete) F3,F4,F5,F6 : All this by remaining at the same level, in the same row
5 When the patient uses elements from different rows in elaborating an experience, an emotion, a perception or a thought (see arrows)
6 When a large field of elements in the GRID is covered during the ORI (see arrows)
81

82
Figure 5. The Differentiation-Relatedness Scale (Diamond et al., 1993).
Level/Scale Point Description
1 Self/Other boundary
compromised
Basic sense of physical cohesion or integrity of
representations are lacking or breached
2. Self/other boundary confused Self and other are represented as physically intact and
separate, but feelings and thoughts description may consist of
single global impressionistic quality or a flood of details with
a sense of confusion and vagueness
3. self/other mirroring Characteristics of self and other, such as physical appearance
or body qualities, shape or size, are virtually identical
4. self/other idealisation or
denigration
Attempt to consolidate representations based on unitary,
unmodulated idealisation or denigration. Extreme,
exaggerated one-sided description.
5. Semi-differentiated, tenuous
consolidation of representations
through splitting (polarisation)
and/or by an emphasis on concrete
part properties
Marked oscillation between dramatically opposite qualities or
an emphasis on manifest external features
6.Emergent, ambivalent constancy
(cohesion) of self and an emergent
sense of relatedness
Emerging consolidation of disparate aspects of self and other
in a somewhat hesitant, equivocal, or ambivalent integration.
A list of appropriate conventional characteristics but they
lack a sense of uniqueness. Tentative movement toward a
more individuated and cohesive sense of self and other
7. Consolidated constant (stable)
self and other in unilateral
relationships
Thoughts, feelings, needs and fantasies are differentiated and
modulated. Increasing tolerance for and integration of
disparate aspects. Distinguishing qualities and characteristics.
Sympathetic understanding of others.
8. Cohesive, individuated,
empathically reated self and
others
Cohesive, nuanced, and related sense of self and others. A
definite sense of identity and interest in interpersonal
relationships and a capacity to understand the perspective of
others
9. Reciprocally related integrated Cohesive sense of self and others in reciprocal relationships
82

83
unfolding self and others that transform both the self and the other in complex,
continually unfolding ways
10. Creative integrated
constructions of self and other in
empathic, reciprocally attuned
relationships
Integrated reciprocal relations with an appreciation that one
contributes to the construction of meaning in complex
interpersonal relationships
To obtain data for the DRS (Diamond et al., 1991, 1993), Blatt and Auerbach (2003)
use the Object Relation Inventory (ORI), a simple semi-structured interview with open
questions about parents, self, therapist, and significant other.
Assessing the RFS is very time consuming. It takes intensive training to administer the
AAI and another intensive training to rate the RFS on it. The procedure of interviewing and
rating takes more than 10 hours for one interview. Therefore, rating the RFS on the ORI
instead of on the AAI is an idea favoured by Fonagy and colleagues and by Blatt (Target &
Blatt, 2001; Vermote, Vertommen, Corveleyn & Peuskens, 2002; Blatt & Auerbach, 2003).
As the ORI consists of questions about self-other relationships and RF is about the capacity to
deal mentally with self-other relationships, it is reasonable to do so.
The Bion Grid Scale can be scored as well on the ORI, as the ORI questions are open
and invite for association and creative elaboration. A problem in measuring the two facets of
mentalization on an ORI is that the Bion Grid scale is about the spontaneous, preconscious,
creative generation and use of thoughts in all levels of sophistication, while the RFS is about a
conscious, intentional stance evoked by ‘demand questions’ of the AAI. In other words,
assessing the first aspect of mentalization requires an open attitude which leaves the
possibility to associate freely, while the second needs thought-provoking, stringent questions.
We decided, therefore, to let patients freely associate on the four questions of the interview in
the first part and to ask for elaboration on specific examples in a second part of the interview
(Blatt & Auerbach, 2003; Vermote, Vertommen,Verhaest et al., 2004).
83

84
We constructed the Felt Safety Scale to be rated on the ORI. The scale is about the
way patients experience the safety of the therapy and is rated on the parts of the ORI
concerning the therapist and on the elaboration of an example about the therapist.
The DRS and RFS were translated into Dutch. For the DRS manual, difficulties in
interpreting were submitted by way of ORI examples to S. Blatt (Yale University, New
Haven), author of the scale, to be sure that we interpreted the DRS categories correctly.
The original manual constructed for rating the Reflective Functioning on the AAI was
adapted for use with the ORI. This was done with permission of P. Fonagy and colleagues
(University College London) and with permission of S. Blatt. To be sure that valid
interpretations were made, one of us (M. Franssen) engaged in RF rating training with Fonagy
and Target. The FS and B-Grid scale are newly constructed scales to be rated on the ORI.
Each of the scales went through several phases of development: the construction or translation
of the scale, the making of a manual, pilot testing, revision of the manual, and subsequent
retesting. For all the scales, a global mean was determined, relying on an overall impression
after all the parts of the interview were rated separately.
Table 1
Parts of the ORI Which are Rated for Each Scale Measured on the ORI
Scale Mother Father Self Therapist Ex.Mother Ex.father Ex. Th. Global
DRS
RFS
BGS
FSS
X
X
X
X
X
X
X
X
X
X
X
X
X
(x)
X
X
(x)
X
X
(x)
X
X
X
X
X
X
X
(x): these fragments were rated in addition in study 1
84

85
Reliability and Validity of Measuring the Three Dimensions of the Psychoanalytic Process on
the ORI
Study 1. Inter-Rater Reliability of the ORI Measures
Method
Participants.
Reflective Functioning and Bion Grid Scale: Fifteen patients (6 males, 9 females)
referred to a setting for hospital treatment on psychoanalytic lines for personality disorders at
the University Center Kortenberg, Belgium (see Chapter 1) were tested. The patients were
between the ages of 17 and 45 years (M = 24.4, SD = 7.36). Two patients were single, 10
lived with their parents, 2 were married or living together with a partner, 1 was divorced. One
patient completed primary education, 6 secondary education and 8 higher education or
university.
Based on the SCID-II we assigned all patients to one of the three clusters of personality
disorders according to the DSM-IV axis II (APA,1994): Cluster B, 11 (7 borderline, 3
narcissistic, 1 histrionic and borderline); Cluster C, 3 (1 avoidant and dependent, 2 NOS);
Clusters B+C, 1. DSM-IV Axis I diagnoses were 5 mood disorders, 1 anxiety disorder, 3
adjustment disorders and 2 substance related disorders.
Differentiation Relatedness Scale: Fifteen patients (4 males, 11 females) referred to a
setting for hospital treatment on psychoanalytic lines for personality disorders at the
University Center Kortenberg, Belgium (see Chapter 1) were tested. The patients were
between the ages of 19 and 58 years (M = 35.4, SD = 11.33). Ten patients were single, 3 were
85

86
married or living together with a partner, 2 were divorced. Three patients completed
secondary education and 12 higher education or university.
Based on the SCID-II we assigned all patients to one of the three clusters of personality
disorders: B, 13 (10 borderline, 2 narcissistic, 1 narcissistic and borderline); C, 1 (1 NOS);
B+C, 1. DSM-IV Axis I disorders were 5 mood disorders, 4 adjustment disorders, 1 eating
disorder, and 1 substance-related disorder.
Felt Safety Scale: Fifteen patients (3 males, 12 females) referred to a setting for hospital
treatment on psychoanalytic lines for personality disorders at the University Center
Kortenberg, Belgium (see Chapter 1) were tested. The patients were between the ages of 18
and 37 years (M = 27.73, SD = 6.0). Seven patients were single, 4 lived with their parents, 4
were married or living together with a partner. Six patients completed secondary education
and 9 higher education or university. Based on the SCID-II we assigned all patients to one of
the three clusters of personality disorders according to DSM-IV: cluster B, 9 (5 borderline, 1
narcissistic, 2 narcissistic and borderline, 1 histrionic and borderline); cluster C, 3 (2
dependent, 1 NOS); clusters B+C, 3. DSM-IV Axis I disorders were 7 mood disorders, 2
adjustment disorders, and 1 substance-related disorder.
Instruments and procedure.
The ORI interviews were assessed by a psychologist who did not have any other contact
with the patients. The transcripts of the interviews were rated by three raters for each scale,
with a total of 12 raters. Each rater had a short training with the manual of the scale he or she
was rating. There was a difference in the background of the raters. For each scale there was an
experienced psychoanalytic psychotherapist or psychoanalyst and two students of psychology,
86

87
except for the Bion Grid Scale, which was rated by two senior and one junior psychoanalytic
therapist-psychoanalyst. The transcripts were anonymized, distributed at random, and the
raters had no contact with the patients and were blind to any information about them.
Analysis of data.
The inter-rater reliability was measured with Kendall’s coefficient of concordance. As
the inter-rater reliability of the original DRS was measured with the intraclass correlation of
Shrout-Fleiss although it is an ordinal scale, we calculated this correlation as well for the
DRS.
Results
Inter-rater reliability, assessed with Kendall’s coefficient of concordance among three
raters for the global scores of each ORI scale is shown in Table 2. The Shrout-Fleiss intraclass
correlation coefficient for the DRS global score, fixed set mean scores, is.83.
Study 2. Validity of the ORI Measures
The validity of the four ORI scales is further studied by examining the correlations
among the different parts of the ORI and examining their independence from
sociodemographic variables. To study the hypothesized relationship of the scale scores with
87

88
level of psychopathology, we examined the correlations among three kinds of scales
(structural, descriptive, symptomatic) measuring degree of psychopathology.
Table 2
Inter-Rater Reliability of the ORI Scales, Kendall’s coefficient of concordance
Scale W
p
RFS glob, raters 1,2,3 .74 < .0001
BGS glob, raters 4,5,6 .77 < .0001
DRS glob, raters 7,8,9 .70 0.0005
FSS glob, raters 10,11,12 .84 < .0001
(n=15)
Method
Participants.
Eighty-two patients (54 male, 28 female) referred to a setting for hospital treatment (day
hospital and in-patients) on psychoanalytic lines for personality disorders at the University
Center Kortenberg, Belgium (see Chapter 1) were tested. The patients were between the ages
of 17 and 58 years (M = 27.1, SD = 8.5). Sixty-one patients were single, 29 lived with their
parents, 17 were married or living together with a partner, 3 were divorced, and 1 lived in a
community center. Three patients completed primary education, 32 secondary education, and
47 higher education or university. On the basis of the SCID-II, we assigned all patients to one
of the three clusters of personality disorders according to DSM-IV: cluster A, 2 (1 schizoid, 1
88

89
schizotypal); cluster B, 62 (1 histrionic, 44 borderline, 10 narcissistic, 4 narcissistic and
borderline, 3 histrionic and borderline); cluster C, 10 (1 avoidant, 2 dependent, 1 avoidant and
dependent, 6 NOS); clusters A+B, 1; clusters B+C, 10.
According to patients' medical files, there were diagnoses of 33 mood disorders, with 7
cases of major depressive disorder, single episode and 11 cases of major depressive disorder,
mild, recurrent; 5 anxiety disorders; and 14 cases of adjustment disorder, commonly with
mixed anxiety and depressed mood. Other Axis I diagnoses were anorexia nervosa, alcohol
abuse, and cannabis abuse.
Instruments.
The ORI interviews were assessed by a psychologist who did not have any other contact
with the patients. The transcripts of the interviews were rated by the raters from the reliability
study, utilizing the DRS, RFS, BGS and FSS. The transcripts were anonymized, random, and
the raters were blind to all other information.
The Inventory of Personality Organization (IPO) is a self-report questionnaire
(Kernberg, & Clarkin, 1995) which measures intrapsychic structure in three dimensions
according to Kernberg's model of borderline personality organisation: primitive defenses,
identity diffusion, and reality testing. Higher IPO sores correspond with more pathological
personality organizations. The scales show good to excellent reliability and validity
(Lenzenweger et al., 2001; Normandin, 2002). The reliability and validity of the IPO
translated in Dutch was demonstrated by Vermote, Vertommen et al. (2004), Vermote, Smits,
Claes & Vertommen (2005).
89

90
The Symptom Checklist 90 (SCL-90; Derogatis, 1977), translated by Arrindell &
Ettema (1993), consists of 90 items that tap eight psychopathological domains. The
Spielberger State-Trait Anxiety Inventory (Spielberger,1970) was translated by van der Ploeg
et al. (1980), the Spielberger State-Trait Anger Inventory (Spielberger et al., 1980) was
translated Van der Ploeg et al. (1982), and the Beck Depression Inventory (Beck, 1961) was
translated by Bouman et al. (1985).
The Structured Clinical Interview for DSM-IV Axis II disorders (SCID-II) is a
structured clinical interview for DSM-III-R Personality Disorders by Spitzer & Williams
(1985). The SCID-II Personality Questionnaire has been translated in Dutch and adapted for
DSM-IV by Weertman, Arntz & Kerkhofs, 2000.
Analysis of data.
Correlations among the four measures of the ORI were examined with the Pearson
Product Moment Correlation Coëfficient. The relationship with sociodemographic variables
was examined by correlating age with the various measures with the Pearson Product Moment
Correlation Coëfficient, by studying the effect of three levels of education by an analysis of
variance with the General Linear Model procedure and by evaluating gender differences with
a classical t-test. Correlations between the DRS and the IPO, SCID, and SCL-90 were also
analyzed by using the Pearson Product Moment Correlation Coëfficient.
Results
The Pearson correlations of the global scores on DRS, BGS, RFS, and FS with the
components of the ORI are shown in Table 3.
90

91
Table 3
Pearson Correlation Coefficients for ORI Global Scale Scores with Separate ORI Interview
Scale Scores
Scale Moth Fath self therap Ex. moth Ex. Fat Ex Ther
RFS glob .68* .70* .60* .53* .64* .74* .69*
BGS glo .79* .74* .78* .79* .79* .73* .77*
DRS glob .81* .77* .82* .76*
FSS glob .70* .81*
(n = 82). * p <.001
There is a significant correlation between the two measures of mentalization and
between RFS and FSS, as shown in Table 4.
Table 4
Pearson Coefficients for ORI Global Scale Score Intercorrelations
Scale RFS BGS DRS FSS
RFS — .42** .26 .30*
BGS — .15 .12
DRS — .10
FSS —
(n = 82). *p <.01, **p <.001
The DRS and RFS ratings were not influenced by sociodemographic variables. There
was a significant correlation between FS and age (r = .23, p = .03) and a significant
91

92
correlation of the BGS with age (r = .36, p = .001) and with education (r = .48, p = .01).
There were no gender differences on any of the four scales. The correlations among the four
global scores and the subscales of the IPO are significant only for DRS, with IPO primitive
defense mechanisms (r = -.21, p < .05), with IPO identity diffusion (r = -.21, p < .05) and
with IPO reality testing (r = -.24,p < .05).
When correlating the DRS, RFS, BGS, and FSS scores of 82 patients with their SCID-II
measures, we found a significant correlation between the DRS and avoidant (r = -.30, p =
.005), dependent (r = -.24, p = .03), and borderline (r = -. 23, p = .037) categories, as well as
a marginally significant correlation with the paranoid category (r = -. 20, p = . 067). We found
a significant negative correlation of the RFS with avoidant (r = -27, p = .01) and depressive
(r = -.22, p = .05) categories and a positive correlation with the narcissistic category (r = .21,
p = .05). The BGS tended to correlate negatively with the antisocial category (r = -.20, p =
.06). The FSS correlated with the schizotypal category (r = .33, p = .002) and tended to
correlate with the dependent category (r = .21, p = .06).
The correlations of the four global scales with the symptom scales on 82 patients
showed a significant negative correlation of the DRS with the SCL-90 (r = -.24, p < .05), with
the BDI (r = -.26, p < .05), with the STAI (r = -.22, p < .05) and with the STAXI (r = -.24, p <
.05) and a significant negative correlation of the global RFS scores with the Spielberger Trait
Anxiety Scale (r = -.22, p < .05). The FS and the BGS showed no reliable correlations with
SCL-90, BDI, STAI and STAXI.
The DRS Self subscale correlated significantly with nearly all SCL subscales, ranging
between r = -.26 and -.39 (p's < .01), except for the subscales of insufficiency in thought and
problems with sleep.
92

93
Table 5
Pearson Correlation Coefficient of the Global Scores of the ORI-Scales With Symptom Scales
Scale RF B-Grid DRS FS
SCL-90 -.08 .03 -.24* .19
BDI -.17 -.04 -.26* .05
STAI -.22* .01 -.22* -.03
STAXI -.18 -.01 -.24* -.11
(n = 82) *p < .05
Study 3. Convergent Validity of the ORI-RFS and AAI-RFS
To further study the convergent validity of the ORI-RFS with the original AAI-RFS, we
correlated both measures. In this analysis, however, there were three variables which we
correlated: two different interviews (AAI and ORI) on which the RFS was rated, two different
raters, and two manuals: the original and one adapted for rating the RFS on the ORI.
Method
Participants.
Fourteen personality-disordered patients were included in this study (5 M, 9 F). Their
DSM-IV Axis II diagnoses were as follows: 8 borderline, 3 narcissistic, 1 schizoid-
narcissistic, 1 dependent, and 1 NOS.
93

94
Procedure.
An ORI interview and AAI interview were assessed within the same week by two
separate psychologists, one trained in assessing the ORI and one trained in assessing the AAI
by the Main method (George, Kaplan, Main, 1985), both blind to other patient data. The
anonymized ORI transcripts were rated with the Leuven manual (see ORI Leuven 1 in table
6). The anonymized transcripts of the AAI were rated with the original RFS manual (Fonagy
et al., 1998) by a psychologist, trained at the UCL by Fonagy and colleagues (AAI Ucl
manual in table 6).
Given the poor reliability between RFS rated with the Leuven manual on the ORI and
that rated with the original manual on the AAI, we decided to ask the psychologist trained by
Fonagy and colleagues to rate the RFS as closely as possible to the London method but on the
ORI instead of on the AAI (ORI,UCL training in table 6). We wanted to exclude the
possibility that the poor correlation was based on a misinterpretation of rating Reflective
Functioning. Vice versa, we asked the psychologist trained with the Leuven ORI manual to
rate the AAI as good as possible without having got a proper training in doing this (AAI
Leuven1 in table 6).
Because two separate psychologists rated the RFS on the ORI, we have an extra rating
of the ORI (ORI Leuven 2 in table 6).
94

95
Results
The results of the Pearson correlation between mean global RFS scores for 14 patients
rated with the several methods are given in Table 6. The results of the correlation with
Kendall Tau B correlation coefficient are similar and therefore not presented.
Table 6
Pearson Correlation Coefficient of the Correlation Between Mean Global RFS Scores Rated
With Several Methods.
Method AAI Ucl man AAI Leuven1 ORI Ucl train ORI Leuven1 ORI Leuven2
AAI Ucl man
AAI Leuven1
ORI Ucl train
ORI Leuven1
ORI Leuven2
— .44
—
.32
.01
—
.09
-.14
.78 **
—
.12
-.15
.76 *
.95 **
—
(n = 14) *p <.01, **p <.001
The correlation between the scores of the RFS-AAI by the rater trained by Fonagy and
Target using their RFS-AAI manual (AAI Ucl man) and the RFS-ORI by a rater using the
Leuven-RFS-ORI manual (ORI Leuven 1), is only r = .09, n.s. When the AAI rater scored the
ORI adhering to the London training, the correlation between his ORI-RFS (ORI Ucl training)
and his AAI-RFS scores (AAI Ucl manual) was r = .32, n.s.
When the ORI-RFS scores, London method, (ORI ucl training) were correlated with the
ORI-RFS of the rater using the Leuven ORI-RFS manual (ORI Leuven 1), the correlation is
95

96
significant, r = .78, p < .001. When this rater scores the RFS on the AAI (AAI Leuven 1), but
without a training for doing this and using the Leuven ORI manual, the correlation is non-
significant. The correlations of the ORI-RFS scores of this rater with the ORI-RFS scores of a
second ORI-rater using the same method (ORI Leuven 2) were high, r = .95, p < .001.
Discussion
Every psychoanalyst agrees that he does not concentrate on symptom or behaviour
changes, but on inner psychic changes, the development of a psychoanalytic process.
Psychoanalysts have the experience from their own analysis and from what they see happen in
patients that such a process exists. Yet this approach is based on concepts. One cannot hear,
see or smell such a process. Till now, there even exists no clear comprehensive definition of
it. Most of the literature links it with working through, resistance and undoing repression
(Weinshel, 1988, Vaughan-Roose, 1995). From clinical experience and literature we
argument (see chapter 1) that such a model may fit for patients with a neurotic personality
organisation but falls short for patients with personality disorders. As an alternative we
developed in Chapter 1, a tentative three-dimensional model, in which we integrated the
former one by distinguishing two layers, a classic and a basic psychoanalytic process. This
model is not just a intellectual exercise, but a tool for daily work and for the organisation and
evaluation of the treatment of personality disordered patients.
Such treatments are often painful and long term and demand a lot of motivation,
engagements of patients, their families and therapists and they cost to the community.
Therefore we cannot retreat comfortably to ideological convictions, but it is necessary to try
to examine in an objective way, the psychoanalytic process which is at the heart in such a
treatment. Although necessary, it is however a somewhat grandiose project. The
96

97
psychoanalytic process is not only imperceptible by the senses, it varies considerably between
patients and it does not have a linear course but progresses and retreats. It is understandable
that many psychoanalysts think that only a heuristic approach of it is possible and that the
empirical approaches are doomed to fail, even worse that such an approach may damage this
process and our understanding of it.
We thought that our three-dimensional model (Chapter 1) offers a chance for a
comprehensive approach of the psychoanalytic process in personality disordered patients. In a
first phase we searched for scales measuring each of the three dimensions of the model. For
measuring the first dimensions of the subjective experience patients have of the safety of the
therapy, we had to create a scale. For the second dimension of mentalization, we made a scale
based on Bion's Grid. Seeing his GRID as a map with sveral zones of mental functioning
made this possible (see Chapter 2). Another facet of mentalization, reflective functioning
became more important in recent years - an advantage of it, being that it is already based on
empirical grounds with an existing scale, the RFS. For the third dimension, the object
relations, we found an empirical approach on the concepts of mental representations in the
work of Blatt and his DRS scale.
To gather material to rate these dimensions of the process, we choose an indirect and
psychoanalytical way in which patients could answer freely to questions. These questions
offer information about representations of self and others and are at the same time emotional
stimuli stimulating the mental functioning of the patients in the here and now. The ORI
offered these characteristics. The rating of the DRS on ORI-transcripts had already a proven
reliability and validity. The ORI questions such as ' tell my about your mother, tell me about
yourself ' provoke affects and their processing and make it possible to rate the Bion Grid
Scale on it. To measure the RF of Fonagy and colleagues we had to find a way to get
equivalents of the AAI demand questions which sollicit to think about relations. Therefore we
97

98
let patients freely associate in the first part of the interview for being able to rate the Bion
GRID scale and asked for an elaboration of the examples in a second part of the interview, a
substitute of the AAI demand questions, to have material to rate the RFS on it.
The next step was to make manuals of each of the scales. For the DRS we only needed a
translation, for the other scales new manuals were made to rate the scales on the ORI. After a
phase of training with the scales and try outs, scales and manual were further refined.
We found a sufficient inter-rater reliability as well for the global scores of the four
scales as for the scores of each component of the ORI. This proved the feasibility to measure
the concepts on which the model is based. Given the fact that the raters were of a different
background (students and experienced therapists), we may conclude that the rating by means
of the four manuals is reliable. This is remarkable as the DRS and the B-Grid scale are theory
driven and based on difficult concepts. The inter-rater reliability of the DRS on the ORI was
examined by Stayner (mentioned in Blatt & Auerbach, 2003) with a Shrout-Fleiss correlation
coefficient of .83., which equals the Shrout-Fleiss correlation coefficient which we found for
the DRS global score.
As the patients are blind to what is rated on their answers to the four open questions and
examples, in which there is no mention made of the process. This way chances of test-retest
confounds are minimised. The ORI's are rated on transcripts of the interviews and the raters
had no information about the patients, nor about the moment in therapy that interviews took
place. There were three raters for each scale, 12 raters in total, so that there is no confound
between the scoring of several measures.
In the next phase different raters for each of the 4 measures, rated 82 ORI to test the
measures for reliability and convergent and divergent validity.
98

99
The correlations between scores from each of the components of the ORI and mean
global scores proved to be high for each scale. Given this correlation and for reasons of space,
we will limit ourselves to the discussion of the global scores. This high correlation for each of
the four measures of the scores of the components of the ORI with the global scores, is an
argument that it may be possible to use global scores instead of the component scores when
studying the process with the three-dimensional model and to reduce the number of variables
this way.
When looking at the intercorrelation of the four measures, we found a correlation
between the global scores of the two measures of mentalization, which come from very
different backgrounds as discussed in Chapter 2. This intercorrelation is significant but not
strong enough to consider them as measuring the same aspects of mentalization. We find a
small correlation between RFS and FS, which is to be expected as they are, in fact, measuring
aspects of the psychoanalytic process that are theoretically supposed to influence each other
(Fonagy, Gergely et al., 2002). The remaining correlations between the four measures,
however, were not significant, which is an argument to use all four scales in measuring the
dimensions of the psychoanalytic process.
The global RFS and DRS scores as measured on the ORI were not associated by
sociodemographic factors such as age, gender, or educational level. This is not obvious for the
RFS which expresses the way of thinking about oneself and others, and for which an
influence of education and intelligence could haven been possible. This is the case, however,
for the BGS, which reflects the way that experiences are dealt with in a creative, associative
way. Education and age play a role here, which we shall take into account when refining the
scale and in the analysis of the data when using the Bion Grid Scale. There is a slight positive
association between age and the FS scale. A possible explanation may be that the mean age of
the patients in partial hospitalization was higher than that of patients in full hospitalization.
99

100
Patients in partial hospitalisation tend to show a lower level of pathology than do the patients
in full hospitalisation, certainly when admitted at a higher age.
To test the convergent and divergent validity of the four scales measured on the ORI, we
correlated them with symptom sales and scales related to personality structure and personality
categories. For the symptom related scales, the DRS correlates significantly with the
Symptom Check List (SCL) positive symptom total, with the Beck Depression Inventory
(BDI), the Spielberger Trait Anxiety (STAI) and Anger Inventories (STAXI). The correlation
with symptoms is even higher for the DRS self-score because the questions about the self
reflect more a present state of mind than the questions about father, mother, therapist. This
significant correlation with symptom scores may be explained by the fact that less integrated
mental representations give rise to more psychopathology and more symptoms. Furthermore,
of the other three ORI scales, it is only the RFS which shows a correlation - it correlates
negatively with anxiety as measured on the STAI. The fact that there are few correlations of
the ORI measures with symptom scales is not unexpected, the ORI scales are supposed to
measure inner psychic change and not symptoms. There might however be a methodological
explanation for these low correlations: the data are obtained by two different methods, the
ORI-scales are assessed on the transcripts of an interview, while the other scales are self-
report scales.
For the personality related scales, we would expect a correlation with the Inventory of
Personality Organisation (IPO), a scale about the level of psychic organisation. We found that
the DRS correlates in a significant way with all three IPO subscales. This is in line with the
fact that both scales are theory driven, relying for a great deal on Object Relations theory and
concepts such as splitting. The fact that the BGS and the FSS do not correlate may be
explained by the fact that both scales are less correlated with severity of personality
organisation and must be judged in the context of other data. Associative processing of
100

101
emotions, as we intend to measure with the BGS may be easier for a creative borderline
patient than for a rigid narcissistic patient, for instance, although the borderline patient may
present more severe psychopathology. In regard to the SCID-II results, we expect negative
correlations for the DRS, with the Axis II categories considered in terms of differentiation and
relatedness. This is the case for the avoidant and dependent categories in which the
relatedness dimension is prominent, and for the borderline and the paranoid categories, in
which the differentiation dimension is prominent. The positive significant correlation of Felt
Safety with the dependent category is as expected. The BGS scores correlate in a significant
negative way with the antisocial category, which can be accounted for by the fact that
antisocial personality disorders have the least mental processing of their emotional
experiences. The finding of the positive significant correlation of the RFS with the narcissistic
category and not with the borderline category is unexpected and raises questions about the
validity of measuring the RFS on the ORI. The RFS as measured on the AAI correlates
significantly negative with BPS (Fonagy et al., 1996) and the mentalization based treatment of
borderline personality disorders (Bateman and Fonagy, 2004) is based on enhancing the
reflective functioning in borderline patients. Furthermore, these findings were corroborated by
an own pilot study mentioned in Chapter 2, in which the RF as measured on the AAI
correlated as well significantly negative with BPS.
The project of rating the RFS on the ORI, was favoured by Fonagy , Target & Blatt
(preconference workshop, J.Sandler Research Conference 2001) as rating the RF on the AAI
is very time consuming and when our method proved to be valid, it would mean that the RFS
could be used on a larger scale in psychotherapy research. Therefore we examined the validity
in greater detail, correlating RF scores, rated on the AAI and rated on the ORI, ORI and AAI
being administered in the same week. The correlation between the RFS rated on the ORI and
the RFS rated on the AAI by a psychologist trained in the latter was not significant, which
101

102
means that we failed in our attempt to find an alternative for the original time-consuming
AAI-RFS rating. We were interested to know whether this was because of the difference in
manual and method of rating, because of the difference in raters or because of the type of
interview. This is explored in the third study. The reason of the lack in correlation was not
that the meaning of Reflective Functioning is not well respected in the ORI-RFS Leuven
manual compared with the UCL AAI-RFS manual; indeed the correlation between RFS rated
on the ORI by the psychologist trained with Fonagy and colleagues using this way of scoring
as much as possible and the RFS rated on the ORI using the Leuven manual is significant at
the p = .002 level. The reason was not the difference in raters, because when the UCL-trained
rater scores both the ORI and the AAI, the correlation between mean global RFS values
remains non significant, just as when the Leuven rater rates both the AAI and the ORI.
Moreover, the correlation between the two Leuven raters RF on ORI is high.
The major reason must be, therefore, the difference in interview, in material on which the
raters rate. Indeed the AAI offers 'permit’ questions and ‘demand’ questions in which the
subject is pushed to the limit to see whether she or he can reflect or not on self-other
relationships, mostly situations of loss in the past. In contrast, the ORI is composed of four
open questions about father, mother, self, and therapist and often concern present
relationships. We tried to compensate for the lack of demand questions by asking patients to
elaborate on examples in a second part of the ORI interview, safeguarding the possibility for
associative exploration in the first part of the interview. This way of dealing with these
examples was clearly not demanding enough to be an alternative to the demand questions of
the AAI. According to these results, the RFS training with Fonagy and colleagues plays a role
as well, but to a lesser degree than the interview format itself.
It appears like that with the ORI-RFS, present or state Reflective Functioning is
measured during the interview, while the AAI-RFS measures the capacity for Reflective
102

103
Functioning. This may explain as well the significant correlation of the ORI-RFS with the
SCID-II category of narcissism: it is peculiar for this category to use more rationalisation and
cognitive processing of their experiences, rather than being highly emotionally involved.
In conclusion, we may say that all scales may be reliably rated on the ORI, in their
present state. The validity of the DRS is further established with this study. Felt Safety is a
new scale, and all correlations are in the direction which we expected, which is a strong
indication of its validity. The BGS is promising; the scores may best be interpreted in relation
with other results. The ORI-RFS did not prove to be an alternative for the time consuming
AAI-RFS but is a reliable instrument in itself, which measures the state Reflective
Functioning rather than the capacity for Reflective Functioning, as the AAI-RFS does. The
ORI-RFS and the BGS correspond with the way we conceptualised the two sides of
mentalization (Chapter 2).
103

104

105
Chapter 4
Het Meten van Psychische Verandering met de Leuvense
Psychotherapie Schaal: het Perspectief van de Patiënt.3
Wanneer men bij persoonlijkheidsstoornissen onderzoek doet naar de effectiviteit van
verschillende therapieën, wordt vaak een beroep gedaan op metingen van veranderingen van
klinische symptomen (Bateman & Fonagy, 1999, 2001 en Linehan et al., 1993). Specifieke
verdiensten van specifieke therapieën zijn door deze focus op symptoomreductie niet steeds
vaststelbaar (Luborsky et al., 1994), zeker niet wanneer de focus van de therapie breder is dan
enkel maar symptoomreductie. Daarom is een focus op de psychologische processen
onderliggend aan deze symptoomvermindering aangewezen. In het vorig hoofdstuk bespraken
we de meetinstrumenten die vanuit psychoanalytische hoek ontwikkeld zijn om de innerlijke
psychische processen te meten vanuit het oogpunt van een extern onderzoeker. We wilden
ook een meting van het therapeutisch veranderingsproces vanuit het oogpunt van de patiënt,
maar op een manier die niet symptoomgericht was. Vermits de schalen over het
veranderingsproces die door therapeuten of door externe onderzoekers gescoord worden
meestal vanuit de theorie gevormd zijn, leek een empirisch gevormde schaal het meest
geschikt voor ons doel. Empirisch gevormde schalen om het innerlijk veranderingsproces bij
de behandeling van persoonlijkheidsstoornissen te meten vanuit het standpunt van de patiënt
zelf bestaan nog niet.
Het maken van een zelfrapporteringsschaal die een innerlijke verandering poogt weer te
geven die kan optreden tijdens een intensieve klinische psychotherapie en die niet vanuit de
3 An English translation is available on request.
105

106
theorie maar vanuit de ervaring van patiënten zelf uitgaat, was de doelstelling van het
onderzoek.
In het huidig onderzoek worden de constructie van de nieuwe psychotherapieschaal en
de waarde van deze schaal besproken in een vijftal studies.
Studie 1: Constructie van de Experimentele Versie van de Leuvense Psychotherapie Schaal
Om een schaal te construeren uitgaande vanuit de ervaring van de cliënten, steunden we
ons op de uitspraken van cliënten zelf, zoals weergegeven door alle teamleden van een
afdeling klinische psychotherapie.
Proefgroep
Aan het team van de afdeling voor Klinische Psychoanalytische Psychotherapie van
Persoonlijkheidsstoornissen (KLIPP) van het Universitair Centrum St. Jozef te Kortenberg
(Vermote, 1997, Vermote en Vansina, 1998, Pieters en Vermote, 2002), beschreven in
hoofstuk 1, werd gevraagd uitspraken van patiënten te verzamelen die iets zeggen over hoe ze
veranderingen tijdens klinische psychotherapie ervaren. De 11 teamleden die meewerkten
waren vijf verpleegkundigen, de creatieve therapeut, de muziektherapeut, de
bewegingstherapeut, twee psychotherapeuten en de psychiater.
106

107
Procedure
Er werd aan deze teamleden gevraagd om uitspraken op te schrijven zoals ze die van
patiënten horen en die een verandering weerspiegelen die de patiënt ervaart tijdens de
therapie. Vijf verpleegkundigen verzamelden 100 uitspraken, de groepstherapeut 17, de
creatieve therapeut 21, de psychomotorisch therapeut 39 en de psychiater 38.
Opbouw van de Experimentele Versie
Alle uitspraken werden op afzonderlijke kaartjes genoteerd, waarna dubbels of gelijke
inhouden eruit gehaald werden. Vervolgens werd de structuur van de items aangepast, zodat
het uitspraken werden met volgende kenmerken: de items geven niet de verandering zelf aan,
maar een toestand die kan veranderen. De graad van aanwezigheid van deze toestand wordt
aangeduid op een Likert-schaal met vijf antwoordmogelijkheden gaande van helemaal oneens
(score 1) tot helemaal eens (score 5). De items werden omgezet tot enkelvoudige items (die
maar één ervaring weergeven), en geformuleerd zijn in de ik-vorm en in de tegenwoordige
tijd. Uitspraken die twee ervaringen weergaven werden gesplitst of weggelaten. Alle
frequentie–aanduidingen zoals ‘soms’, ‘dikwijls’, ‘altijd’, ‘nooit’, werden weggelaten. De
items kunnen positieve ofwel negatieve ervaringen weergeven, maar worden steeds in
positieve zin geformuleerd (dus: ik ben triestig of ik ben gelukkig, maar niet: ik ben niet
gelukkig of ik ben niet triestig). Op basis van deze criteria werden 26 items van de 215
verzamelde items niet opgenomen. Daardoor ontstond de experimentele versie van de
Leuvense Psychotherapie Schaal, bestaande uit 189 items opgesplitst in twee delen 1.
uitspraken over jezelf en 2. uitspraken over de afdeling.
107

108
Studie 2: Constructie en kenmerken van de definitieve versie van de LPS
Voor de schaalconstructie werd op exploratieve wijze nagegaan welke psychologische
constructen gemeten worden door de experimentele versie van de LPS en hoe betrouwbaar
deze metingen zijn. Op grond van dit resultaat werden de definitieve schalen samengesteld.
De eigenschappen van de schaal en de invloed van demografische factoren en psychiatrische
diagnose werden nagegaan.
Methode
Proefgroep
De proefgroep werd samengesteld uit patiënten die een behandeling volgen op een
residentiële therapeutische afdeling in negen centra uit Vlaanderen. In verband met de
uitspraken over de afdeling diende de opnameduur minstens een week te bedragen en
patiënten die een acute psychotische episode doormaakten of mentaal geretardeerd waren,
werden uitgesloten. Vragenlijsten die onvoldoende volledig ingevuld waren, werden niet
meegenomen in het onderzoek.
Dit leverde een onderzoeksgroep op van 482 deelnemers, waarvan 180 mannelijke
(37.3%) en 292 vrouwelijke (60.6%) patiënten. Van 10 patiënten (2.1%) bleef het geslacht
onbekend. Er werden vijf verschillende leeftijdscategorieën onderscheiden: 17-20 jaar (12%),
21-30 jaar (36.7%), 31-40 jaar (24.4%), 41-50 jaar (17.8%) en 51-65 jaar (4.9%). Wat de
burgerlijke status betreft, was 58.3% van de patiënten ongehuwd, 20.1% was gehuwd, 13.9%
was gescheiden en 5.2% was samenwonend. Van 12 patiënten (2.5%) was de burgerlijke staat
onbekend. Wat het opleidingsniveau betreft, was 4.8% enkel in het bezit van een diploma
108

109
lager onderwijs 47.7% van een diploma middelbaar onderwijs, 28.2% had het hoger onderwijs
succesvol doorlopen en 10% bezat een universitair diploma. Van 44 deelnemers (9.1%) bleef
het opleidingsniveau onbekend.
Het merendeel van de deelnemers was minder dan drie maanden in opname (70.8%). Bij
8.3% van de deelnemers was de opnameduur minder dan zes maanden, bij 4.4% was deze
minder dan negen maanden, en bij 3.5% minder dan één jaar. De opnameduur was langer dan
één jaar bij 3.6% van de deelnemers en onbekend bij 9.4% van de deelnemers.
Tijdens de opname volgden 80.1% van de deelnemers een psychoanalytisch of
psychodynamisch georiënteerde psychotherapie, 15.5% een gedragstherapeutische
behandeling en 4.4% een systeemtheoretische therapie. Verder kreeg 65.1% van de
deelnemers alleen groepstherapie, 31% zowel groeps- als individuele therapie en 4% enkel
individuele therapie.
Wat de aanwezigheid van As-I diagnoses binnen de onderzoeksgroep betreft, ziet de
vertegenwoordiging er als volgt uit: Stemmingsstoornis: 49.6% (239 patiënten),
Angststoornis: 7.9% (38 patiënten), Stoornis gebonden aan middelengebruik: 7.6% (37
patiënten), Schizofrenie en andere psychotische stoornissen: 7.2% (35 patiënten),
Aanpassingsstoornis: 7% (34 patiënten), Eetstoornis: 3% (14 patiënten), Stoornis in de
impulscontrole: 1.5% (7 patiënten), Dissociatieve stoornis: 1.2% (6 patiënten), Somatoforme
stoornis: 0.8% (4 patiënten), Andere stoornissen: 3.2% (15 patiënten). Bij 44 deelnemers werd
geen As-I stoornis vastgesteld en 9 deelnemers werden geclassificeerd in de categorie
Uitgestelde of onbekende As-I diagnose.
Wat de aanwezigheid en vertegenwoordiging van As-II diagnoses betreft, werden 286
patiënten (59.3%) van de onderzoeksgroep gediagnosticeerd aan de hand van een Cluster B
persoonlijkheidsstoornis, waarvan aan 188 patiënten (39%) een borderline
persoonlijkheidsstoornis, aan 50 (10.4%) een narcistische, aan 42 (8.7%) een theatrale en aan
109

110
6 (1.2%) een antisociale persoonlijkheidsstoornis toegeschreven werd. Bij 51 patiënten
(10.6%) werd een Cluster C persoonlijkheidsstoornis vastgesteld, bestaande uit 28 patiënten
(5.8%) met een afhankelijke persoonlijkheidsstoornis, 12 patiënten (2.5%) met een obsessief-
compulsieve en 11 patiënten (2.3%) met een vermijdende persoonlijkheidsstoornis. Tot de
groep met een Cluster A persoonlijkheidsstoornis behoorden 26 patiënten (5.4%), waarvan 11
(2.3%) met een schizoïde, eveneens 11 met een schizotypische en 4 (0.8%) met een paranoïde
persoonlijkheidsstoornis. Bij 59 patiënten (12.2%) werd geen As-II stoornis vastgesteld en 60
deelnemers (12.5%) werden geclassificeerd in de categorie Uitgestelde of onbekende As-II
diagnose.
Instrument
De experimentele LPS uit studie 1 met 189 items.
Analyse
Schaalconstructie.
Op de resultaten van de afname van de experimentele LPS op de proefgroep werden de
dimensies van de LPS bepaald met een Principale Componenten Analyse gevolgd door
VARIMAX-rotatie. Er werd gezocht naar de meest eenvoudige en best interpreteerbare
oplossing, waarbij een voldoende aantal ladende items op de verschillende factoren werd
nagestreefd. Voor de interpretatie werden alle ladingen hoger dan.30 in rekening gebracht.
Het samenstellen van de voorlopige schalen gebeurde op basis van dit factorpatroon. Een item
dat op meer dan één factor.30 laadde, werd enkel opgenomen als het verschil in lading meer
110

111
dan.15 bedroeg. In dat geval behoorde het item tot die schaal, waar het de hoogste lading op
had. De interne consistentie werd gemaximaliseerd door repetitief Chronbach’s alpha
coëfficiënt te berekenen, met een successief weglaten van items die een lage correlatie met de
schaal vertoonden, totdat de alpha coëfficiënt niet meer steeg.
Op die manier werden de definitieve schalen samengesteld. Hiervan werden de
gemiddelde scores, de standaarddeviaties en de intercorrelaties nagegaan. Het benoemen van
de schalen gebeurde in de positieve richting, zodat de hoogste score de meest gezonde positie
aanduidt.
Invloed van demografische variabelen en psychiatrische diagnose.
De invloed van demografische variabelen (geslacht, leeftijd, opleiding en burgerlijke
staat) en psychiatrische diagnose (zoals door de behandelende psychiater in het medisch
dossier vermeld) werden met variantie-analyse onderzocht volgens de GLM-procedure zonder
uitspraak te doen over de richting van het effect. Multivariate- (Wilks’ Lambda) en univariate
analyses, als ook post-hoc toetsen (volgens het principe van Tukey) werden uitgevoerd.
Resultaten
Factoranalyse van de Experimentele LPS
Een exploratieve factoranalyse met varimaxrotatie leidt op grond van scree-test en
interpretatiecriteria tot een acht factorenoplossing. Deze verklaart 38.62% van de initiële
totale variantie. De acht factoren met telkens het percentage van de verklaarde variantie en de
hoogstladende items zijn de volgende:
111

112
1. positief welbevinden (14.76%): ‘Ik ben depressief’, ‘Ik voel me goed’, ‘Ik voel me leeg’ en
‘Ik heb zin in het leven’.
2. openheid tegenover anderen (5.02%): ‘Ik kan praten over mijn gevoelens’, ‘Ik durf iets van
mezelf te laten zien’ en ‘ Ik kan moeilijk mensen dicht bij mij verdragen’.
3. autonomie (4.98%): ‘Ik heb veel bescherming van anderen nodig’, ‘Ik kan tegen een duw’
en ‘Ik durf weinig’.
4. affectregulatie en afwezigheid van splitsing (3.34%): ‘Ik voel nu nog een oud verdriet’,
‘Het verleden achtervolgt mij’ en ‘ Ik ben voor iemand of tegen’.
5. zelfgerichtheid (3.19%): ‘Ik ben meer met de noden van anderen bezig dan stil te staan bij
wat ik zelf wil’, ‘De mening van anderen vind ik erg belangrijk’ en ‘Negatieve
opmerkingen van anderen raken mij niet’.
6. positieve ervaring van de afdeling (3.10%): ‘Ik wil hier weg’ en ‘De afdeling is een beetje
mijn thuis’.
7. ervaring van steun en erkenning (2.14%): ‘De mensen zijn tegen mij’ en ‘Ik ervaar echte
bekommernis’.
8. controle over het gebruik van genotsmiddelen (2.09%): ‘Ik heb genotsmiddelen nodig’ en
‘Ik kan gemakkelijk stoppen met het gebruik van genotsmiddelen’.
Schaalsamenstelling van de LPS
Met de beschreven procedure werden de voorlopige schalen uit dit factorpatroon
afgeleid. Van de oorspronkelijke 189 items bleven op die manier 132 items over. Door het
maximaliseren van de interne consistentie verdwenen nog eens 20 items, waardoor de
definitieve schaal uit 112 items bestaat. Op basis van de itemanalyse konden we zeven
schalen weerhouden, waarvan de alpha coëfficiënt groter is dan.50 (Tabel 1). De
112

113
oorspronkelijke schaal zeven ‘Ervaring van steun en erkenning’ bleek slechts uit drie items te
bestaan en niet intern consistent te zijn (α = .07) en werd niet weerhouden.
Tabel 1
Itemverdeling Voorlopige en Definitieve Versie van de Leuvense Psychotherapie Schaal Met
Vermelding Chronbach α
Subschaal Voorlopig Definitief α
Schaal ‘Positief welbevinden’ 61 57 0.97
Schaal ‘Openheid tegenover anderen’ 18 18 0.85
Schaal ‘Autonomie’ 11 7 0.81
Schaal ‘Affectregulatie’ 10 9 0.64
Schaal ‘Gerichtheid op zichzelf’ 10 5 0.64
Schaal ‘Positieve ervaring afdeling’ 13 13 0.82
Schaal ‘Ervaring van steun’ 3 0
Schaal ‘Genotsmiddelen’ 6 3 0.85
Totaal 132 112
113

114
Basiskenmerken van de LPS
In tabel 2 geven we de basiskenmerken van de zeven definitieve schalen, waaruit blijkt
dat er voldoende variabiliteit is op de schalen.
Tabel 2.
Gemiddelde, Standaarddeviatie, Minimum- en Maximumscore van de Zeven Definitieve
Schalen van de Leuvense Psychotherapie Schaal
In tabel 3 geven we de intercorrelaties tussen de schalen. Daaruit blijkt dat alle schalen
positief met elkaar correleren. De hoogste correlatie vinden we tussen S1 ‘Positief
welbevinden’ en S3 ‘Autonomie’ met r = .63. Verder is er ook een hoge correlatie tussen S1
‘Positief welbevinden’ en S2 ‘Openheid tegenover anderen’ met r = .57, en tussen S1 en S4
‘Affectregulatie’ met r = .54. De ‘Positieve ervaring met de afdeling’ heeft de laagste
positieve verbanden met de andere schalen, die alle persoonlijke ervaringen weergeven.
Subschaal N M SD Min Max
‘Positief welbevinden’ (PW) 482 2.76 0.76 1.21 4.89
‘Openheid tegenover anderen’(OA) 482 3.27 0.61 1.50 4.78
‘Autonomie’ (AUT) 482 2.78 0.83 1.00 5.00
‘Affectregulatie’ (AR) 482 2.61 0.58 1.00 4.20
‘Gerichtheid op zichzelf’ (GZ) 482 2.53 0.73 1.00 5.00
‘Positieve ervaring afdeling’(PEA) 482 3.40 0.65 1.23 4.92
‘Genotsmiddelen’ (GEN) 482 3.06 1.33 1.00 5.00
114

115
Tabel 3
Intercorrelaties Tussen de Schalen van de Leuvense Psychotherapie Schaal
*p <0.05, **p <0.01, ***p <0.001
De LPS en Demografische Variabelen
De multivariate variantie analyse (MANOVA) toont dat er een significant effect is van
geslacht op het antwoordpatroon met Wilks’ Lambda F(7, 463) = 6.075 met p < .0001.
Mannen scoren gemiddeld genomen hoger dan vrouwen op S1 ‘Positief welbevinden’ (2.97
vs. 2.64), S2 ‘Openheid tegenover anderen’ (3.37 vs. 3.21), S3 ‘Autonomie’ (3 vs. 2.65), S4
‘Affectregulatie’ (2.80 vs. 2.55) en S5 ‘Gerichtheid op zichzelf’ (2.74 vs. 2.39).
Er is geen effect van leeftijd verdeeld over vijf leeftijdsgroepen, met Wilks’ Lambda
F(28, 1631) = 1.305 met p = .132. Ook een univariate analyse van de zeven schalen
afzonderlijk toont geen effect van leeftijd.
Subschaal 1 2 3 4 5 6 7
1. Positief welbevinden _ 0.57*** 0.63*** 0.54*** 0.38*** 0.27*** 0.34***
2. Openheid tegenover anderen _ 0.36*** 0.26*** 0.22*** 0.31*** 0.13**
3. Autonomie _ 0.43*** 0.26*** 0.08 0.22***
4. Affectregulatie _ 0.36*** 0.16*** 0.27***
5. Gerichtheid op zichzelf _ 0.04 0.08
6. Positieve ervaring afdeling _ 0.10*
7. Genotsmiddelen _
115

116
De drie opleidingsgroepen (lager, middelbaar en hoger onderwijs) tonen geen effect aan
van opleiding, met Wilks’ Lambda F(14, 856) = 0.956 met p = .50, ook niet wanneer de
schalen afzonderlijk bekeken worden.
Er is een significant effect van burgerlijke staat (verdeeld in vier groepen (ongehuwd,
samenwonend, gehuwd en gescheiden) met Wilks’ Lambda F(21, 1318) = 3.052 met p <
.0001). Een univariate analyse van de zeven schalen afzonderlijk toont dat burgerlijke staat
een significant effect heeft op S2 ‘Openheid tegenover anderen’ met F(3, 465) = 2.815 met p
< .05, op S3 ‘Autonomie’ met F(3, 465) = 2.934 met p < .05, op S4 ‘Affectregulatie’ met
F(3, 465) = 5.840 met p < .01 en op S6 ‘Positieve ervaring van afdeling’ met F(3, 465) =
5.028 met p < .01.
De LPS en de DSM IV As-II Persoonlijkheidsdiagnoses
Alleen van de DSM IV-as II diagnoses die voldoende frequent aanwezig zijn in de
proefgroep (borderline, narcistische, theatrale en afhankelijke persoonlijkheidsstoornis) kan
het effect op de beantwoording van de LPS worden berekend. De andere as II diagnoses zijn
niet voldoende vertegenwoordigd in de proefgroep, evenmin als de DSM IV- as I diagnoses.
Uit de multivariate analyse van de vier bestudeerde groepen persoonlijkheidsstoornissen
blijkt dat er een significant effect is van de As-II stoornis, met Wilks’ Lambda F(21, 850) =
3.82, p < .0001. De univariate analyse van de zeven schalen toont aan dat er een significant
effect is op S1 ‘Positief welbevinden’ met F(3, 302) = 5.56, p < .01, op S2 ‘Openheid
tegenover anderen’ met F(3, 302) = 3.30, p < .05, op S3 ‘Autonomie’ met F(3, 302) = 8.97, p
< .0001, op S4 ‘Affectregulatie’ met F(3, 302) = 7.37, p < .0001 en op S7 ‘Controle over
genotsmiddelen’ met F(3, 302) = 3.05, p < .0001. De post-hoc toets van Tukey geeft aan dat
de narcisten significant gemiddeld hoger scoren op S1 ‘Positief welbevinden’(3.1) dan
116

117
borderlines(2.57) en afhankelijken (2.55), die onderling niet significant verschillen. Op S2
‘Openheid tegenover anderen’ en op S 7 ‘Controle over genotsmiddelen’ is er een significant
verschil tussen narcisten (3.43 en 3.30) en borderlines (2.57 en 2.73). Op S 3 ‘Autonomie’
scoren de borderlines (2.76) significant hoger dan de theatralen (2.22). Verder blijkt dat de
narcisten (3.08) op deze schaal een significant hogere score behalen dan de theatralen (2.22)
en afhankelijken (2.57). Op S 4 ‘Affectregulatie’ is de gemiddelde score van de narcisten
(2.90) significant hoger dan die van borderlines (2.47) en theatralen (2.50). Hoewel de
univariate analyse aangaf dat er geen significant effect van diagnose was op schaal 5
‘Gerichtheid op zichzelf’, blijkt uit de post-hoc toets van Tukey dat er op deze schaal een
significant verschil is tussen narcisten (2.72) en borderlines (2.42).
Samenvattend kunnen we stellen dat narcisten gemiddeld een hogere score op de LPS
behalen dan de andere drie As-II diagnoses, die onderling niet veel verschil tonen. De
borderlines scoren enkel op schaal 3 ‘Autonomie’ significant hoger dan de theatralen.
Studie 3: Congruente Validiteit: de LPS en de Neo-Pi-R
Methode
Proefgroep
Bij zesenvijftig deelnemers uit de proefgroep van studie twee waarvan de leeftijd
varieerde tussen 17 en 65 jaar werd ook de NEO-PI-R afgenomen. Onder hen bevonden zich
39 vrouwen. 16 deelnemers waren gehuwd of samenwonend, 14 gescheiden, 26 ongehuwd. 3
volgden lager onderwijs, 31 middelbaar, 15 hoger en van 7 deelnemers was de opleiding
onbekend. Diagnostisch werd aan 47 deelnemers de borderline persoonlijkheidsstoornis
117

118
toegeschreven. Vier deelnemers hadden een narcistische persoonlijkheidsstoornis, 3 hadden
een histrionische persoonlijkheidsstoornis, 1 een afhankelijke persoonlijkheidsstoornis en 1
onbekend.
Instrument
De Neo-PI-R werd afgenomen als algemene persoonlijkheidsvragenlijst met
aangetoonde begripsvaliditeit. We maakten gebruik van de Nederlandstalige NEO-PI-R
persoonlijkheidsvragenlijst (Hoekstra, 1996). Deze vragenlijst telt 240 items, die elk vijf
antwoordmogelijkheden hebben, gaande van helemaal oneens tot helemaal eens. De NEO-PI-
R meet de vijf belangrijkste dimensies van de persoonlijkheid volgens het vijf factoren model
van Costa en Widiger, 1994, en binnen elk van die vijf domeinen een zestal facetten, die de
dimensie in kwestie definiëren. De schalen van de NEO-PI-R zijn neuroticisme, extraversie,
openheid, altruïsme, consciëntieusheid.
Analyse
We berekenden een Pearson Correlatie Coëfficient van alle scores van de LPS schalen
met alle scores van de NEO-PI-R.
Resultaten
S1 ‘positief welbevinden’ van de LPS correleert negatief met de N ‘neuroticisme’ schaal
van de Neo-PI-R (-.27, p < .05), vooral met N 2 ‘ergernis’ (-.37, p < .01) en N3 ‘depressie’
118

119
(.36, p < .01). De correlatie is positief met de E ‘extraversie’ schaal (.37, p < .01), vooral met
E1 ‘hartelijkheid’ (.25, p < .25), E4 ‘energie’ (.35, p < .01) en E6 ‘vrolijkheid’ (.26, p < .05).
Verder is er nog een significante correlatie met de C5 ‘zelfdiscipline’ (.24, p < .05) van de C
‘conscientieusheid’ schaal.
S2 ‘openheid tegenover anderen’ van de LPS correleert significant positief met de E
‘extraversie’ schaal van de Neo-Pi-R (.37, p < .01), vooral met E1 ‘hartelijkheid’ (.30, p <
.05) en E5 ‘avonturisme’(.23, p < .05). Er is verder een positieve correlatie met de O3
‘gevoelens’ (.28, p < .05) uit de O ‘openheid ‘ schaal. De A ‘altruïsme’ schaal correleert
eveneens significant positief (.27, p < .05), met vooral A2 ‘oprechtheid’(.29, p < .05) en A4
’inschikkelijkheid’ (.23, p < .05) en A6 ‘medeleven’ (.26, p < .05). Uit de C
‘consciëntieusheid’ schaal correleert C6 ‘bedachtzaamheid’ (.27, p < .05).
Met S3 ‘ autonomie’ van de LPS correleren de volgende schalen uit de Neo-Pi-R
significant: N3 ‘ergernis’ (.37, p < .01) en N6 ‘ kwetsbaarheid’ (.32, p <01) uit de N
‘neuroticisme’ schaal. E 4 ‘ energie’ (.37, p < .01) en E6 ‘vrolijkheid’ (.25, p < . 05) uit de E
‘extraversie’ schaal. Verder de O ‘openheidsschaal’ (.40, p < .05) met O1’ fantasie’(.26, p <
.05) en O2 ‘esthetiek’(.25, p < .05) en O3 ‘gevoelens’(.22, p < .05) en O4
‘veranderingen’(.31, p < .01). Uit de C ‘consciëntieusheid’ schaal, de C1 ‘doelmatigheid’
(.29, p < .05)en de C5 ‘zelfdiscipline’(.23, p < .05).
De S4 ‘ affectregulatie’ van de LPS correleert significant negatief met de N
‘neuroticisme’schaal van de Neo-Pi-R (-.38, p < .01), vooral met N1 ‘angst’ (-.23, p < .05),
N2’ergernis’(-55, p < .01), N5 ‘impulsiviteit’ (-.27, p < .01). Er is een positieve correlatie met
E1’hartelijkheid’ (.26, p < .05) uit de E ‘extraversie’schaal en C3 ‘betrouwbaarheid’(.23, p <
.05) uit de C ‘ consciëntieusheid’ schaal en een negatieve correlatie met de A2
‘oprechtheid’(.25, p < .05) uit de A ‘altruïsme’schaal.
119

120
S5 ‘gerichtheid op zichzelf’ uit de LPS correleert sterk met de N ‘neuroticisme’ schaal
(.35, p < .01), vooral met N2 ‘ergernis’(.27, p < .05) en N5 ‘impulsiviteit’ (.32, p < .01).
Verder E6 ‘avonturisme’ (.41, p < .01) uit de E ‘extraversieschaal’ en C3
‘betrouwbaarheid’(.37, p < .01) uit de C ‘consciëntieusheid’ schaal.
S6 ‘ positieve ervaring van de afdeling’ uit de LPS correleert significant met E6
‘vrolijkheid’ uit de E ’extraversie’ schaal en met A1 ‘vertrouwen’ (.34, p < .01) uit de A
‘altruïsme’ schaal van de Neo-Pi-R.
S7 ‘ controle over het gebruik van genotsmiddelen’ van de LPS correleert negatief met
de N ‘neuroticisme’ schaal (-.30, p < .05) van de Neo-Pi-R, vooral met N1 ‘angst’(-.24, p <
.05), N3 ‘depressie’ (-.38, p < .01), N4 ‘schaamte’ (-.32, p < .01). Verder is er nog een
significante correlatie met C1 ‘doelmatigheid’ (.33, p < .01) van de C ‘ consciëntieusheid’
schaal.
Studie 4: Congruente Validiteit van de LPS.
Naast de correlatie van de LPS met een dimensionele persoonlijkheidsdiagnose (Neo-PI-
R) werd ook de correlatie met een beschrijvende categoriale diagnose (SCID) en met een
structurele diagnose (IPO) nagegaan.
Methode
Proefgroep
Deze steekproef bestond uit 82 deelnemers (54 vrouwen en 28 mannen) bij opname op
de klinische psychotherapie afdeling voor persoonlijkheidsstoornissen in het U.C. St.-Jozef te
120

121
Kortenberg (Vermote & Vansina, 1998, Pieters & Vermote, 2002, Hoofdstuk 1), 58
residentieel en 24 in daghospitaal. De patiënten waren tussen 17 en 58 jaar met een
gemiddelde leeftijd van 27,1 jaar (SD = 8.5). Tweeëndertig deelnemers waren alleenwonend,
29 woonden bij hun ouders, 17 waren gehuwd of samenwonend, 3 deelnemers waren
gescheiden en 1 woonde in een gemeenschapshuis. Wat het studieniveau betreft volgden 3
deelnemers alleen het lager onderwijs, 32 volgden het secundair of technisch onderwijs en 47
hoger of universitair onderwijs. Op basis van de SCID II wezen we alle patiënten toe aan één
van de drie clusters van de DSM IV, as II: cluster A: 2 (1 schizoid, 1 schizotypisch); cluster
B: 62 (1 histrionisch, 44 borderline, 10 narcistisch, 4 narcistisch en borderline, 3 histrionisch
en borderline); cluster C: 10 (1 ontwijkend, 2 afhankelijk, 1 ontwijkend en afhankelijk, 6 niet
anders omlijnd); clusters A+B: 1;clusters B+C: 10.
Volgens het medisch dossier waren er 33 diverse stemmingsstoornissen, met 7 maal een
majeure depressieve stoornis, eenmalig voorkomen en 11 maal een majeure depressieve
stoornis, matig, heroptredend als meest voorkomende en 5 angststoornissen en 14 maal een
aanpassingsstoornis, met gemengde angst en depressieve gevoelens als meest voorkomende.
Instrumenten
Het Structured Clinical Interview for DSM-III Axis II Personality Disorders (SCID-II)
is een semi-gestructureerd interview ontworpen om een betrouwbare DSM-IV diagnose te
stellen. In het onderzoek werd de SCID-II Persoonlijkheidsvragenlijst gebruikt, een
Nederlandse vertaling en aanpassing voor de DSM IV door Weertman, Arntz & Kerkhofs,
2000.
De Inventory of Personality Organization (IPO) is een zelfrapportage vragenlijst
gebaseerd op psychodynamische constructen die afgeleid werden van Kernbergs theorie over
121

122
persoonlijkheidsorganisatie en persoonlijkheidsstoornissen (Kernberg, O., Clarkin, J., 1995).
Tot op heden werd de IPO bij meer dan 1500 personen afgenomen. De schalen vertonen een
goede tot excellente interne consistentie en test-hertest betrouwbaarheid. Constructvaliditeit
werd aangetoond door factor analyse. Daarenboven kon de IPO patiënten met As II
pathologie differentiëren in neurotische, high en low borderline categoriëen van
persoonlijkheidsorganisatie, zoals door de theorie is verondersteld (Lenzenweger et al., 2000).
Het onderzoek naar de constructvaliditeit werd door Normandin et al. (2002) gerepliceerd. De
IPO werd in het Nederlands vertaald en gevalideerd met exploratieve factoranalyse (Vermote
et al., 2003) en met confirmatorische factoranalyse (Vermote, Smits, Claes en Vertommen,
2005).
Analyse
Een Pearson Correlatie Coëfficient werd berekend van alle gemiddelde scores van de
LPS schalen met alle gemiddelde scores van de SCID II en van de IPO.
Resultaten
Deze resultaten worden weergegeven in tabel 4 en 5.
122

123
Tabel 4
Correlaties Tussen de Gemiddelde Scores van de LPS –schalen en de Gemiddelde Scores Op
de SCID II Categorieën
Subschaal LPS Subschaal
SCID 1 2 3 4 5 6 7 M
Ontwijkend -0.53*** -0.52*** -0.48*** -0.16 -0.29** -0.09 -0.27* -0.55***
Afhankelijk -0.41*** 0.05 -0.35** -0.28* -0.21 0.03 -0.39*** -0.43***
Obs.-Comp. -0.28* -0.15 -0.24* -0.23* -0.16 -0.11 0.02 -0.22*
Passief Agr. -0.26* 0.01 -0.15 -0.53*** -0.20 -0.07 -0.24* -0.35**
Depressief -0.58*** -0.21 -0.48*** -0.45*** -0.34** -0.18 -0.37*** -0.61***
Paranoid -0.38*** -0.24* -0.20 -0.58*** -0.36*** -0.03 -0.29** -0.51***
Schizotypisch -0.31** -0.35** -0.23* -0.15 -0.21 0.10 -0.12 -0.32**
Schizoid -0.38*** -0.44*** -0.30** -0.24* -0.09 -0.05 -0.09 -0.35**
Theatraal 0.11 0.37*** 0.16 -0.19 -0.04 0.05 -0.12 0.04
Narcistisch -0.01 0.04 0.03 -0.33** 0.12 0.02 -0.12 -0.07
Borderline -0.45*** -0.14 -0.38*** -0.52*** -0.38*** -0.10 -0.33** -0.56***
Antisociaal -0.06 -0.12 -0.10 -0.30** -0.03 -0.10 -0.22* -0.22*
Som -0.49*** -0.23* -0.38*** -0.57*** -0.31** -0.07 -0.36*** -0.59***
(n = 82) *p <0.05, **p <0.01, ***p <0.001
123

124
Tabel 5
Correlaties Tussen de Gemiddelde Scores Op de Schalen van de LPS en de Gemiddelde
Scores Op de Schalen van de IPO
Subschaal LPS
Subschaal IPO 1 2 3 4 5 6 7 M
Primitieve defensies -0.41*** -0.20 -0.17 -0.58*** -0.49*** -0.05 -0.37*** -0.57***
Identiteitsdiffusie -0.50*** -0.16 -0.37*** -0.49*** -0.42*** -0.03 -0.35** -0.58***
Realiteitstoetsing -0.19 -0.13 -0.09 -0.30** -0.16 -0.07 -0.07 -0.22*
Som-score -0.46*** -0.23* -0.28* -0.57*** -0.40*** -0.11 -0.32** -0.57***
(n = 82) *p <0.05, **p <0.01, ***p <0.001
Schaal 1 van de LPS (positief welbevinden) correleert significant negatief met de
ontwijkende, afhankelijke, depressieve, paranoïde, schizoïde en borderline categorieën van de
SCID II, Schaal 2 (openheid naar anderen) correleert significant negatief met de ontwijkende,
schizoïde, schizotypische en positief met de theatrale categorieën van de SCID II. Schaal 3
(autonomie) correleert in hoge mate significant negatief met de ontwijkende, afhankelijke,
depressieve, schizoïde, borderline en in mindere mate met obssessief-compulsieve en
schizotypische categorieën van de LPS. Schaal 4 van de LPS (aanwezigheid en omgaan met
negatieve gevoelens) correleert significant negatief met de passief-agressieve, depressieve,
paranoïde, borderline, antisociale en narcistische categorieën van SCID II. Schaal 5 van de
LPS (gerichtheid op zichzelf) correleert significant negatief met de borderline, paranoïde en
de depressieve en ontwijkende categorieën van de SCID II. Schaal 6 van de LPS (positieve
gevoelens tegenover de afdeling) heeft geen significante correlatie met categorieën van de
SCID II. Schaal 7 van de LPS (gebruik van genotsmiddelen) correleert het meest significant
124

125
negatief met de depressieve, ontwijkende, borderline, paranoïde en afhankelijke categorieën
van de SCID II.
De gemiddelde schaalscore van de LPS correleert significant negatief met de som score
en met de drie subschalen van de IPO, waarvan het meest met primitieve defensies en
identiteitsdiffusie en het minst met realiteitstoetsing. Schaal 1 en 7 van de LPS correleren
significant negatief met de primitieve defensies en identiteitsdiffusie subschalen van de IPO,
schaal 2 en 6 van de LPS hebben geen significante correlaties met de subschalen van de IPO,
schaal 3 van de LPS correleert negatief met de identiteitsdiffusie subschaal van de IPO, schaal
4 correleert significant negatief met de drie subschalen van de IPO.
Discussie
De LPS is een zelfbeoordelingsschaal om de innerlijke verandering bij psychotherapie te
meten zoals die wordt ervaren door de patiënten zelf. De schaal werd samengesteld op basis
van uitspraken van patiënten zelf. Op basis van factoranalyse en repetitieve Chronbach alfa,
kwamen we tot een schaal van 112 items en zeven subschalen. (tabel 1 & 2).
Op tabel 3 zien we dat de meeste schalen positief intercorreleren, maar niet van die aard
dat we kunnen besluiten dat ze hetzelfde meten. We kunnen ervan uitgaan dat elke schaal een
afzonderlijk concept meet, dat voor een deel overlap vertoont met de concepten gemeten door
de andere schalen. De positieve correlaties wijzen er ook op dat alle schalen steun geven aan
de veronderstelling van een onderliggend concept dat een dimensie van pathologie naar
gezondheid meet.
Wat de invloed van demografische variabelen betreft (tabel 4), zien we dat mannen
globaal beter scoren op de LPS dan vrouwen, dit lijkt voor alle schalen zo te zijn behalve voor
controlegevoel over middelengebruik en positieve ervaring van de afdeling. Dit fenomeen zou
125

126
kunnen verklaard worden door het feit dat er in de proefgroep relatief meer mannen met een
narcistische problematiek zijn dan vrouwen en dat de groep met de narcistische problematiek
globaal beter scoort in hun zelfbeleving zoals die in verschillende dimensies gemeten wordt.
Er is geen effect van leeftijd of opleiding, wel van de burgerlijke staat: ongehuwden vinden
zichzelf ‘autonomer’ en beschrijven een betere affectregulatie bij zichzelf dan
samenwonenden en gescheiden personen. Gescheiden personen geven aan dat ze de afdeling
als positiever ervaren dan ongehuwden en samenwonenden. Dit zou kunnen verklaard worden
door het feit dat verschillende patiënten eerder recent gescheiden waren, zodat het alleen
leven een wat nieuwe situatie was – waarbij ze veel emotionele feedback van de afdeling
kregen. Leeftijd en opleiding hebben geen effect op de resultaten van de LPS. Deze resultaten
kunnen van dienst zijn bij de normering van de vragenlijst bij toekomstig onderzoek.
Het voorkomen van de afzonderlijke DSM as I stoornissen was niet talrijk genoeg om de
invloed op de antwoorden op de LPS na te gaan. Vier van de tien DSM IV as II
persoonlijkheidsdiagnoses, zoals door de psychiater klinisch waren gediagnosticeerd, waren
dat wel (tabel 5). Narcisten beschrijven meer positief welbevinden dan borderlines en
afhankelijken, ze vinden zich ook meer in staat tot het aangaan van authentieke relaties en zijn
meer gecontroleerd in middelengebruik dan borderlines. Ze rapporteren een betere
affectregulatie dan borderlines en histrionici en geven aan meer gericht te zijn op zichzelf.
Deze verschillen komen uiteraard overeen met de zelfbeleving die bij de narcistische
persoonlijkheidsstoornis hoort, zeker in vergelijking met de borderline en ook met de
afhankelijke persoonlijkheidsstoornis. Dit komt ook overeen met het klinisch beeld van deze
verschillende psychopathologieën: de borderline heeft veel moeite met het aangaan van
relaties, lijdt daaronder en heeft een groter middelengebruik (Zanarinini et al., 2003). De
affectregulatie is klinisch ook beter bij narcistische patiënten dan bij borderlines en theatrale.
126

127
Daarenboven komt het gevonden onderscheid tussen narcistische pathologie enerzijds
en borderlines, afhankelijke en theatrale anderzijds—met weinig verschil binnen deze laatste
groep, overeen met het onderscheid dat Blatt (1992) vond bij zijn analyse van de resultaten
van de Menninger studie. Het betreft een onderscheid tussen introjectieve patiënten (zoals de
narcistische patiënten) en de anaclytische patiënten (zoals de borderlines en de histrionische
patiënten).
De eerste schaal, positief welbevinden, hangt sterk samen met alle andere schalen en
kan als een gemeenschappelijk ervaringselement gezien worden. De alfa-coëfficient van 0.97
toont dat deze schaal homogeen is. De correlaties met de subschalen van de Neo-Pi-R (tabel
6) laten toe de eerste schaal meer specifiek te omschrijven: ze correleert positief met energie,
hartelijkheid, vrolijkheid, extraversie en negatief met neuroticisme, ergernis en depressie. De
naam ‘positief welbevinden’ geeft dit goed weer. De correlatie met de SCID II – categorieën
ligt in dezelfde lijn. Deze correlatie is het meest negatief met de meest angstige en
depressieve categorieën (ontwijkend, afhankelijk, depressief, paranoid, schizoid, borderline en
dwangmatig), maar toont een groter positief zelfbevinden in de categorieën van de SCID die
minder met hun psychisch lijden in contact zijn zoals de narcistische, de antisociale en de
theatrale categorieën. Vermits het een algemene maat is, correleert ze negatief met de
subschalen van de IPO die primitieve defensies en identiteitsdiffusie weergeven, en aldus de
ernst van de structurele persoonlijkheidsstoornis weergeven.
De tweede schaal over ‘openheid naar anderen’ is met een alfa-coëfficient van 0.85 ook
vrij homogeen. Uit de correlatie met de Neo-Pi-R zien we dat ze correleert met extraversie en
hartelijkheid maar ook met gevoelens, oprechtheid, avonturisme, altruïsme, medeleven en
inschikkelijkheid. Meer specifiek gaat het dus eerder om de mogelijkheid om authentieke
relaties aan te gaan zoals beschreven vanuit een innerlijke beleving, dan om een sociale
vaardigheid. Dit is een van de hoofddoelen van de klinische psychotherapie.
127

128
De schaal correleert zoals verwacht negatief met de meest introverte categorieën uit de
SCID II met name de ontwijkende, schizoïde en schizotypische categorie en positief met de
meest extraverte met name de theatrale.
De derde schaal ‘autonomie’ heeft met een Chronbach coëfficiënt van 0.81 ook een
goede interne consistentie. Uit de correlatie met de Neo-Pi-R blijkt dat de schaal negatief
correleert met ergernis en kwetsbaarheid en positief met energie, openheid, veranderingen
(openstaan voor nieuwe veranderingen), gevoelens(innerlijke gevoelswereld), fantasie
(innerlijk verbeeldingsleven), esthetiek (interesse voor kunst), doelmatigheid en
zelfdiscipline. De meeste verbanden zijn met de O-schalen en de laagscoorders op deze
schalen hebben een uitgesproken voorkeur voor het vertrouwde. De aanvankelijke term
autonomie is niet misplaatst, maar het accent ligt zoals bij de tweede schaal ook duidelijk op
een innerlijke attitude en niet op een sociale vaardigheid. Het gaat om de mogelijkheid om op
een autonome en creatieve manier met nieuwe gebeurtenissen om te gaan, wat een belangrijke
therapeutische verworvenheid is.
De correlatie met de IPO subschalen is niet zo hoog, waarschijnlijk omdat de
moeilijkheid in het omgaan met nieuwe gebeurtenissen niet alleen een borderline structuur
typeert maar ook bij hogere persoonlijkheidsorganisaties zoals de obsessief-compulsieve terug
te vinden is.
De vierde schaal ‘ affect regulatie’ correleert op de Neo-Pi-R vooral met de
aanwezigheid van negatieve gevoelens (neuroticisme, angst, ergernis, impulsiviteit) en hoe
ermee omgegaan wordt (oprechtheid, hartelijkheid en betrouwbaarheid). In die zin is de term
affectregulatie niet verkeerd gekozen maar is ‘de aanwezigheid en de omgang met negatieve
gevoelens’ een meer specifieke omlijning. Vermits we in deze schaal zowel de aanwezigheid
als de omgang met negatieve gevoelens vinden, is het niet verwonderlijk dat de interne
consistentie hier lager is (α = .64).
128

129
Het is niet verwonderlijk dat deze schaal sterk correleert met de twee eerste IPO
schalen. Volgens Kernberg worden low level borderlines immers getypeerd door de sterke
aanwezigheid van splitsing en een infusie van agressie. De correlatie met de agressieve SCID
II categorieën zoals de paranoide, de passief-agressieve, de borderline en de antisociale ligt in
dezelfde lijn.
De vijfde schaal over de ‘zelfgerichtheid’ correleert op de Neo-Pi-R niet negatief met
extraversie, sociabiliteit en altruïsme zoals we zouden verwacht hebben maar wel negatief met
de aanwezigheid van negatieve gevoelens, vooral impulsivitiet, en sterk positief met
avonturisme en betrouwbaarheid. De schaal lijkt dus niet zozeer een negatieve zelfgerichtheid
weer te geven maar een mogelijkheid tot zelfreflectie en een vertrouwen om met nieuwe
zaken om te gaan en waarvan impulsiviteit het omgekeerde is. Het is inderdaad zo dat de
klinische psychotherapie bij persoonlijkheidsstoornissen zich de laatste jaren focust op de
omgang met impulsen en tegelijk op het openstaan voor en het opdoen van nieuwe
ervaringen. Dit wordt door Fonagy en collega’s (1996) als de mogelijkheid tot zelfreflectie
beschreven. In deze schaal wordt een duidelijke focus van de behandeling weerspiegeld,
zonder dat het om een eenduidig construct gaat (α = .64).
Het is niet onverwacht dat deze schaal positief correleert met de narcistische SCID II
categorie die inderdaad een grotere gerichtheid op zichzelf inhoudt terwijl ze negatief
correleert met het merendeel van de categorieën van persoonlijkheidsstoornissen. Het meest
negatief correleert ze met de borderline en de paranoide categorieën, die volgens Fonagy et al.
(2002) een expliciet tekort aan deze zelfreflectie vertonen.
De sterke correlatie van deze schaal met de eerste twee IPO schalen is ook verwacht,
vermits de mogelijkheid tot zelfreflectie omgekeerd evenredig is met de primitieve defensies
en de grensverwarring bij low level borderlines.
129

130
De zesde schaal over een ‘positieve ervaring van de afdeling’ correleert weinig met
persoonlijkheidstrekken van de SCID en ook niet van de Neo-Pi-R. Wel met kenmerken die
een algemeen gevoel weergeven zoals vrolijkheid en vertrouwen. Zoals verwacht is er ook
een sterke correlatie met schaal 1 (positief welbevinden) en schaal 4 (negatieve gevoelens en
hoe ermee omgaan). Het is een homogene schaal (α = .82).
De zevende schaal over het ‘middelengebruik’ correleert zoals verwacht sterk met de
aanwezigheid van negatieve gevoelens uit de ‘neuroticisme’ schaal van de Neo-Pi-R, maar
ook met doelmatigheid (wat de aanwezigheid van een innerlijke beleving van inefficiëntie bij
middelengebruik weerspiegelt). Het is een homogene schaal (α = .85). De schaal komt
overeen met SCID II persoonlijkheidscategorieën waar middelengebruik klinisch meer lijkt
voorte komen zoals de ontwijkende, de afhankelijke, de depressieve en de borderline. De
correlatie met de IPO en dus met de mate van borderline structuur is ook verwacht gezien het
middelengebruik als manier om niet psychisch met spanning te moeten omgaan ook een
borderline kenmerk is.
Uit de omlijning van de schalen blijkt dat de patiënten in hun eigen woorden
hoofdzakelijk innerlijke belevingen weergeven. Dit komt tegemoet aan het doel van de schaal,
die het specifieke van de klinische psychotherapie voor persoonlijkheidsstoornissen wou
weergeven. De focus van de behandeling is een innerlijk psychisch proces.
Op basis van de het onderzoek naar de congruente validiteit van de LPS veranderden we
namen van de subschalen in de experimentele versie in namen die beter weergeven wat de
subschalen meten: S1: Positief Welbevinden; S2: Openheid voor authentieke relaties; S3:
Capaciteit om met onverwachte en stressrijke situaties om te gaan; S4: Capaciteit om met
negatieve emoties om te gaan; S5: Capaciteit voor verinnerlijking; S6 : Positieve ervaring van
de setting; S7: Controle over middelengebruik.
130

131
Besluit
De LPS is een empirisch samengestelde zelf rapportage vragenlijst aan de hand van items die
gebaseerd zijn op uitspraken van patiënten zelf. De bedoeling is om het innerlijke
veranderingsproces te meten tijdens psychotherapie vanuit de beleving van de patiënt. De
congruentie van deze schaal blijkt uit de correlaties met verschillende
persoonlijkheidsschalen. Uit deze resultaten blijkt ook dat eerder innerlijke veranderingen
worden weergegeven dan puur klinisch symptomatische.
131

132
132

133
Chapter 5
Measuring the Psychoanalytic Process with the Psychoanalytic
Process Rating Scale: The Perspective of the Therapist.
In the two previous chapters, we discussed the measuring of the process of inner change
in personality disordered patients in psychoanalytically informed hospitalization-based
treatment, with treatment specific measures from the perspective of independent researchers
and from the perspective of the patients themselves. We found few scales to rate the process
from the therapist's point of view. There are scales to measure the relationship between
patient and therapist such as the Helping Alliance Questionnaire (Luborgsky et al., 1996), the
Barett-Lennard Relationship Inventory (Barett-Lennard, 1962) and the Working Alliance
Inventory (Horvath & Greenberg, 1994). We found one scale measuring several facets of the
psychoanalytic process, the Psychoanalytic Process Rating Scale or PRS (Gerber, Fonagy,
Bateman & Higgitt, 2004). A Dutch scale is constructed based on a translation and
elaboration of this PRS: the Psychoanalytic Process Rating Scale (PPRS, Stoker & Beenen,
2001; Stoker & Zevalkink, 2005). This scale has an established reliability by now, but was in
a stage of development when we started using it. Therefore, before using it in the process-
outcome study we did some preliminary tests about the reliability and the validity of the scale
in using it with personality disordered patients. Our data confirmed the validity of the scale
(Stoker & Zevalkink, 2005).
133

134
Construction and Reliability of the PPRS
Beenen and Stoker (1997) constructed the ' Periodische Beoordelingsschaal' (PBS) or
Periodic Rating Scale (PRS). Their scale consisted of 262 items and was a translation and
elaboration of the 'Anna Freud Centre Session Rating for Children and Adolescents' and the '
Anna Freud Centre Young Adult Weekly Rating Scale' (Gerber, Fonagy, Bateman & Higgitt,
2004). The PRS consists of three parts: 1. General aspects of the process, 2. Characteristic
Aspects, 3. Interaction. The PRS was assessed every two weeks on a random session from
1996 by analysts of the Dutch Psychoanalytic Institute.
To construct the PPRS (Psychoanalytic Process Rating Scale), 860 PRS of 60
psychoanalyses were analysed in 2000. The sensitivity of the items to change was examined
and the items were cluster analysed for each of the parts of the scale. Based on the results of
this processing, a scale on 'contact development' was made with items from part 1 and 3. This
scale consists of 44 items. 23 items describe the patient by the analyst during the session, 21
items describe how the therapist feels and interacts in regard to this session. A factor analysis
on 612 PRS revealed five factors, explaining 37.5 % of the total variance with an internal
consistency ranging between.70 and.82 (table 1).
These factors were:
1. Co-operation: this refers to a productive and positive process focused on the
therapeutic relationship, to a close co-operation, to the process of alliance, to the degree of
clicking between analyst and patient.
2. Tension: this refers to negative feelings in both participants and can be used as an
indication of the degree the two participants clash. It does not mean that the treatment has
reached a deadlock or does not progress well.
134

135
3. Exploration: this refers to the therapist making thoughts and feelings of the patient
explicit, showing implicit patterns to the patient and linking with external reality.
4. Disillusion: this refers to the degree that the analysand experiences, expresses and
behaves as a rejected lover.
5. Contentment: this factor refers to a positive idealising counter-transference in which
the analyst may see the analysand as an ideal son, daughter or partner. Moreover it refers to
the healthy affective mood of the analysand and his enjoying the treatment.
Table 1
Subscales of the Psychoanalytic Process Rating Scale Version 1
Subscales Internal consistency
1997-1999
number of items
Cooperation (CP) .82 11
Tension (TE) .82 12
Exploration (EX) .70 6
Disillusion (DI) .81 13
Contentment (CT) .70 5
135

136
The analyses of part 2 revealed too many clusters to be useful. The items can be used for
a qualitative examination of content. Together with the 44 item 'Contact Development Scale'
they formed the first version of the PPRS (Beenen & Stoker, 2001). In our study we used the
44 item Contact Development Scale of the PPRS.
Recently a new factor analysis was done based on the results of this 44 item Contact
Development Scale of 969 PPRS assessed between 2001 and 2003. This factor analysis
revealed the same five factors with an increased explanation of the total variance (48.3 %) and
with some items jumping to another scale. This 44 item scale forms the PPRS version 2
(Stoker & Zevalkink, 2005).
Table 2
Subscales of the Psychoanalytic Process Rating Scale Version 2
Subscales Internal consistency
2001-2003
% of variance
explained
number of items
Cooperation (CP) .82 16% 11
Tension (TE) .86 9.75% 12
Exploration (EX) .72 8.76% 6
Disillusion (DI) .71 7.29% 13
Contentment (CT) .78 6.50% 5
PPRS-total .84 48.3%
136

137
Convergent and Divergent Validity of the PPRS
The aim of this study is to examine the intercorrelation of the subscales of the PPRS, the
correlation with related measures and the correlation with clinical scales. Related measures
were the 4 expert measures on the ORI, discussed in Chapter 3, and the LPS, discussed in
Chapter 4. These correlations were investigated on a group of patients on admission and a
group of patients with a variable length of stay, because the PPRS measures contact
development. Clinical scales were a symptom scale (SCL), a scale based on the categorical
approach to personality disorders (DSM-IV Axis II, evaluated with the SCID-II
questionnaire) and a scale based on the structural approach to personality disorders (IPO). The
intercorrelation of the PPRS scales and the correlation with the LPS was also investigated
with a comparable group of PD patients which were of varying length in therapy, this to see
whether the correlations changed with therapy, compared to a first assessment.
Method
Participants.
The first study group consists of 82 PD patients at the very beginning of their treatment.
These are first assessments with the PPRS. The study group is the same group as described in
Chapter 3. Eighty-two patients (54 male, 28 female) referred to a setting for hospital treatment
(day hospital and in-patients) on psychoanalytic lines for personality disorders at the
University Center Kortenberg, Belgium (see Chapter 1) were tested. The patients were
between the ages of 17 and 58 (M = 27.1, SD = 8.5). Sixty-one patients were single, 29 lived
137

138
with their parents, 17 were married or living together with a partner, 3 were divorced and 1
lived in a community center. Three patients completed primary education, 32 secondary
education and 47 higher education or university. On the basis of the SCID-II, we assigned all
patients to one of the three clusters of personality disorders according to DSM-IV: cluster A,
2 (1 schizoid, 1 schizotypal); cluster B, 62 (1 histrionic, 44 borderline, 10 narcissistic, 4
narcissistic and borderline, 3 histrionic and borderline); cluster C, 10 (1 avoidant, 2
dependent, 1 avoidant and dependent, 6 NOS); clusters A+B, 1; clusters B+C, 10.
The second study group consists of 53 PD patients that were in therapy for a duration
of varying length. This group is part of the first group. Fifty-three patients (18 males, 35
females) followed a treatment on psychoanalytic lines for personality disorders (see
Chapter1), 39 in an open-ended residential setting and 14 in a day-treatment setting. The
patients were between the ages of 17 and 58 (M = 28.0, SD = 9.2). Twenty-one were single,
19 lived with their parents, 10 were married or living together with a partner and 4 were
divorced. Nineteen patients completed secondary education and 34 higher education or
university. On the basis of the SCID II we assigned all patients to one of the three clusters of
personality disorders according to DSM IV; cluster B: 40 (28 borderline, 9 narcissistic, 2
narcissistic and borderline, 1 histrionic and borderline); cluster C: 6 (1 dependent, 1 avoidant
and dependent, 4 not otherwise specified); clusters B+C: 7.
These patients had a variable length of stay at the moment of assessment of the PPRS: <3
months: 1; 3-6 months: 5; 6-9 months: 4; 9-12 Months: 30; >12 Months: 13.
138

139
Instruments.
Besides the Psychoanalytic Process Rating Scale version 1 (PPRS, Beenen and Stoker, 2001),
we used the Felt Safety Scale, the Differentiation Relatedness Scale, the Reflective
Functioning Scale and the Bion Grid Scale as related measures; these measures are discussed
in Chapter 3. Another related measure is the Leuven Psychotherapy Scale (LPS), discussed in
Chapter 4.
As clinical measures we used the Symptom Checklist 90 (SCL-90) (Derogatis,1977),
translated by Arrindell & Ettema (1993), which consists of 90 items that tap eight
psychopathological domains. As personality related scales we used the SCID-II questionnaire
that has been translated in Dutch and adapted for DSM-IV by Weertman, Arntz & Kerkhofs
(2000). The SCID-II is a structured clinical interview of DSM-III-R Personality Disorders by
Spitzer & Williams, 1985) and the Inventory of Personality Organization (IPO). The IPO is a
self-report questionnaire (Kernberg & Clarkin, 1995) measuring the intrapsychic structure in
three dimensions according to the model of Kernberg on borderline personality organization.
The reliability and validity of the IPO translated in Dutch has been demonstrated by Vermote,
Vertommen et al., 2004 and by Vermote,Smits, Claes & Vertommen, 2005.
Procedure.
After written informed consent, the PPRS was assessed by the therapist, in this case two
senior psychoanalytic psychotherapists and one psychoanalyst. The patients were in group
analysis (three times, 1.5 hour sessions three times a week), and because it was difficult to
139

140
rate all patients on one session, the PPRS was assessed every three months for each patient
about the last two weeks of treatment.
The other scales were assessed by an independent researcher in the same periods of the
assessment of the PPRS. The Object Relation Inventories were assessed at the same moments
of assessment of the Clinical Symptom Measures by a psychologist having no contact with
the therapeutic staff and the patients besides of these interviews. The transcripts of the
anonimised ORI’s were distributed in a random order to raters blind to any information about
the patients and having had no contact with them. The ORI – interviews were rated on the Felt
Safety Scale, the Differentiation Relatedness Scale, the Reflective Functioning Scale and the
Bion Grid Scale by a group of 12 raters, three raters for each scale, who had a previous
training with the manual of their scale and a satisfactory reliability test for this scale.
Analysis of data.
The correlations between the scales of the PPRS were examined using the Pearson
Product Moment Correlation Coëfficient. The correlations of the PPRS scales with the LPS,
FSS, DRS, RFS, BGS, IPO, SCID, SCL scales were also analysed by using the Pearson
Product Moment Correlation Coëfficient.
140

141
Results
The Pearson correlation coëfficients of the intercorrelation between the five PPRS scales
for study group 1 after a first assessment are given in table 3.
Table 3
Correlations between PPRS Subscales at the First Assessment, Pearson Correlation
Coefficient (n=82)
Subscale PPRS CP PPRS TE PPRS EX PPRS DI PPRS CT
PPRS CP ― .32** .37*** .27* .39***
PPRS TE ― .53*** .41*** -.36***
PPRS EX ― .20 .15
PPRS DI ― -.18
PPRS CT ―
* p < .05, ** p < .01, *** p < .001
The correlations of the PPRS scales at a first assessment with the SCID-II categories are
given in table 4.
There were few significant correlations of the PPRS scales with the SCID-II
categories.These correaltions are show in table 5. Weak negative correlations were found
between the obsessive compulsive category and the PPRS ct (r = -.25, p = .03), the depressive
category and the PPRS cp (r = -.27, p = .01) and the PPRS ct (r = .22, p = .04) and between
the schizoid category and the PPRS ct (r = -.31, p < .01).
141

142
Table 4
Correlations between PPRS Subscales of a Group of Patients for a Duration of Therapy of
Varying Length, Pearson Correlation Coefficient (n=53)
Subscale PPRS CP PPRS TE PPRS EX PPRS DI PPRS CT
PPRS CP ― .14 .38** .57*** .54***
PPRS TE ― .53*** .26 -.49***
PPRS EX ― .33* -.08
PPRS DI ― -.02
PPRS CT ―
* p < .05, ** p < .01, *** p < .001
The only significant correlations between the scales of the SCL-90-R and the scales of
the PPRS were with the PPRS te scale: with SCL interpersonal sensitivity (r = .22, p = .05)
and with SCL anxiety (r = .23, p = .03).
When correlating the three IPO subscales (Pdm, Id, Rt) with the 5 PPRS subscales, there
were only significant correlations with the PPRS di scales: with IPO Pdm (r = .22, p = .05)
and with IPO Rt (r = .27, p = .01).
The correlation between the subscales of the LPS and the subscales of the PPRS of a
group of patients for a duration of therapy of varying length, is shown in table 6.
The correlation between the four ORI expert rating scales (FS, DRS, RFS, BGS) and the
PPRS scales is shown in table 7.
142

143
Table 5
Correlation of the SCID II and PPRS at the First Assessment, Pearson Correlation
Coefficient ( n=53)
Subscale PPRS CP PPRS TE PPRS EX PPRS DI PPRS CT
SCID AVOI -.02 .04 -.11 .08 -.22
SCID DEP -.03 -.02 -.11 .10 -.06
SCID OC -.10 .11 -.01 .06 -.25*
SCID PA -.09 .02 -.09 .17 .01
SCID DEPR -.27* .11 -.08 .10 -.22*
SCID PAR -.07 .10 -.06 .08 -.02
SCID SCT .06 .07 .00 .013 -.03
SCID SCI -.14 .12 .03 .17 -.32**
SCID HIS .10 -.02 -.07 .06 .14
SCID NAR -.06 .01 -.00 .14 .06
SCID BOR .01 .08 -.14 .21 -.06
SCID ANT -.03 .11 .06 .18 -.13
* p < .05, ** p < .01
143

144
Table 6
Correlation between LPS and PPRS of a Group of Patients for a Duration of Therapy of
Varying Length, Pearson Correlation Coefficient (n=53)
Subscale PPRS CP PPRS TE PPRS EX PPRS DI PPRS CT
LPS Pos. Well Being .03 -.33* -.33* -.09 .32*
LPS Cap. Authentic Relationships. -.13 -.19 -.25 -.05 .16
LPS Cap. Unexpected Events .01 -.10 -.18 -.11 .22
LPS Cap. Negative Emotios -.09 -.41** -.35** -.22 .17
LPS Cap. Interiorization -.03 -.17 -.07 -.03 .14
LPS Positive Exp. Setting .06 -.48*** -.17 .00 .53***
LPS Control Substance Use -.19 -.26 -.43** -.33* .24
* p < .05, ** p < .01, *** p < .001
Table 7
Correlation between LPS and Felt Safety Scale(FSS), Differentiation Relatedness Scale(DRS),
Refelctive Functioning Scale(RFS) and Bion Grid Scale (BGS) of a Group of Patients for a
Duration of Therapy of Varying Length, Pearson Correlation Coefficient (n=53)
Subscale PPRS CP PPRS TE PPRS EX PPRS DI PPRS CT
FSS global .16 -.32* -.05 .08 .42**
DRS global .30* -.10 .15 .05 .48***
RFS global -.06 .02 -.06 -.17 .17
BGS global .18 -26 .09 -.15 .49***
* p < .05, ** p < .01, *** p < .001
144

145
Discussion
The PPRS issued from factor analysing the 262-item PRS and was constructed with the
theory driven items from the general process and the interaction parts of it. The scale is
designed to reflect the therapist’s impression about the analytic relation in a way that makes it
possible to rate it and to subject those ratings to statistical analysis. The scale contains
therapist and patient related items.
Not all scales correlate; the correlations that we find are in line with clinical logic. The
Exploration scale correlates with the Cooperation scale and goes with increased Tension. We
find a positive correlation between Disillusion and Tension scales, a negative correlation
between Tension and Contentment and a positive correlation between Cooperation and
Contentment. These correlations are expected and remain during therapy as table 3 and 4
show. We see that in patients under therapy, a former correlation between Cooperation and
Tension stops, just as between Disillusion and Tension. It is indeed expected to be an effect of
the therapy that patients are less afraid to cooperate and that they support negative feelings
better. These intercorrelations fit with what we expect and support the validity of the scale.
Although most scales are interrelated, they measure separate qualities of the process, as the
highest value is.57.
We wondered whether the factors describing the process by the PPRS, the result of a factor
analysis of theory driven items in a sample of patients in psychoanalysis, were related to the
factors describing the process by the LPS, the result of a factor analysis of items based on
statements of the patients in a PD sample. There is indeed a correlation between the LPS scale
of 'Positive well-being' and the PPRS scale 'Contentment' and a negative one with 'Tension'.
145

146
Both the PPRS and LPS seem to reveal this same basic dimension. There is indeed a
correlation between the LPS scale 'Capacity to deal with negative emotions' and the PPRS
scale ' Exploration', but to our surprise it is a negative one. We further see that this
'Exploration' scale correlates negatively with the LPS scale 'Positive Well-Being'. An
explanation may be that the items of the PPRS scale 'Exploration' are all therapist related
items in which the therapist confronts or interprets and such sessions may be accompanied
with a temporally lessening of the feeling of well-being. At least in this sample of PD patients
this seems to go with a diminished State Reflective Functioning. The fact that the LPS '
Capacity to deal with negative emotions' correlates negatively with the PPRS ' Tension' is as
expected. A correlation in the relation dimension, between the LPS scale 'Openness to
authentic relations' and the PPRS scale ' Cooperation' is not found.
When correlating the PPRS with the dimensions of the three dimensional model as
measured on the ORI (see Chapter 3), we found the Felt Safety Scale correlating positively
with the PPRS scale ' Contentment' and negatively with the ' Tension' scale, which is
expected. The DRS that is about the mental representation of self-other relationships,
correlates with the PPRS scale ' Cooperation ' scale, which is the PPRS scale which is most
relationship orientated and with the 'Contentment ' scale. The Reflective Functioning Scale as
measured on the ORI, does not correlate with any of the PPRS scales. The Bion Grid Scale,
which reflects the way of creatively linking experiences with several levels of thought
correlates with the PPRS scale ' Contentment', indeed it is only in a relaxed contact between
patient and therapist that this mental functioning goes well.
We found few relations of the PPRS with personality related measures. With the SCID
II, consisting of items describing the DSM IV Axis II categories, we see a negative correlation
between the PPRS 'Cooperation' scale with the Depressive category and a negative correlation
between the PPRS 'Contentment' scale with the Obsessive Compulsive, the Depressive and
146

147
the Schizoid categories. These correlations fit clinical logic as well, because these categories
consist of anxious patients that are just admitted and adjusting to the setting. The PPRS does
not seem to measure personality traits, but the relations are coherent As far as the level of
Borderline Personality Organisation is concerned, we see only a correlation with the PPRS
'Disillusion' scale. This is in line with Kernberg's (1996) model on which the IPO is based,
lower degrees of personality organisation going with higher IPO scores are characterised by a
greater aggression infusion. There are no relationships with the SCL-90 except for the PPRS '
Tension ' scale with the SCL interpersonal sensitivity and anxiety subscales.
In conclusion we may say that for a sample of personality disordered patients, the
relations of the PPRS are coherent with the process measures assessed on the ORI and the
self-report LPS scales. There are few relations with clinical scales, but these are coherent.
This is a good argument for validity given the fact that the study group, personality disordered
patients in psychoanalytically informed hospitalization, differs greatly from the group on
which the scale has been constructed, clients in ambulant psychoanalysis or psychoanalytic
psychotherapy. We wonder however whether the PPRS reflects sufficiently the domains of
change which we judge important in personality disorders; felt safety seems to be present, and
a relation aspect as well but the mentalisation dimension is not really present. This may due to
the fact that in the client group on which data the PPRS is constructed, mentalization is not
the main issue. There are no others groups of patients to compare with at the moment. These
data are the first data on validity of the PPRS. The PPRS is part of an evaluation procedure of
the Dutch Psychoanalytical Institute which will provide soon data that will be used studying
the validity of the PPRS. In Italy the PPRS is used in the evaluation of Psychoanalytic
Psychotherapies by G. Amadei (Stoker and Zevalkink, 2005).
147

148
148

149
PART 3
OUTCOME, OUTCOME TRAJECTORIES AND DROP-OUT
149

150
150

151
Chapter 6
Outcome and Outcome Trajectories of Personality Disordered
Patients During and After Hospitalization on Analytic Lines
According to meta-analyses (Perry et al., 1999; Leichsenring & Leibing, 2003),
psychotherapy on analytic lines proves to be an effective treatment for personality disorders.
A specific form of this therapy, namely partial hospitalization premised on analytic lines, has
been examined for borderline patients utilizing an RCT design by Bateman and Fonagy
(1999, 2001). These researchers found significant clinical improvement during a treatment
period of 18 months; gains which continued in a follow-up period of 18 months.
RCT studies such as the Bateman and Fonagy study are bound to a selection of
homogeneous groups but in most settings for a hospital-based treatment of personality
disorders, the population includes a wider range of persons within the spectrum of borderline
personality organization (Kernberg, 1996).
Several authors (Bergin & Lambert, 1979; Frank, 1979; Clarkin & Levy, 2004) suggest
that the largest proportion of variance in therapy is due to personal characteristics and
qualities of the clients. Lambert (1992) attributes as much as 40 % of client improvement in
psychotherapy to client variables and extratherapeutic influences. When it is possible to
discern differences in outcome trajectories, it must be possible to study the impact of pre-
treatment client variables by post-hoc examinations, given a naturalistic composition of the
study group.
In the present two-year prospective study, the study group comprises 78 consecutively-
admitted patients in a psychoanalytically informed hospitalization based treatment of
personality disorders. Symptom change during therapy (which is open-ended with a
151

152
maximum duration of 1 year) and during a 1 year follow-up period is measured. In this study,
we want to examine whether it is possible to reliably discriminate different trajectories and
whether these trajectories are linked to patient characteristics at the start. We may suppose
that the effect of patient variables on outcome trajectories is prominent as the patients all get
the same treatment and because extratherapeutic influences are relatively small in this
hospitalization-based treatment. In the future clinical settings might benefit from these
findings to adapt and specify treatment technique and treatment goals to the needs of the
patients.
For the process-outcome study (Chapter 8) we study a part of this group, namely the 47
consecutively admitted patients that stayed at least nine months, this to have enough moments
to study the process. We can compare the results of this group with the results of the treatment
group of the Bateman and Fonagy, 1999 RCT study, because in their treatment group of
nineteen patients, the drop-outs neither were included and the treatment duration was up to 18
months. This comparison may compensate partly for the fact that there is no control group in
the naturalistic design of our project. The Kortenberg group comprises consecutively admitted
DSM IV axis II personality disordered patients within the spectrum of borderline personality
organisation, while the group of the Bateman and Fonagy study is a selected group of DSM
IV axis II borderline patients.
First study: Outcome of Personality Disordered Patients During and After a Hospitalization
on Analytic Lines.
In this study we examined the change in clinical symptoms and in personality related
characteristics of personality disordered patients in psychoanalytically informed
hospitalization for 9 to 12 months and during one year follow-up.
152

153
We compared anxiety and depression scores assessed with the same scales, at the same
time intervals in our study as in the therapy group of the RCT study of Bateman and Fonagy
(1999).
Treatment
The treatment of patients with personality disorders with psychoanalytically informed
hospitalization based treatment (KLIPP) at the University Centre Kortenberg is to a large
extend comparable to the treatment model, recently described by Bateman and Fonagy
(2004). The Kortenberg hospitalization treatment focuses on three dimensions of inner
change: felt safety, mentalization, and self-other object relations (Chapter 1). Two setting
characteristics distinguished the present setting and that described by Bateman and Fonagy
(1999, 2001, 2004): the present study was an open-ended in-patient and day-hospital
treatment with a maximum stay of approximately one year, rather than the 18 months of day
hospital treatment in the Bateman and Fonagy (1999) study.
Method
Participants
The study group comprised personality disordered patients who stayed long enough in
psychoanalytically informed open-ended hospitalization to have at least the measures at
outset, after 3, 6 and nine months of treatment and three months after discharge. This in order
to have enough moments to enable a process study and to have a length of time of treatment
comparable with the Bateman and Fonagy study (1999).
153

154
47 consecutively admitted patients corresponded to these conditions. Of these 47
patients, three patients were not included in the statistical analysis because of missing data at
discharge or follow-up: one refused further assessments, the self-reports of another got lost by
mail delivery service, one patient was acutely discharged because of selling drugs.
Thus the patients in the study consisted of 44 admitted patients (13 males, 31 females).
31 were in-patients, 13 in a day-treatment setting. The patients were between the ages of 17
and 58 (M = 28.0, SD = 9.24). 17 were single, 17 lived with their parents, 7 were married or
living together with a partner, 3 were divorced. 16 patients completed secondary education
and 28 higher education or university. Using the SCID-II we assigned all patients to one of
the three clusters of personality disorders according to DSM-IV; cluster B: 32 (23 borderline,
7 narcissistic, 1 narcissistic and borderline, 1 histrionic and borderline); cluster C: 7 (1
dependent, 1 avoidant and dependent, 5 not otherwise specified); clusters B+C: 5. On DSM-
IV Axis I, we diagnosed 20 mood disorders, 2 anxiety disorders, 1 eating disorder, 5
adjustment disorders and 5 substance related disorders.
Measures
Clinical symptom measures.
The Self Harm Inventory (SHI) (Sansone et al., 1998) applied to the last two weeks, the
Symptom Checklist 90 (SCL-90) (Derogatis, 1977) translated by Arrindell & Ettema(1993),
the Spielberger State-Trait Anxiety Inventory translated by van der Ploeg et al. (1980), the
Spielberger State-Trait Anger Inventory translated by Van der Ploeg et al. (1982) and the
Beck Depression Inventory translated by Bouman et al. (1985).
154

155
Personality characteristics.
The questionnaire of the Structured Clinical Interview for DSM-IV Axis II disorders
SCID-II (Spitzer & Williams, 1985) has been translated in Dutch and adapted for DSM-IV,
by Weertman, Arntz & Kerkhofs (2000). The Inventory of Personality Organization
(Kernberg & Clarkin,1995), is a self-report questionnaire used to assess a structural diagnosis
according to the model of Kernberg on borderline personality organization. The reliability and
validity of the IPO translated in Dutch was demonstrated by Vermote,Vertommen et al.
(2004) and Vermote, Smits, Claes and Vertommen (2005). The Inventory of Interpersonal
Problems, circumflex version (Alden et al., 1990) measures interpersonal functioning in 8
scales: domineering (pa), vindictive (bc), cold (de), socially avoidant (fg), nonassertive (hi),
exploitable (jk), overly nurturant (lm), intrusive (no). All these measures have a well
established reliability and validity.
Procedure
The patients were assessed within the first two weeks of admission, every three months,
and upon discharge. In follow-up, the patients were assessed three months and one year after
discharge.
The personality related measures were assessed within the first two weeks of admission,
upon discharge and three months and one year after discharge.
155

156
Data Analysis
Analysis of variance, repeated measures. We used the Cohen d in comparing the
Kortenberg outcome with the London outcome.
Table 1
Mean Scores of the Sansone Self Harm Inventory (SHI), Symptom Check List (SCL) and Beck
Depression Inventory (BDI)
SHI SCL psn BDI
Time M SD M SD M SD
admission 2.32 2.53 210 57.9 25.4 10.4
3m. 2.41 2.5 210 60.4 23 11.1
6m. 1.88 2.23 211 69.5 22.5 10.9
9m. 1.73 2.21 201 62.6 18.9 9.84
12m. 1 1.54 175 56.7 15.2 9.59
15m 1.25 2.8 169 65.5 14.3 11.5
24m 1.29 1.95 170 60.6 13.7 11.1
156

157
Results
The results for each of the symptom outcome measures are shown in table 1 and 2.
There was a significant change over time. We noted an improvement during therapy and the
year after discharge for the following symptom measures: SHI last two weeks (F
(6,248)=4.88, p =< .0001), SCL pst (F (6,248)=9.99, p =< .0001) with a cubic component
(F(1,248)=5.77, p = .02) and BDI (F (6,248)=16.45, p =< .0001), STAI state (F
(6,248)=5.84, p =< .0001), STAI trait (F (6,248)=14.52, p =< .0001), STAXI trait (F
(6,248)=5.21, p =< .0001) with a quadratic component (F(1,248)=4.32, p = .04). These
changes over time are presented in figures 1-7. Only the STAXI state showed no significant
change (F (6,248)= .95, p = .45).
Table 2
Mean Scores of the Spielberger State and Trait Anxiety and Anger Inventory (STAI & STAXI)
Every Three Months During Treatment and Three Months and a Year After Discharge
STAI state STAI trait STAXI state STAXI trait
Time M SD M SD M SD M SD
admission 47.6 13.4 59 10.2 11.9 3.1 17.8 6.77
3m. 46.1 11.5 57.4 10.6 11.7 3.93 18.1 5.97
6m. 45.2 12.1 56.5 11.9 12.1 4.6 17.9 5.15
9m. 43.8 11.9 54.1 10.7 12.3 5.06 18.1 4.06
12m. 40.9 13.9 49.4 13 11.1 2.65 16.8 4.96
15m 39.3 14 48 14.6 11.6 4.66 15.5 4.76
24m 39.5 13 46.8 14.7 10.9 1.11 15.5 3.89
157

158
Figure 1. Time related change of the mean scores on the Self Harm Inventory (SHI) of the
last two weeks (Moment 4 is at discharge).
Figure 2. Time related change of the mean scores of the Symptom Checklist (SCL-90-R)
Positive Symptom Total.
158

159
Figure 3. Time related change of the mean scores of the Beck Depression Inventory (BDI).
Figure 4. Time related change of the mean scores of the Spielberger Anxiety Inventory, State
(STAI).
159

160
Figure 5 . Time related change of the mean scores of the Spielberger Trait Anxiety Inventory
(STAI).
Figure 6. Time related change of the mean scores of the Spielberger State Anger Inventory
(STAXI).
160

161
Figure 7. Time related change of the mean scores of the Spielberger Trait Anger Inventory
(STAXI).
The results for each of the personality related measures are shown in table 3. There
was a significant change over time. We noted an improvement during therapy and the year
after discharge for all personality related characteristics: IPO sum (F (3,128)=6.15, p =
.0006), SCID II sum (F (3,128)=11.17, p < .0001) with a quadratic trend (F(1,128)= 4.22, p =
.04) and the IIP mean (F (3,128)=6.84, p = .0003). These changes over time are presented in
figures 8-10.
161

162
Table 3
Mean Scores of the Sum Scores of the Inventory Personality Organisation (IPO), of the Sum
Scores of the Structured Interview DSM III-R ( adapted questionnaire for DSM IV) and of the
Inventory Interpersonal Problems (IIP)
IPO SCID II IIP
Time M SD M SD M SD
admission 136.93 42.09 1.69
12m. 129.18 35.57 1.56
15m 126.54 31.86 1.41
24m 122.20 31.41 1.35
Figure 8. Time related change of the mean sum scores of the Inventory of Personality
Oganisation ( IPO), at 4 moments.
162

163
Figure 9. Time related changes of the mean sum cores of the Structural Clinical Interview of
DSM III -R Personality Disorders ( SCID II), adapted questionnaire for DSM IV.
Figure 10. Time related changes of the mean scores of the Inventory of Interpersonal
Problems (IIP).
163

164
The results of the STAI state, the STAI trait and the BDI of this study group
(Kortenberg group) can be compared with the results of the therapy group in the Bateman &
Fonagy (1999) RCT study (London group), at similar time moments.This is shown in table 4.
The therapy in the Kortenberg group stops at 12 months, in the London group at 18 months.
The baseline of the two groups is different, the Kortenberg group starts at a lower level of
pathology than the London. The effect sizes (Cohen d) of the STAI state are .6 for this study
group and 2.1 for the London group, of the STAI trait are 1.1 for this study group and 1.6 for
the London group and of the BDI they are 1.1 for this study group and 1.6 for the London
group.
Table 4
Comparison of the Mean Scores of the Spielberger State and Trait Anxiety Scores (STAI) and
of the Beck Depression Inventory Scores (BDI) Between the Study Group in Kortenberg and
the Therapy Group of the Bateman & Fonagy 1999 RCT in London, with Assessments Every
Three Months
STAI state STAI trait BDI
Kortenberg London Kortenberg London Kortenberg London
Time M SD M SD M SD M SD M SD M SD
admission 47.6 13.4 68.4 7 59 10.2 66.5 6.1 25.4 10.4 36 7.6
3 months 46.1 11.5 66.7 7.9 57.4 10.6 65.8 5.8 23 11 36.2 7.3
6 months 45.2 12.1 64.3 12.1 55.8 11.9 62.3 9.8 21.6 10.9 36.3 8.9
9 months 43.8 11.9 60.2 12.1 54.2 10.7 60.3 7.4 18.9 9.8 30.7 10.4
12 months 40.9 13.9 55.6 9.8 49.4 13 60.4 7.4 15.2 9.6 26.7 8.7
15 months 39.3 14 53.8 9.7 48 14.6 56.4 8.9 14.3 11.5 23.7 5.7
164

165
Study 2. Outcome trajectories of personality disordered patients during and after a
hospitalization on analytic lines.
In this study, we want to examine whether it is possible to reliably discriminate different
trajectories and to study whether different trajectories are linked to patient characteristics. To
study the trajectories of treatment, we asked all consecutively admitted patients
between.23/5/01-15/7/02 to participate in this prospective study. The group of the first study
is part of this larger group.
Method
Treatment
The treatment is the same as in Study 1.
Participants
78 consecutively-admitted patients in a psychoanalytically informed hospitalization
based treatment of personality disorders (in-patient and partial hospitalization) were asked to
participate at the present prospective study. Two patients refused and six patients left within
the first days and could not be implied because of missing data. They are discussed in Chapter
7 on attrition. After obtaining written informed consent, the remaining 70 consecutively-
admitted patients (24 males, 46 females) were enrolled in the study. Of the 70 participants, 51
were treated in an open-ended hospital setting and 19 in a day-treatment setting. Patients were
165

166
between the ages of 17 and 58 (M = 27.7, SD = 9.0). 26 were single, 26 lived with their
parents, 14 were married or living together with a partner, 3 were divorced and 1 lived in a
community centre. 2 patients completed primary education, 27 secondary education and 41
higher education or university. Using the SCID-II we assigned all patients to one of the three
clusters of personality disorders according to DSM-IV; cluster A: 2 (1 schizoid, 1
schizotypal); cluster B: 52 (1 histrionic, 36 borderline, 9 narcissistic, 3 narcissistic and
borderline, 3 histrionic and borderline); cluster C: 8 (2 dependent, 1 avoidant and dependent,
5 not otherwise specified); clusters A+B: 1; clusters B+C: 7. On DSM-IV Axis I, we
diagnosed 35 mood disorders, 4 anxiety disorders, 1 eating disorder, 20 adjustment disorders
and 8 substance related disorders. Five patients stayed 0 to 3 months; 16 patients, 3 to 6
months; 5 patients, 6 to 9 months; 30 patients, 9 to 12 months; and 14 patients, 12 to 13
months.
Procedure
The patients were assessed within the first two weeks of admission, every three months, and
upon discharge. In follow-up, the patients that stayed less than six months were again
assessed one year after admission. All other patients were assessed three months and one year
after discharge. Personality related measures were assessed at the first moment of assessment.
Measures
The same clinical outcome measures and measures of personality related characteristics
were used as in study 1.
166

167
Data Analysis
Multiple clinical outcome measures were reduced to a single global outcome component
score, the Global Symptom Score (GSS) by means of a Principal Component Analysis (PCA)
on the seven outcome-scores of the patients at the first measurement. The PCA yielded a first
factor accounting for 50 % of the total variance. The loadings on this principal component
were the following:.53 for the Self Harm Inventory (SHI 2w),.86 for the Symptom Check
List-Positive symptom total (SCLpst) ,.85 for the Beck Depression Inventory (BDI), .73 for
the Spielberger State, .89 for the Trait Anxiety Inventory Scores (STAI) and .44 for the
Spielberger State and .49 for the Trait Anger Inventory Scores (STAXI). This PCA solution
was used to calculate GSS component scores at each of the seven time moments.
We modelled global outcome trajectories during treatment and during the year after
treatment using the SAS TRAJ procedure (Nagin, 1999; Jones, Nagin & Roeder, 2001). This
procedure allows for the identification of distinct groups of trajectories and allows to estimate
the proportion of each trajectory group. Trajectory modelling is based on a semi-parametric,
group-based modelling strategy, which is of use in the statistical analysis of trajectories.
Technically this trajectory model is a mixture of probability distributions that are suitably
specified to describe the data that are to be analysed. Differences in patient characteristics
between groups with a different trajectory were studied with univariate analysis of variance
followed by unadjusted t-tests.
Results
The model with one trajectory for the Global Symptom Scale (GSS) showed the global
outcome change of the total study population (see Fig. 1). We found a significant
167

168
improvement during therapy, with a continuation of improvement during the post-treatment
phase, as reported in the Bateman and Fonagy study. The trend of the GSS decreases over the
period of treatment and follow-up was cubic (z = 2.19, p = < .05). This cubic trend shows that
there is a small improvement in the first three months of the therapy, a major improvement in
the following period which continues in the first three months after discharge, followed by a
stabilisation till the next assessment one year after discharge.
Figure 11. Global trajectory of seventy patients treated with hospitalization on analytic lines
for 1 year and 1 year follow-up.The dotted line presents the cubic trend.
Outcome Trajectories
The principal aim of this study was to differentiate outcome trajectories within the study
group. On the basis of information criteria (Akaike’s Information Criterion, Bayesian
Information Criterion) of models with 2, 3 and 4 trajectories, the four - trajectory model was
chosen. Analyses presented here are based on this four-trajectory model, with individuals
168

169
being assigned to their most likely trajectory of outcome change over time, using posterior
probabilities. The trends of the four trajectories are presented in Table 5, the membership of
the four trajectories in Table 6; the trajectories in Figure 12.
Table 5
Model of Four Change Trajectories of 70 Patients
Group Parameter Estimate SE t p
1 Intercept 80.08 1.66 48.33 .0000
2 Intercept 101.70 1.39 73.02 .0000
Linear -0.39 0.06 -6.37 .0000
Quadratic 0.00 0.00 4.07 .0001
3 Intercept 100.40 0.91 110.75 .0000
4 Intercept 112.14 2.38 47.09 .0000
Linear -0.00 0.12 -0.02 .9843
Quadratic -0.00 0.00 -2.14 .0330
169

170
Table 6
Group Membership of the Four Trajectories of Symptom Change
Group membership (%) SE t p
1 8.97 3.87 2.31 .0211
2 49.55 7.14 6.94 .0000
3 25.90 6.86 3.77 .0002
4 15.58 5.43 2.87 .0044
Figure 12. Four trajectory model of seventy patients treated with hospitalization on analytic
lines for 1 year and 1 year follow-up. The dotted lines presents the trends.
170

171
The first trajectory group consists of patients with very few symptoms at the start and
for whom the Global Symptom Score does not change significantly during the treatment or in
follow-up. This group includes 9 % of the patients.
The second and the third groups show the same level of clinical symptoms at the start.
Together they account for 75% of the patients. The second group shows a significant change
in clinical symptoms over time, while the third group does not. The relation of the symptom
change of the second group with the length of treatment is quadratic. The change in clinical
symptoms starts after three months and continues after discharge. This group comprises 49.5
% of the patients. The third group, starting at the same level of symptoms but with no change
over time, includes 25.9% of the patients.
The fourth group includes the patients with the highest severity of symptoms at the start.
This group includes 16 % of the patients. The trajectory of this group shows a quadratic trend:
change beginning late in the treatment with a marked continuing improvement in the post-
treatment period. One year after discharge, the level of symptoms of the fourth group is the
same as that of the second group.
Differences in Pre-Treatment Variables Among Groups of Outcome Trajectories
The difference in the manifestation of symptoms at the start is reflected in the difference
of intercept values of the Global Symptom Score of the four trajectories. These intercept
values show that the first group presents very few symptoms and the fourth group shows
many symptoms at the start, while group two and three start at the same intermediate level of
the global symptom score. The groups had a comparable percentage of drop-outs within the
171

172
first six months, except for the third which had a lower drop-out rate: traj. 1: 33%, traj. 2: 27
%, traj. 3: 12 %, traj. 4: 27 %.
The differences between the four groups in patient characteristics showed that the history of
self harm as measured with the SHI is discriminative (F(3,66) = 5.66, p = .002) for the four
groups. Self-reported traumatic experiences from the past, measured with the TEQ, were
significantly different among groups for the total number of experiences (F(3,66) = 3.63, p =
.02). Groups also differed regarding emotional traumatic experiences (F(3,66) = 5.40, p =
.002) and the difference between groups approached significance for physical traumatic
experiences (F(3,66) = 2.60, p = .06). There were no reliable differences with respect to
sexual traumatic experiences (F(3,66) = 2.14, p = 3,66, n.s.).
For the personality-related characteristics, the level of borderline personality
organization was significantly related to the trajectories. This was found for each of the three
dimensions of personality organization according to Kernberg (1996): primitive defences
(F(3,66) = 5.88, p = .001), identity diffusion (F(3,66) = 5.41, p = .002) and reality testing
(F(3,66) = 4.16, p = .009). The sum score of the IPO, a measure of the severity of borderline
personality organization, is significantly different for the four groups (F(3,66) = 6.57, p =
.0006).
Of the IIP scales, the BC 'vindictive' scale is the most different among the four groups
(F(3,66) = 4.76, p = .004). Additionally, PA 'domineering' (F(3,66) = 3.96, p = .01), DE 'cold'
(F(3,66) = 3.83, p = .01), JK 'exploitabe' (F(3,66) = 3.16, p = .03) and LM 'overly nurturant'
(F(3,66) = 3.25, p = .03) scales are significantly different. This is not the case for the FG
'socially avoidant', HI 'non assertive' and NO 'intrusive' scales. The total IIP score is highly
discriminative (F(3,66) = 6.68, p = .0005).
For the item scores on the DSM-IV axis II categories (SCID), we found a significant
distinction between the groups for the avoidant (F(3,66) = 6.11, p = .001), paranoid (F(3,66)
172

173
= 6.11, p = 002), borderline (F(3,66) = 5.40, p = .002), schizotypal (F(3,66) = 5.09, p = 003),
schizoid (F(3,66) = 4.14, p = .009) and depressive (F(3,66) = 4.54, p = .006) categories.
We further investigated the particular differences between these groups by post-hoc
tests. The results of the post-hoc comparisons are shown in table 7 and 8.
Table 7
Mean scores and Standard Deviations of the 4 Trajectory Groups on the Inventory of
Personality Organisation(IPO), SCID Scales and Symptom Checklist (SCL-90-R), With
Indication of Significant Differences Between the Groups
Trajectory group 1 Trajectory group 2 Trajectory group 3 Trajectory group 4
Scale M SD M SD M SD M SD
IPO sum 105.83bcd 29.08 140.03ad 26.5 135.24ad 23.12 164.55abc 30.98
SC avoid 1.67bcd 1.21 3.94ad 2.22 3.65ad 1.54 5.73abc 1.62
SC dep 2 1.41 3.19 1.93 3.59 2.4 4.36 1.75
SC oc 4 2.28 4.61 1.83 4.41 2 5.36 1.86
SC pa 1.33bd 1.37 3.22a 1.96 2.59 1.37 3.55a 1.7
SC sct 1.5 1.52 3.14 1.94 3.71 1.9 5.09 2.12
SC schiz 0.67bd 0.52 2.22ac 1.53 1.35bd 0.7 2.36a 1.29
SC par 1.33 0.85 3.81ac 2.14ac 2.65b 0.51 5.00ac 0.63
SC his 0.83b 2.04 2.17a 1.78 1.59 1.28 1.64 1.69
SC nar 2.33 3.39 4.22c 3.03 2.77b 1.79 4 2.97
SC bor 4.00bcd 2.9 8.69a 2.98 8.41a 3.3 10.36a 3.59
SC ant 0.83 2.04 0.75 1.08 1.12 2.21 1.46 1.86
SCL hos 7.00bd 1.49 9.89a 0.61 8.53bd 0.88 11.45ac 1.1
SCL pst 114.17 17.96 199.72 55.75 207.71 46.86 296.36 41.28 abcd differs significantly from (a) traj 1 (b) traj 2 (c) traj 3 (d) traj 4
173

174
Table 8
Mean scores and Standard Deviations of the 4 Trajectory Groups on theInventory of
Interpersonal Problems (IIP), Self Harm Inventory (SHI) ever, Traumatic Experiences
Checklist (TEQ) and Length of Stay, With Indication of Significant Differences Between the
Groups
Trajectory group 1 Trajectory group 2 Trajectory group 3 Trajectory group 4
Scale M SD M SD M SD M SD
IIP pa 1.5 1.64 8.97 5.88 8.71 3.98 9.82 5.53
IIP bc 3.33bd 2.34 10.56ac 5.99 7.12bd 4.36 12.27ac 6.94
IIP de 5.50bcd 5.79 12.53a 4.92 11.77a 7.47 15.46a 5.91
IIP fg 12 9.32 16.06 7.77 16.06 8.3 21.36 7.19
IIP hi 12 11.95 18.53 7.99 18.53 8.05 20.55 6.09
IIP jk 8.83 6.43 16.75 6.69 17 7.13 19.55 7.87
IIP lm 8 7.62 15.94 6.93 16.59 7.19 19.18 7.69
IIP no 3.67bcd 2.58 10.97a 6.66 11.24a 7.01 10.18a 6.31
SHI ever 3.50cd 3.33 6.56d 3.96 7.71 ad 4.09 11.82 a b c 6.51
TEQ emo 0.33 b c d 0.82 1.77 a d 1.57 1.88 a d 1.22 3.36 a b c 2.11
TEQ phy 0.00 c d 0 0.51 0.66 0.77 a 0.75 0.91 0.94
TEQ sex 0.67 b 0.82 0.86 c 1.26 1.71 b 1.49 1.46 1.21
TEQ tot 3.50 b c d 1.69 6.77 a d 0.7 7.23 a d 1 10.18 a b c 1.25
length stay 2.34 0.64 2.53 c 0.26 3.47 b 0.38 2.55 0.47 abcd differs significantly from (a) traj 1 (b) traj 2 (c) traj 3 (d) traj 4
We focused in the post-hoc comparison on the differences between group two and three,
which account for 75 % of the patients and were not discerned at a symptom level at the start
but showed a very different outcome trajectory. Distinguishing the patient characteristics of
these two groups is therefore clinically very meaningful. Comparing the Trajectory 2 group,
that shows change, with the Trajectory 3 group, who does not, we see that Trajectory 2 group
is not different in IPO sum (t = 0.61, p = .54) and SHI (t = -0.88, p = .38), but the Trajectory
2 group has higher scores on SC schiz (t = 2.31, p = .02), SC par (t = 1.89, p = .06), SC nar (t
174

175
= 41.77, p = .08) and IIP bc (t = 2.09,p = .04). Trajectory 3 group shows the highest TEQ
scores and the longest stay of all trajectory groups and showed more sexual traumata in the
past TEQ sex (t = -2.23, p = .03) and a longer length of stay (t = -2.04, p = .05) in
comparison with the Trajectory 2 group.
Comparing the Trajectory 3 group (poor change), with the Trajectory 4 group (more
severe pathology, but changes), we see differences in IPO sum (t = -2.83, p = .006) and SHI
(t = -2.39, p = .02) which reflect the difference in level of borderline personality organisation,
and further in SC avoid (t = -2.80, p = .007), SC par (t = -2.92, p = .005), SCL host (t = -
2.07, p = .04) and IIP bc (t = -2.37, p = .02).
Comparing the Trajectory 1 group (less severe pathology) with the Trajectory 2 and 3
groups, we see lower scores than the Trajectory 2 on TEQ total (t = -1.78, p = .08), IPO (t = -
2.90, p = .005), SC avoid (t = -2.69, p = .009), SC bord (t = -3.38, p = .001) and lower scores
than Trajectory 3 group on the same scales: TEQ total (t = -1.78, p = .08), IPO (t = -2.32, p =
.02), SC avoid. (t = -2.17, p = .03), SC bord (t = -2.94, p = .004). However the Trajectory 1
group has significantly lower scores than the Trajectory 2 group on SC pa. (t = -2.45, p =
.02), SC schiz. (t = -2.76, p = .007), SC par. (t = -2.70, p = .009), SCL host. (t = -1.8,p =
.08), IIP bc (t = -2.92, p = .005), IIP de (t = -2.72, p = .008) and TEQ emo (t = -2.10, p =
.04) while the differences between the Trajectory 1 group and the Trajectory 3 group are not
significant for the same scales.
Discussion
Time related outcome changes corroborate the findings of the RCT study of Bateman
and Fonagy (1999, 2001) which shows the effectiveness of a hospitalization on analytic lines
for borderline personality disorders. For a group of patients within the spectrum of borderline
175

176
personality organization that was treated by a similar hospitalization on psychoanalytic lines
for a period of at least nine months, we found a significant decrease in self-harm, anxiety,
anger, depression and global symptoms with a continuation of the improvement after
discharge. Compared with the Bateman and Fonagy (1999) treatment group, our study shows
a significant decrease in symptoms but starts and ends with lower mean scores on the
symptom scales. This may be due to the fact that our study group consists of a heterogeneous
group of personality disorders, while the Bateman and Fonagy (1999) treatment group
comprised selected DSM IV axis II borderline patients. Furthermore we waited for the first
assessment till the patients were acclimatized on the setting.
We found a significant decrease during the treatment and a further decrease in the three
months after discharge, not only in symptom scores, but also in all scores related to
personality characteristics. These scores reflect change in broad domains of human life such
as self-experience, coping strategies, affect-regulation, relationships, work. These
characteristics are more stable than volatile symptoms and may reflect so-called structural
changes. This structural change is particularly reflected in the change of the IPO, which
measures the borderline personality organisation. A further argument for structural change
may be found in the continuation of improvement of most measures after discharge from the
treatment program. This was found in the Bateman and Fonagy study as well. However in the
Bateman and Fonagy follow-up study (2001), 75% percent of the patients continued with a
twice weekly group psychotherapy in the post -treatment phase. In our study group no such
organised post- hospitalization treatment was offered at the time of the study, however 50 %
of the patients of the study group started an ambulant psychoanalytical psychotherapy - a
form of treatment that was not successful before the hospital treatment, hence the admission.
176

177
These results suggest that psychic change happens during treatment resulting in an
outcome which is enduring and fruitful in the long term. A psychoanalytically informed
treatment aims at such psychic change rather than being symptom oriented.
Given a continuing improvement after discharge, it is a point for discussion whether in
the studied program with a maximum stay of one year, the patients are discharged too early
and at a moment they are still improving, or whether this time limit is an important
therapeutic factor as the study of Chiesa (2000) shows. Chiesa compared a two-stage model (a
shorter in-patient admission followed by outreached therapy) with a one-stage treatment
model without aftercare and found that the former yielded significantly better results for
borderline patients.
The main aim of the second study of the large group was to examine the trajectories of
change behind the global result of the whole group of heterogeneous personality disorders.
Given the heterogeneity of this group, it was possible to examine whether these outcome
trajectories were related to pre-treatment patient characteristics. With such data, it would be
possible in naturalistic settings to predict more accurate treatment outcome from the outset
and to adapt the treatment to the needs of the patients.
As expected, we saw different trajectories related with the symptom level at the start in
this study group: in relation with the symptom baseline of this studygroup we can speak of a
low symptom level, a medium symptom level and a high symptom level. The majority of the
patients starts at the medium level and then we see two different trajectories: one with a good
and one with a bad outcome.
A small group of patients presented few symptoms at the outset, and hence showing few
clinical outcome change. This group had a high level of borderline personality organisation.
These patients were in crisis when admitted to the hospital, but they recovered fast and were
admitted to the treatment setting on psychoanalytic lines because of the diagnosis of a
177

178
personality disorder in order to obtain a psychic, structural change to prevent similar
breakdowns in the future and in order to increase their quality of life. We cannot judge
whether this is achieved from the clinical symptom outcome.
The trajectory of the group with a high frequency of symptoms at the start is interesting.
This group shows not only significantly more symptoms (SCL- 90-R) but also more suicidal
treat and self-harm, a higher SCID-sum score and higher IPO -scores compared to the other
groups. These patients have a low level borderline personality organisation. Post hoc tests
show that they are considerably more paranoid, hostile and vindictive than the other groups.
They are so-called difficult patients. The trajectory of this group is remarkable: they only start
to improve in the last phase of the treatment, an improvement that goes on in the post-
treatment phase. It is a group that puts a lot of stress on the staff and the other patients and the
poor change across several months often raises questions about their suitability for this kind
of treatment. The marked improvement which starts only after a long period might be an
argument that it is important to have patience with these patients and to tolerate their
disruptive behavior within a holding, secure environment with a coherent treatment program
(Bateman & Fonagy, 2004). The continuing improvement in the post-treatment phase may
indicate that this group found another way of dealing with their problems than before
treatment. It is possible that some structural change occurred and that these patients
internalized other ways of dealing with their inner chaos and suffering, and perhaps finding
ways to rely on others instead of evacuating their inner suffering and distress in self
destructive behavior.
The group with a moderate level of symptoms at the start was our main interest. Not
only because they are the majority of the patients in a naturalistic setting (seventy-five percent
of the patients in this sample), but also because in this group we have a group with a trajectory
of good outcome and a group with a trajectory of poor outcome with the therapy as it is. The
178

179
severity in borderline personality organisation is not different between these two groups at the
start: the level of personality organization (IPO), the level of borderline symptoms (SCID)
and self harm (SHI). Therefore, it was important to study whether we could find other
characteristics which differentiate the two groups at the start. This would it make possible to
predict more accurate outcome according to patient characteristics and to adapt the treatment
program in the future to the needs of specific patients. We found that the group with good
outcome showed more avoidant, schizoid, paranoid and narcissistic features (SCID) and had
more difficulty in expressing emotions, as rated with the IIP than the group with poor
outcome. This also corroborates with the findings of attachment studies which show that
patients with a dismissive-avoidant attachment style respond best to psychodynamically
oriented interpretive, explorative therapy (Blatt and Shahar, 2004; Gabbard 1994; Fonagy,
1996). The similarity of this group with Blatt's categorisation of a introjective group (Blatt
and Blass, 1996, Blatt & Auerbach, 2003) is striking. He found the introjective group being
composed of individuals with paranoid, schizoid, avoidant and narcissistic personality
disorders, exactly the personality characteristics which specify this trajectory group with a
good outcome. In his analysis of the data of the Menninger Psychotherapy Research Project
(Wallerstein, 1986), he found that these patient characteristics discerned the group of patients
that clearly benefited from a psychoanalytic, explorative treatment. In fact Blatt and
colleagues integrated the diversity of personality disorders into two primary organisations
based on the differentiation of two polarised psychological dimensions of relatedness and
self-definition. The introjective group shows a greater preoccupation with issues of self-
definition, while the anaclitic group shows a greater preoccupation with issues of relatedness.
In this study, the group with a poor outcome shows all the borderline features but without a
concomitant controlling and introjective attitude. This group reported far more sexual abuse
and corresponds to the so-called anaclitic group, where Blatt found the mutative dimension
179

180
being a supportive therapeutic relationship and not so much insight and self-understanding as
in the introjective group. They tend to improve by relying on the support and structure of the
therapy rather than by an explorative attitude, certainly in the beginning of the therapy. Their
dependency on relationships makes that this facet needs to be addressed first before there is
enough therapeutic alliance to work in an explorative way. The poor outcome of the third
group is probably related to the fact that the specific needs of this group were not met enough.
Our impression is that these patients were included too soon in a classic psychoanalytic
interpretive group approach and were easily overwhelmed by this approach. Our process-
outcome study of the inner change during treatment of borderline patients points in the same
direction (see Chapter 8). According to Blatt and Shahar (2004), this group develops an
unrealistic positive therapeutic alliance towards the end of the treatment and the ending of the
therapy is difficult. The fact that they stay longer and show fewer drop-out despite of a poor
outcome may reflect the insecure attachment (Holmes, 2001) described in such patients.
The late improvement of the group with the very difficult patients, borderlines with a
high level of psychopathology at the start, is thought-provoking. It may be that these patients,
which clinical picture is more severe, are more demanding and threatening and therefore for a
very long time approached in a supportive way with many individual contacts and a long
delay before they take part in the intensive group therapy sessions. They do not run the risk to
be included too soon in an explorative phase of the treatment, consisting of classic
psychoanalytic group approach. It may be that the psychic change within the treatment under
study happens mainly by a mechanism of introjection and that it takes a long time before this
change becomes manifest in a decrease of clinical symptoms, and the longest time for the
most severe group.
180

181
Both groups with a good outcome show a more introjective style than the groups with a
poor outcome. This difference in personality style seems to have a strong impact on outcome
within a certain kind of treatment, maybe a stronger impact than the level of psychopathology.
In conclusion:
We found an improvement in the symptom and personality measures of personality
disordered patients during psychoanalytically informed treatment. The improvement
continued in the post- treatment period, as in the Bateman and Fonagy RCT study
(1999,2001). This is an argument for a structural change, which will be further examined in a
process-outcome study (Chapter 8).
The major finding of this study is that the outcome trajectories of personality disordered
patients in psychoanalytic informed hospitalisation relate to pre-treatment patient
characteristics. We discerned a high, medium and low level borderline symptomatology at the
start. Patients with a high level of borderline symptomatology, that are more hostile and
paranoid, take a very long time before they start to change. This change seems related with
introjective mechanisms. But it is worth to have patience and to contain for a long time their
threatening behaviour because post treatment changes are prominent. In the medium level of
borderline symptomatology, the different trajectories depend on the personality style. The
introjective, more avoidant and narcissistic borderline patients profit from the
psychoanalytically informed therapy as it is, while the more anaclitic borderline patients seem
to depend on relational support which has probably to be continued after the treatment. These
patients seem to gain less profit from this explorative treatment as it is. Clinicians tend to
predict outcome according to the severity of symptoms. This study shows that this distinction
is lacking and that differences in personality style, reflected in coping and interpersonal
relation, and already present at the outset of the therapy are important factors in predicting the
181

182
response to the treatment and the outcome. Knowing this distinction, it becomes possible to
adapt the treatment to the needs of the patients and further ameliorate the effectiveness of the
psychoanalytically informed hospitalisation based treatment for personality disorders.
182

183
Chapter 7
Patient Attrition in a Psychoanalytically Informed Hospitalization-
Based Treatment for Personality Disorders
As discussed by Clarkin and Levy (2004) and Chiesa (2000), clients with personality
disorders have been found to be at high risk for dropping out of treatment. The authors argue
that with the large numbers of patients who leave therapy prematurely, study in this area
should be given high priority. Indeed, the literature on personality disorders shows drop-out
rates varying between 32% and 66% in hospitalization-based treatments (Skodol et al., 1983;
Gunderson et al., 1989; Kelly et al., 1992; Chiesa, 2000). Reported differences between
attrition groups and therapy groups related to client characteristics are contradictory. Skodol
et al. (1983) found more severe symptoms in the attrition group at the start of the treatment
but Gunderson et al. (1989), Kelly et al. (1992), Chiesa (2000) and Thunissen (2004) did not
find such differences. In a study of ambulant psychodynamic psychotherapy, Smith,
Koenigsberg, Yeomans, Clarkin & Selzer (1995) found more hostility in the attrition group of
borderline patients.
The present study was designed to examine the attrition group in a psychoanalytically
informed hospitalisation based treatment for patients with personality disorders, comparable
to the setting in the study of Chiesa (2000) and to the treatment program described by
Bateman and Fonagy (2004).
We were interested in identifying factors that predicted early termination. Therefore, in
a first study we examined the differences on pre-treatment variables between the attrition
group and the longer- term therapy group and in a second study we looked at differences in
the psychotherapeutic process at three months for the drop-out group that stayed at least that
183

184
long. In a third study we compared the outcome of the drop-out group and the therapy group
one year after admission regarding clinical symptoms, personality related measures and
psychoanalytic process measures.
Treatment
Hospitalization-based treatment (inpatient and dayhospital) on psychoanalytic lines for
personality disorders at the University Center Kortenberg, Belgium. The treatment focuses on
three dimensions of inner change: felt safety, mentalization, and self-other object relations
(Chapter 1).
Definition of Drop-out
For this study, we applied a definition of attrition in a small and in a broad sense.
In a small sense we rely on the definition of Chiesa (2000), who defined drop-out in a
similar setting as any premature termination of treatment not mutually negotiated and agreed
upon by staff and patient occurring within 14 weeks of admission. We call it the 'small DO
group'.
In a large sense we define the drop-out group as all patients who left the therapy
program earlier than was negotiated at the beginning of treatment. This is relative as the
treatment was open-ended with a focus on the psychic process and on the possibility to switch
to an ambulant psychotherapy when possible. However, prior to therapy, there was an
agreement with the patients that the hospitalization based treatment (in-patient or partial
hospitalization) takes at least 6 months till a maximum duration of 12 months. By this
184

185
definition every patient who fell out of the program in the first six months was part of the
attrition group. We call it the 'large DO group'.
In the second study we examine the therapeutic process parameters at three months for
the drop-out patients which stayed at least three months. We call this group the ' late DO
group'.
Table 1 shows how these DO groups are related.
Table 1
Large Drop Out group, With Indication of the Small Drop Out Group
With advice of staff Against advice of staff
Charateristic Less than 3M 3M-6M Less than 3M 3M-6M
Male 1 1 2 1
Female 2 5 7 5
In-patiënts 2 5 8 5
Day-hospital 1 1 1 1
n 3 6 9 6
Late DO
group
Small DO
group
Late DO
group
185

186
Study 1
In this study we compared the pre-treatment variables of the patients of the small DO group
and of the large DO group with those of the therapy group that stayed at least nine months.
Method
Participants.
For a process-outcome study (Chapter 1), we assessed 78 consecutively-admitted
patients in the period from 23 -5-01 to 17- 7-02. Of these 78 patients, one patient refused to
take part in the study and one patient with an experience in psychological tests was deemed to
have unreliable test results and was removed from the study. Of the 76 consecutively admitted
patients, 26 left treatment within the first six months. With two of these 26 patients, a short
stay was already decided at the start of the treatment (one because of professional reasons,
one because of a rehospitalization). With the broad definition of drop-out, this brings the
drop-out before six months of treatment to 24 (15 female, 9 male), this is the large drop-out
group (see table 1).
Of the large DO group, twenty patients followed a treatment in an open-ended hospital
setting and 4 in a day-treatment setting. 15 patients were between the ages of 17 and 46 (M =
25.8, SD = 7.8). Ten were single, 9 lived with their parents, 4 were married or living together
with a partner and 1 lived in a community center. Three patients completed primary
education, 10 secondary education and 11 higher education or university. Using the SCID-II
we assigned patients to one of the three clusters of personality disorders according to DSM-
186

187
IV; cluster A: 2 (1 schizoid, 1 schizotypal); cluster B: 17 (1 histrionic, 13 borderline, 1
narcissistic, 1 narcissistic and borderline, 1 histrionic and borderline); cluster C: 2 (1
dependent, 1 not otherwise specified); clusters A+B: 1; clusters B+C: 1; unknown: 1. No
patients committed suicide. As shown in table 1, of this 'large DO group', 9 patients left with
advice of the staff: 3 patients were transferred to a setting for young psychotic patients, 2
patients were transferred to a secluded ward because of severe danger for suicide, 1 patient
was discharged because of breaking the rules and antisocial behavior, 2 because of
improvement and stepping over to ambulatory treatment and 1 patient was transferred to a
rehabilitation setting because he did not respond to the treatment approach.
In the ' small DO group' drop-outs were limited to the 9 patients who left against advice
during the first three months. Of the two males and seven females, eight followed a treatment
in an open-ended hospital setting and one in a day-treatment setting. The patients were
between the ages of 17 and 38 years (M = 23.2, SD = 6.0). Four were single and 5 lived with
their parents. Three patients completed primary education, 2 secondary education and 4
higher education or university. Using the SCID-II we assigned patients to one of the three
clusters of personality disorders according to DSM-IV; B: 7 (7 borderline); C: 1 (1 not
otherwise specified); unknown: 1.
The large DO group and the small DO group were compared with the therapy group
defined in the prospective process-outcome study (see Chapter 6 & 8) as the group that stayed
at least 9 months. The latter group consisted of 47 patients (14 males, 33 females); 34
followed a treatment in an open-ended residential setting and 13 in a day-treatment setting.
The patients were between the ages of 17 and 58 (M = 27.8, SD = 9.1). Eighteen were single,
19 lived with their parents, 7 were married or living together with a partner and 3 were
divorced. Seventeen patients completed secondary education and 30 higher education or
university. Using the SCID-II we assigned all patients to one of the three clusters of
187

188
personality disorders according to DSM-IV; B: 34 (23 borderline, 7 narcissistic, 2 narcissistic
and borderline, 2 histrionic and borderline); C: 7 (1 dependent, 1 avoidant and dependent, 5
not otherwise specified); B+C: 6. On DSM-IV Axis I we daignosed 19 mood disorders, 4
anxiety disorders, 1 eating disorder, 6 adjustment disorders and 6 substance related disorders.
Procedure.
The patients were assessed within the first two weeks of admission.
Instruments.
In assessing the clinical symptoms we used the Self Harm Inventory about the last two
weeks (Sansone et al., 1998), the Symptom Checklist 90 (SCL-90) (Derogatis, 1977),
translated by Arrindell and Ettema (1993), the Spielberger State-Trait Anxiety Inventory (
Spielberger, 1970) translated by Van der Plaog et al. (1980), the Spielberger State-Trait Anger
Inventory (Spileberger et al., 1980) translated by Van der Ploeg et al. (1982) and the Beck
Depression Inventory (Bouman et al., 1995). As personality related measures we used the
Dutch translation of the SCID II ( DM III-R, Axis II; Spitzer and Williams,1985)
questionnaire and its adaptation for the DSM IV by Weertman, Arntz & Kerkhofs (2000), the
Inventory of Personality Organisation (Kernberg and Clarkin, 1995) translated ad validated in
Dutch by Vermote, Vertommen et al. (2004) and Vermote, Smits, Claes, Vertommen (2005)
and the Inventory of Interpersonal Problems, circumflex version (Alden et al., 1990) which
measures the interpersonal functioning. The Self Harm Inventory (Sansone et al., 1998)
measures the history of suicidality and self-harm. The Traumatic Experience Checklist
(Nijenhuis et al., 2002) measures the history of trauma.
188

189
Data analysis.
A classical t-test was performed to compare the means of scores on admission between
the therapy group and the large DO group and the small DO group. Because we had the
clinical impression that the large DO group consisted of patients who had a personality
organisation (measured by the IPO) that was either too severe to be treated in the setting or,
conversely, too high functioning to be treated in the setting, comparing mean scores could
have given a false impression. Therefore, we calculated the difference between the score of
each individual patient and the mean for the patients of the large DO group for whom this
value could be assessed (n = 14) and for the 47 patients in the therapy group. Then we
compared these difference scores of the large DO group and the therapy group with a t-test.
Results
When comparing the small DO group with the therapy group (Table 2), we found that
the group that left within the first three months against advice of the staff was less depressed
(SCL, SCID, BDI), less anxious (Spielberger scales), but, again, significantly angrier (STAXI
state and trait). These characteristics were also reflected in the frequency of items from DSM-
IV Axis II categories: the therapy group had more items rated in the controlling, anxious and
depressive sphere, associated with the avoidant, obsessive-compulsive and depressive
diagnostic categories.
When comparing the large DO group and the therapy group we found no significant
differences on the following measures: TEQ, SHI, BDI, IIP, SCID, SCL, IPO, and patient
189

190
Table 2
Significant Differences in Pre-Treatment Variables for the Small DO group and the Therapy
Group
Small DO group Therapy group Statistical Test
Subscale M SD M SD t p
SCID avoidant 2.44 1.42 4.11 2.056 -2.31 .02
SCID obsess-comp 3 2.29 4.53 1.92 -2.13 .04
SCID depressive 3.67 2.5 5.30 1.64 -2.50 .02
IIP fg avoidant 8.71 4.07 16.94 7.97 -2.66 .01
IIP Hi nonassertive 11.71 6.70 19.28 7.77 -2.44 .02
SCL depressive 34.14 12.81 47.17 14.80 -2.20 .03
SCL sensitive 30 7.19 38.60 10.81 -2.03 .05
BDI 15.17 8.38 25 10.57 -2.19 .03
STAI state 33.33 23.340 44.05 13.22 -2.32 .02
STAI trait 33 23.13 59.17 10.46 -4.91 < .0001
STAXI state 27.67 21.27 11.87 3.04 5.02 < .0001
STAXI trait 32.67 14.60 18.06 6.61 4.34 < .0001
age. Patients of the large DO group were significantly angrier as measured by the Spielberger
Trait Anger Inventory Score, t(64) = 2.89, p = .005) with a mean score of 24.35 (SD = 10.98)
versus the therapy group mean of 18.06 (SD = 6.61). A similar pattern was found with the
Spielberger State Anger Inventory Score, t(64) = 2.43, p = .02, with mean scores of 16.9 (SD
= 13.50) versus 11.87 (SD = 3.04), respectively. On the Spielberger Trait Anxiety Inventory,
the large DO group was characterized by less anxiety as a trait than the therapy group, t(64) =
190

191
-2.12, p = .04, 51.2 (SD = 20.37) versus 59.17 (SD = 10.46). When comparing the IPO
difference scores, between the large DO group and the therapy group, we found the
differences to be significant, t(58) = 3.21, p = .002. The results indicated that it was the
patients with the most and the least severe personality organisation who terminated early.
Study 2
In this study, we compared the change in process measures of drop-out patients who
stayed long enough to have a second assessment at three months with comparable data for the
therapy group. The process is judged by the therapists and by the patients themselves.
Method
Participants.
Patients of the attrition group who stayed between three and six months, the so called '
late ' DO group, were included in these analyses and were compared to those in the therapy
group from the first study. (Table 1: 12 patients, but missing data of 2 of 10).
This late DO group consisted of 10 admitted patients (4 males, 6 females); 8 followed a
treatment in an open-ended residential setting and 2 in a day-treatment setting. The patients
were between the ages of 18 and 46 (M = 25.2, SD = 7.9). Three were single, 4 lived with
their parents and 3 were married or living together with a partner. Six patients completed
secondary education and 4 higher education or university. Using the SCID-II we assigned
191

192
patients to one of the three clusters of personality disorders according to DSM-IV; cluster A:
2 (1 schizoid, 1 schizotypal); cluster B: 6 (1 histrionic, 4 borderline, 1 narcissistic and
borderline); cluster C: 1 (1 dependent); clusters B+C: 1.
Instruments.
We used the Leuven Psychotherapy Scale (LPS), to study the process from the
perspective of the patients. The construction of the Leuven Psychotherapy Scale (LPS), its
reliability and congruent and divergent validities are discussed in Chapter 4. The LPS self-
report scale is composed of 112 items that are ordered, following factor analysis, into 7
subscales: General positive well-being, Openness to authentic relationships, Capacity to deal
with unexpected events, Capacity to deal with negative emotions, Capacity for interiorisation,
Positive experience of the setting and Control over substance use
We used the Psychoanalytic Process Rating Scale (PPRS, Stoker & Zevalkink, 2001,
2005) to study the psychoanalytic process from the therapist's point of view. The scale
focuses on the features of contact development in a session and is based on theory driven
items. The scale counts 47 items divided over 5 subscales: Cooperation, Tension, Exploration,
Disillusion, Contentment..
The Nurse's rating scale of the process, is a 6- point categorical scale that the nurses
use in rating the treatment process from their point of view ( Table 3).
192

193
Data analysis.
The results of the change between the first and second measurements were examined
with Analysis of Variance, with one repeated measures factor (time) and one between subjects
factor (group).
Table 3
Nurses Rating Scale of the Treatment Process
Nurses Rating Scale of the treatment process
-1
0
destructive, poor contacts, very poor participation, breaking the rules
discussion about breaking the rules is possible, poor participation
1
2
conscious about what is going wrong, more control, partcipation but not active
control, good participation at therapies, trust, contacts with fellow patients
3 good control, infrequent breaking of the rules, attached to at least one fellow
patient
4 active participation in therapies, absence of destructive acts, close relationship
with more than one fellow patient
5 rich contacts, active and independent partcipation, helps in creating psychic
space and therapeutic culture at the setting
193

194
Results
The late DO group and the therapy group did not differ significantly on the difference in
process measures between the beginning of treatment and assessment at three months. These
findings held for the process as evaluated by the patients with the LPS and the process as
evaluated by psychotherapists with the PPRS. There was, however, a group by time
interaction effect for the nurses' rating of the process (F(1,54) = 6.50, p = .01), with the late
DO group starting at a mean score of.90 (SD = 1.10) at moment 1 and decreasing to a mean
score of.40 (SD = 1.58) three months later. This is in contrast with the therapy group, which
starts with a mean score of 1.00 (SD = 0.69) and increases to a mean score of 1.6 (SD = 0.90)
three months later.
Study 3
In this study we compared the outcome of the large DO group with the outcome of the
therapy group for symptom measures, personality related measures and measures related with
the therapeutic process.
Method
Participants.
We obtained assessments of 13 patients from the DO group one year after admission.
Two patients were left out of the study group, as they were treated during this year in a
different intensive hospitalization based behaviour treatment setting in our hospital (Pieters &
194

195
Vermote, 2002). Thus the final DO follow-up group consisted of 11 patients (5 males, 6
females); 8 followed a treatment in an open-ended hospital setting and 3 in a day-treatment
setting. The patients were between the ages of 18 and 46 (M = 25.3, SD = 7.8). Four were
single, 5 lived with their parents and 2 were married or living together with a partner. Two
patients completed primary education, 4 secondary education and 5 higher education or
university. Using the SCID-II we assigned the patients to one of the three clusters of
personality disorders according to DSM-IV; cluster A: 1 (1 schizoid); cluster B: 8 (1
histrionic, 6 borderline, 1 narcissistic and borderline); cluster C: 1 (1 dependent); clusters
B+C: 1. Four patients continued a psychiatric follow-up, no one of the DO group received
further psychotherapy.
The therapy group is the same as in study 1 and 2.
Procedure.
DO patients who left the treatment were assessed again one year after admission and the
therapy group was assessed three months after discharge. The lapse of time between
admission and this follow-up assessment was comparable for the therapy group and the DO
group.
Measures.
In assessing the clinical symptoms we used the Self Harm Inventory about the last two
weeks (Sansone et al., 1998), the Symptom Checklist 90 (SCL-90) (Derogatis, 1977),
translated by Arrindell and Ettema (1993), the Spielberger State-Trait Anxiety Inventory (
Spielberger, 1970) translated by Van der Ploeg et al. (1980), the Spielberger State-Trait Anger
195

196
Inventory (Spielberger et al., 1980) translated by Van der Ploeg et al. (1982) and the Beck
Depression Inventory (Bouman et al., 1995). As personality related measures we used the
Dutch translation of the SCID II ( DM III-R, Axis II; Spitzer and Williams,1985)
questionnaire and its adaptation for the DSM IV by Weertman, Arntz & Kerkhofs (2000), the
Inventory of Personality Organisation (Kernberg and Clarkin, 1995) translated ad validated in
Dutch by Vermote, Vertommen et al. (2004) and Vermote, Smits, Claes, Vertommen (2005)
and the Inventory of Interpersonal Problems, circumflex version (Alden et al., 1990) which
measures the interpersonal functioning. The Self Harm Inventory (Sansone et al., 1998)
measures the history of suicidality and self-harm. The Traumatic Experience Checklist
(Nijenhuis et al., 2002) measures the history of trauma.
In assessing the psychoanalytic process, we used the measures to be rated on the ORI by
independent researchers (discussed in Chapter 3): the Felt Safety Scale, the Differentiation
Relatedness Scale, the Reflective Functioning Scale and the Bion Grid Scale and the Leuven
Pyschotherapy Scale, a self-report scale where the process is measured from the perspective
of the patient (discussed in Chapter 4).
Analysis of data.
Analysis of Variance, with one repeated measures factor (time) and one between
subjects factor (group).
196

197
Results
Interaction effects were not significant for the symptom and personality related outcome
measures except for the measure which had differed most at the beginning of the treatment
between the therapy group and DO group: the Spielberger Trait Anger Scale (F(1,53) = 3.23,
p = .08), changed from 22.45 (SD = 7.83) to 17.00 (SD = 5.00) for the drop-out group, while
the therapy group changed from 17.89 (SD = 6.70) to 15.53 (SD = 4.70).
However there is a significant interaction effect between the drop-out group and the
therapy group for measures of the psychoanalytic process such as the Differentiation Relation
Scale about Self: F(1,56) = 4.46, p = .04 with a change from 5.00 (SD = 1.35) to 4.92 (SD =
1.68) for the drop-out group and a change from 4.93 (SD = 1.04) to 6.02 (SD = 1.27) for the
therapy group and a significant difference change in Felt Safety F (1,56) = 3.15, p = .08) with
a change from 2.27 (SD = 0.47) to 3.09 (SD = 1.22) for the drop-out group and a change from
2.30 (SD = 0.66) to 3.72 (SD = 0.93) for the therapy group.
Discussion
We delineated a small DO group according to Chiesa (2000), that is, a group that left
against advice within the first three months. With this definition, the attrition rate of our
treatment setting is low, 11.8% (9 of 76), compared with attrition rates ranging between 32%
and 66 % in hospitalisation based treatments for personality disorders using the same
definition (Skodol et al., 1983; Gunderson et al., 1989; Kelly et al., 1992; Chiesa, 2000).
Reasons for the low drop-out rate are probably found in a strategy which we imply since
197

198
several years and which corresponds for the most part to the recommendations of Blount,
King & Menzies (2002) and Chiesa, Drahorad & Longo, S. (2000) to mitigate drop-out rates.
There is a pre-hospitalization phase during which there are contacts between patient and
family and the psychiatrist and the head nurse; during this phase patients are provided with
coherent information about the treatment program. In this phase a verbal contract is made as
well. When admitted, patients participate in a special sociotherapy group for newcomers and
in an introductory psychotherapy group where the anxiety level is kept low and the goal is
trying to deal mentally with what happens in the here-and-now. During this initial phase, the
patients have individual sessions with the therapists as well. Only when the patients are
capable to do so, they switch to the proper thrice-weekly, 90-minute group analytic sessions.
At the wards, there is a structured and coherent program, 24 hour staffing and several avenues
of response when patients are in emotional crisis. The staff pays special attention to patients'
felt safety, which is seen as one of the primordial mediators of the treatment. This strategy is
discussed at length in chapter 1.
When the pre-treatment variables of the small DO group and the long term therapy
group were compared, the DO group was less anxious, less depressed than was the therapy
group at the start but reported considerably more anger.
We examined drop-out from a broader perspective as well. During the intake sessions,
an agreement was made with patients that a period of at least six months of psychodynamic-
informed inpatient or partial hospitalization would be necessary to promote more lasting
changes. In a broad sense we can consider all patients who stop the treatment earlier as drop-
outs. Even in this broad sense the attrition rate is 32%, which still puts the program at the low
end of the range. However, drop-out is relative in this large DO group as the treatment was
open-ended in this hospital setting, and patients stop the treatment for several reasons with or
without advice of staff often by transfer to ambulatory treatment or to another setting. A
198

199
major reason of this drop-out is that patients who do not meet all indication guidelines are
often given a try-out phase. This may be the reason that there is a tendency for a mainstream
effect and explains why patients in this large DO group are of the highest and the lowest level
of personality organization when compared with the therapy group. When this large drop-out
group was compared with the therapy group, it showed significantly more aggressive features
just as the small DO group did. In our study, anger was clearly shown to be the most
predictive symptom for early termination of the treatment. This corroborates the findings of
Smith et al. (1995), who found more hostility in the attrition group of borderline patients in
ambulant psychodynamic psychotherapy.
We further examined whether the evaluation of the therapeutic process by the therapists
and by the patients themselves yielded a predictive factor. Therefore we compared the
patients of the large DO group, that stayed at least three months, to have an assessment of the
process, (the so called 'late DO group') with the therapy group. There was no interaction effect
for the rating of the process by the patients or the therapists, except for the evaluation of the
process by the nurses. The nurses judged the overall participation on the ward, and we
hypothesize that they possess a broader view of whether a patient was engaging in the
treatment as a whole or not. This suggests that their evaluation is of considerable importance
when trying to improve attrition rates.
As far as outcome differences between drop-out and therapy groups are concerned,
psychotherapy research shows that, in general, drop-outs from different kinds of
psychotherapy do not have a bad outcome, although it is seen as a missed chance (Clarkin &
Levy, 2004). The results of the present study tend to corroborate this: we found no significant
interaction effects between the large drop-out group and the therapy group in any of the
symptom scales or personality-related scales.
199

200
This is in contrast with changes measured with the process parameters. As discussed in
chapter 1, the treatment focused on inner changes in three dimensions: felt safety, integration
of split mental representations of self-other relations, and mentalization. When these inner
changes of the large DO group were compared to the changes in the therapy group one year
after admission, there was a significant interaction effect on the integration of mental
representations or inner object relational patterns, as measured with the DRS (Differentiation
Relatedness Scale) especially on the Self subscale and on Felt Safety. This suggests that
staying longer in treatment has an effect on these dimensions of inner psychic functioning and
this is what is the primary goal of the treatment. The relation between these inner changes and
outcome is studied in detail in a process-outcome study (Chapter 8).
200

201
PART 4
THE RELATION BETWEEN PROCESS AND OUTCOME
201

202
202

203
Chapter 8
The Kortenberg - Leuven Process Outcome Study on Patients with
Personality Disorders in Psychoanalytically Oriented Hospitalisation4
The RCT study of Bateman and Fonagy (1999, 2001) showed the effectiveness of
psychoanalytically informed day hospitalization of patients with a borderline personality
disorder. The specificity of a psychoanalytic therapy is that it is not symptom oriented but
focuses on intrapsychic changes. These intrapsychic changes over time are denoted as the
psychoanalytic process in contrast to symptom outcome. It is only one of the many conflicting
ways of defining the psychoanalytic process (Boesky, 1990; Weinshel, 1990; Abrams, 1990).
Given the effectiveness of this treatment in symptom outcome, we wanted to know whether
these changes in symptoms are related with the intrapsychic changes targeted by the
treatment.
This is important as relations between the psychoanalytic process and outcome were
found to be weak or non-existing (Vaughan and Roose, 1995) while psychoanalytic therapy is
effective (Leichsenring & Leibing, 2003). Furthermore very different therapies for personality
disorders show similar results (Blatt, 2003), which is known as the dodo verdict referring to
Lewis Caroll's ' Alice in Wonderland': all have won, everybody shall have prices. There may
be different reasons for the dodo effect such as the use in RCT studies of uniform symptom
scales which may fail to grasp treatment or model specific changes, or the focus in
effectiveness studies on limited homogeneous groups or the fact that therapeutic agents may
4 Granted by the Research Advisory Board of the International Psychoanalytical
Association in 2001, 2002, 2003 and by Lotto
203

204
be aspecific and shared by all kind of treatments. Given the intensity of a psychoanalytically
informed treatment, it is important to know whether targeted intrapsychic changes are reached
and whether there is a relation with outcome. To avoid aspecific results because of
methodological reasons, it is important to use model specific measures in a naturalistic design
(Clarkin & Levy, 2004).
The relation between the hypothesized intrapsychic changes and symptom changes, is
studied in a group of consecutively admitted patients in a psychoanalytically informed
hospitalization based treatment for personality disorders (day and residential) at the
University Centre St.-Joseph at Kortenberg. The treatment is described in chapter 1 and
comparable to the one described by Bateman and Fonagy (2004). Two setting characteristics
distinguished the present setting and that described by Bateman & Fonagy (1999, 2001): the
present study was an open-ended in-patient and day-hospital treatment with a maximum stay
of approximately one year and a follow- up of one year, rather than 18 months of day hospital
treatment only with a follow-up of 18 months.
We used the same clinical symptom outcome measures as used in the Bateman and
Fonagy (1999, 2001) RCT study and additional clinical measures related to personality
characteristics. These clinical outcome scores and additional personality related scores will be
related with psychoanalytic model specific measures about the process. Measuring the
psychoanalytic process is not evident because, although it is seen as the heart of every
psychoanalytic therapy, it is in itself imperceptible and a highly conceptual hypothesis.
Therefore it seems logical to study the psychoanalytic process from different perspectives and
with different measures to see whether coherent patterns of change are revealed pointing at
the existence of such a process.
In the following studies we examined the intrapsychic changes from three different
perspectives, from the perspective of independent researchers, from the perspective of the
204

205
patients themselves and from the perspective of the therapists and this with very different
measures: theory driven, empirical, based on transcripts, on patient self-reports and on
therapist reports. Then we studied the relation of the process with changes in clinical
symptom and personality related measures.
Study 1 The Relation Between Process and Outcome, with the Process Seen from the
Perspective of Independent Researchers.
According to J. Clarkin (personal communication), the best way to approach the
psychoanalytical process is by a model. We described a three-dimensional model of the
process in personality disorders in Chapter 1. We operationalized the assessment of the
intrapsychic changes according to the three dimensions of this model (Chapter 3), that consist
of increasing felt safety (measured with the Felt Safety Scale, FSS); of developing more
mature and differentiated representations of self and others (measured with the Differentiation
Relatedness Scale, DRS); and of enhancing mentalization (measured with two scales: the
Reflective Functioning Scale, RFS and the Bion Grid Scale, BGS). In this study we examined
the changes over time of these four measures of intrapsychic changes, and we related these
intrapsychic changes with symptom and personality changes during the therapy. Therefore we
first reduced the outcome changes to two component scores: a Global Symptom Score (GSS)
and a Global Personality Score (GPS).
205

206
Method
Participants
The study group comprised patients who were consecutively admitted at the
psychoanalytically informed hospitalization for personality disorders. Of this group we
selected the patients who stayed long enough in the psychoanalytically informed treatment to
have at least the measures at 0-3-6-9 months and three months after discharge. This to have
enough moments to enable a process study. Forty seven consecutively admitted patients
corresponded to these conditions. Of these 47 patients, three patients were not included in the
statistical analysis because of missing data at discharge or follow-up: one refused further
assessments, the self-reports of another got lost by mail delivery service, one patient was
acutely discharged because of selling drugs.
Thus the patients in the study consisted of 44 admitted patients (13 males, 31 females).
31 were in-patients, 13 in a day treatment setting. The patients were between the ages of 17
and 58 (M = 28.0, SD = 9.24). 17 were single, 17 lived with their parents, 7 were married or
living together with a partner and 3 were divorced. Sixteen patients completed secondary
education and 28 higher education or university. Using the SCID-II we assigned all patients to
one of the three clusters of personality disorders according to DSM-IV; cluster B: 32 (23
borderline, 7 narcissistic, 1 narcissistic and borderline, 1 histrionic and borderline); cluster C:
7 (1 dependent, 1 avoidant and dependent, 5 not otherwise specified); clusters B+C: 5. On
DSM-IV Axis I, we diagnosed 20 mood disorders, 2 anxiety disorders, 1 eating disorder, 5
adjustment disorders and 5 substance related disorders.
206

207
Measures
Clinical symptom measures.
The Self Harm Inventory (SHI) (Sansone et al., 1998) about the last two weeks, the
Symptom Checklist 90 (SCL-90) (Derogatis,1977), translated by Arrindell & Ettema (1993),
the Spielberger State-Trait Anxiety Inventory translated by van der Ploeg et al. (1980), the
Spielberger State-Trait Anger Inventory translated by Van der Ploeg et al. (1982), the Beck
Depression Inventory translated by Bouman et al. (1985).
Personality characteristics.
The questionnaire of the Structured Clinical Interview for DSM-III-R Axis II disorders
SCID-II (Spitzer & Williams, 1985) has been translated in Dutch and adapted for DSM-IV,
by Weertman, Arntz & Kerkhofs (1999). The Inventory of Personality Organization
(Lenzenweger, Kernberg, Clarkin, Foelsch, 2001; Normandin et al., 2002) is a self-report
questionnaire used to assess a structural diagnosis according to the model of Kernberg on
borderline personality organization. The reliability and validity of the IPO translated in Dutch
was demonstrated by Vermote, Maes, Vertommen, Corveleyn, Verhaest & Peuskens (2004)
and Vermote, Smits, Claes & Vertommen (2005). The Inventory of Interpersonal Problems,
circumflex version (Alden et al., 1990) measures interpersonal functioning. All these
measures have a well established reliability and validity.
207

208
Intrapsychic process measures.
Intrapsychic Process Measures were measured with the Object Relation Inventory (Blatt
et al., 1996), a semi- structured interview with open questions about parents, self, therapist
and significant other.
The Felt Safety Scale (FSS) is a five point scale to rate on the ORI ranging from 1.No
felt safety in therapy to 5. Strong feeling of safety (Vermote, Vertommen,Verhaest, Franssen,
Geenen, Corveleyn, Peuskens, 2004a; Vermote, Vertommen, Corveleyn, Verhaest,2004b;
Chapter 3).
The Differentiation Relatedness Scale (DRS, Blatt et al., 1996) is a ten point scale which is
rated on the ORI, measuring the degree of self-other differentiation and relation ranging from
1-2:. Lack of basic differentiation between self-other to 10:. Reflectively constructed
integrated representations of self and others in reciprocal and mutual relationships.
The Reflective Functioning Scale (RFS) is a 11 point scale about the capacity to
perceive ones own and others' actions and feelings in terms of mental states (Fonagy et al,
1998) ranging from 1: Lacking RF to 7:. Marked RF. The Reflective Functioning Scale is
used in the Adult Attachment Interview (Main et al., 1991) where it measures the capacity for
reflective functioning (Fonagy et al., 1998), while when used on the ORI, it measures more
the state reflective functioning (Vermote et al., 2004a, b; Chapter 3).
The Bion Grid Scale (BGS) is a 6 point scale in which answers on the ORI can be put in
6 categories according to their level of mentalization, openness to explore and degree of
switching between levels. The scale reflects the degree of creative exploration and mental
transformation of experiences (Vermote et al., 2004a, b; Chapter 3).
These FSS, DRS, RFS, BGS showed a sufficient inter judge reliability and validity
when rated on the ORI.
208

209
Procedure
After written informed consent, clinical symptom measures were assessed on admission,
each three months, at discharge and three months and a year after discharge. Personality
characteristics were assessed on admission, at discharge and three months and a year after
discharge. The Object Relation Inventories were assessed at the same moments of assessment
of the Clinical Symptom Measures by a psychologist having no contact with the therapeutic
staff and the patients besides of these interviews. The transcripts of the anonimised ORI’s
were distributed in a random order to raters blind to any information about the patients and
having had no contact with them. The ORI – interviews were rated on the Felt Safety Scale,
the Differentiation Relatedness Scale, the Reflective Functioning Scale and the Bion Grid
Scale by a group of 12 raters, three raters for each scale, who had a previous training with the
manual of their scale and a satisfactory reliability test for this scale.
Analysis
To study the relation between the clinical symptom measures and the psychoanalytic
process measures, multiple clinical symptom measures were reduced to a single symptom
component score, the Global Symptom Score (GSS) by means of a Principal Component
Analysis (PCA) on the seven symptom-scores of the patients at the first measurement of a
group of seventy patients of which the study group is a part (see Chapter 6). The PCA yielded
a first component accounting for 50 % of the total variance. The loadings on this component
were the following:.53 for the Self Harm Inventory (SHI 2w), .86 for the Symptom Check
209

210
List, positive symptom total, .85, for the Beck Depression Inventory (BDI), .73 for the
Spielberger State, .89 for the Trait Anxiety Inventory Scores (STAI), .44 for the Spielberger
State and .49 for the Trait Anger Inventory Scores (STAXI). These loadings were used to
calculate GSS component scores at each of the seven time moments.
A similar PCA was performed on the total scores of the different personality
characteristics (IPO, SCID, IIP) resulting in a component, the Global Personality Score (GPS)
explaining 74.6 % of the variance. The loadings on this component, were .88 for the IPO, .87
for the SCID and .81 for the IIP. Also in this case, the loadings of the scales at the first time
moment of a larger group of 70 patients were used to calculate component scores at each of
the three time moments.
Changes over time of the clinical symptom measures, the personality related measures
and the psychoanalytic process measures are studied with analysis of variance, repeated
measures analysis. Then separate random coefficient models with patient-specific intercepts
and time moments as a fixed predictor variable were estimated to test for different trends
(linear, quadratic, cubic trends) between time moments and the clinical symptom measures,
the personality related measures and the psychoanalytic process measures.
Next, a model with patient specific intercepts and fixed slopes
for the time moments for the Global Symptom Score was estimated to predict the intrapsychic
process variables (FS, DRS, RF, BGS assessed on ORI). Similar analyses were performed to
predict these intrapsychic process variables on the basis of the Global Personality Score.
210

211
Results
Outcome
The Global Symptom Score (GSS) showed a significant change over time and
improvement (F (6, 246)=14.25, p < .0001). The mean scores at several time moments are
shown in table 1. The trends are shown in table 2. The changes over time are plotted in Figure
1.
There was a significant change over time and improvement during therapy and the year
after discharge for the Global Personality Score (F(3,128)=10.48, p.< .0001). The mean
scores are presented in Table 1. The trend is linear (Table 2). The changes over time are
plotted in Figure 2.
Process
The change over time of the process measures is shown in table 1. There is a significant
change over time and an improvement of the means of the global Felt Safety score
(F(6,242)=17.06,p = < .0001). The same holds true for the Differentiation Relatedness Scores
(F(6,244)=7.49, p = < .0001). The change over time and improvement of the Reflective
Functioning Scores (RF state on the ORI) shows a tendency to be significant
(F(6,255)=1.97,p = .07). The Bion Grid scores show a significant change and improvement
(F(6,245)= 2.82, p = .01). The trends of the change of the mean global scores of the four
ORI-measures is shown in table 2. The changes over time of the four ORI - measures are
plotted in figures 3 - 6.
211

212
Table 1.
Means and SD of the Felt Safety Scale (FSS), the Differentiation Relatedness Scale (DRS), the
Reflective Function Scale State(RFS), the Bion Grid Scale (BGS) and the Global Symptom
Score (GSS) at Different Moments in Time. 12 M Corresponds with Discharge
FSS DRS RFS BGS GSS GPS
Time M SD M SD M SD M SD M SD M SD
admission 2.32 0.67 4.84 1.29 4.34 1.22 3.95 1.01 0.01 0.87 -0.08 0.87
3M 2.81 0.92 5.43 1.28 3.91 1.29 3.93 0.95 -0.10 0.91
6M 2.62 1.03 5.28 1.20 3.79 1.28 3.60 1.18 -0.27 0.93
9M 3.36 0.99 5.44 1.11 4.42 1.18 4.14 1.15 -0.35 0.98
12M 3.56 1.03 5.95 1.18 4.14 1.34 3.95 1.03 -0.78 0.97 -0.43 0.93
15M 3.70 0.95 5.91 1.31 4.20 1.34 4.16 1.20 -0.87 1.23 -0.66 1.64
24M 3.54 1.02 5.64 1.06 4.40 1.39 4.35 1.27 -0.99 1.04 -0.83 1.08
212

213
Figure 1. Change over 7 moments of the Global Syptom Score. The cubic trend is presented
by the dotted line.
Figure 2. Change of the Global Personality Score over 4 moments. The linear trend is
presented by the dotted line.
213

214
Table 2
Trends of Change Over Time for the ORI- measures and the Global Symptom Score
(df=1,249)and for the Global Personality Score (df=1,128)
Subscale Trend df F p
Felt Safety Scale Linear 1,245 70.86 < .0001
Quadratic 1,245 17.59 < .0001
Cubic 1,245 3.34 .07
Differentiation Relatedness Scale Linear 1,247 24.14 < .0001
Quadratic 1,247 11.54 < .001
Cubic 1,247 .65 .04
Reflective Functioning Scale Linear 1,248 1.64 .20
Quadratic 1,248 1.18 .28
Cubic 1,248 2.86 .09
Bion Grid Scale Linear 1,248 8.26 .004
Quadratic 1,248 1.39 .24
Cubic 1,248 2.00 .16
Global Symptom Score Linear 1,249 73.80 < .0001
Quadratic 1,249 4.47 .04
Cubic 1,249 5.47 .02
Global Personality Score (df= 1,128) Linear 1,128 29.57 < .0001
Quadratic 1,128 .63 .43
Cubic 1,128 1.26 .26
214

215
Figure 3. Mean global scores of the Felt Safety Scale over 7 moments. The quadratic trend is
presented by a dotted line.
Figure 4. Mean global scores of the Differentiation Relatedness Scale over 7 moments. The
quadratic trend is presented by a dotted line.
215

216
Figure 5. Mean global scores of the Reflective Functioning Scale over 7 moments. The cubic
trend is presented by a dotted line.
Figure 6. Mean global scores of the Bion Grid Scale over 7 moments. The linear trend is
presented by a dotted line.
216

217
Process-outcome
To test the relation between the four ORI-measures and the GSS, the significance of the
fixed slope was tested. Changes of the global symptom score over 6 time moments are
strongly related with changes in Felt Safety: F(1, 206) = 27.24, p < .0001 and in
Differentiation Relatedness: F(1, 208) = 17.86, p < .0001. We found no significant
relationship with the Reflective Functioning Scale measured on the ORI : F(1, 205) = 2.30, p
= .13 nor with the Bion Grid Scale: F(1, 208) = .20, p = .65.
To test the relation between the four ORI-measures and the GPS, the significance of
the fixed slope was tested. Changes of the global personality outcome score over 3 time
moments are strongly related to changes in Felt Safety: F(1, 86) = 16.86, p < .0001 and
significantly related to changes in Differentiation Relatedness: F(1, 86) = 6.31, p = .01 and
changes in Reflective Functioning: F(1, 86) = 4.16, p = .04. They are not related with changes
in the Bion Grid Scale: F(1, 86) = .46, p = .49.
Discussion
The Global Symptom Score shows a significant improvement with a cubic trend. The
major improvement is during the last phase of the therapy, which suggests that one should be
careful in discharging patients too early. The Global Personality Score shows a linear
improvement during therapy, also with a continuation after discharge. These phenomena point
at a structural change. However, half of the patients recieved ambulatory psychotherapy after
discharge.
217

218
The ORI-process measures improved significantly except for the RFS of which the
change over time tends to be significant. The major improvement of all measures takes place
in the second half of the treatment. In the first half of the treatment, the process measures
show an increase in FSS going with a decrease in RFS. This is peculiar and seems to point at
the phenomenon of regression, people letting their usual defensive attitudes go within a safe
psychoanalytic frame.
Our major goal was to examine whether and how these outcome changes are related to
the process variables of a psychoanalytically informed treatment: felt safety, mental
representations of self-object relations and mentalization. The relation of changes in the
Global Symptom Score and in the Global Personality Score with the change of Felt Safety is
strongly significant. It seems that increasing felt safety is an important condition to generate
changes in symptoms and in more stable personality patterns in personality disorders during a
psychoanalytic informed hospitalization. Felt safety is a basic dimension in most treatments of
personality disorders and can therefore be considered as a non specific factor. More
particularly in the psychodynamic model, felt safety is a central concept about the role of the
environment in the etiology and treatment of personality disorders. It is most related to the
attachment theory (Bowlby, 1969; Sroufe, 1996) but many concepts are related to it, such as
the background experience of safety (Sandler, 1960), holding (Winnicott, 1960), containment
(Bion, 1959). The experiencing of it and the possibility to regain it in moments of distress and
emotional storms is a primary goal of the treatment (see Chapter 1). A creation and
enhancement of felt safety provides the basis for a change in relational patterns such as less
splitting and a greater individuality (Stern, 1985). According to Bion (1962) it is the base of a
psychic processing of affects and according to Fonagy et al. (2002) it is the base of the
reflective functioning in self-other relationships.
218

219
We found a significant relation as well between changes in the Global Symptom Score
and the Global Personality Score and changes on the Differentiation Relatedness Scale. The
DRS is a theory driven scale about the representations of self - object relations, which are
since Kernberg’s approach the central focus in the treatment of personality disordered
patients. The relation of the DRS with symptom change supports the assumption (Blatt &
Auerbach, 2001; Kernberg, 1984) that a decrease in splitting and a more stable and coherent
mental representation of self and others go with a decrease in anxiety and mood lability. The
psychodynamic conceptualisation of personality organisation is closely related to the
construct of self-other representations, hence the relation between the change on the DRS and
changes in the Global Personality Score is expected.
We found no significant relation between changes in the Global Symptom Score and
changes in RF as measured on the ORI, while the relation with the Global Personality Score
was significant. The ORI-RF reflects a state RF rather than the capacity for RF as with AAI-
RF which can be seen as a trait (see Chapter 4). The relation between change in the ORI-RF
and symptom change is probably complex. When anxious for instance, narcissistic patients
are expected to show an increase in their RF to get control, while borderline patients will
show a disintegration of their state RF. The significant relation between the changes in state
RF and the change in the Global Personality Score, while this relation is lacking for the
Global Symptom Score, may be explained by the fact that RF is more about the way a patient
copes with a symptom which is reflected in personality related measures, than with the
manifestation of the symptom itself.
Changes in the BGS show no significant relation with changes in the Global Symptom
Score, nor with changes in the Global Personality Score. A reason may be that the degree of
creatively and associatively transforming experiences is not linked with the level of
personality organisation and with symptoms, for instance severe borderline patients may do
219

220
better for instance than obsessive-compulsive patients (see Chapter 3). It seems difficult to
judge this mentalization score on its own, we may hypothesize that changes in this form of
mentalization may resort in a therapeutic effect for a group of restricted patients while having
adverse effects in patients who need a stronger ego-control like patients with a low level
personality organisation (Fertuck et al., 2004; Blatt, 1992).
Study 2 The relation between process and outcome, with the process seen from the
perspective of the patients.
To study the therapeutic process of personality disordered patients and its relation with
changes in clinical symptom and in personality related measures from the perspective of the
patients, we used the Leuven Psychotherapy Scale (LPS, Chapter 4). This is a self-report
scale, with items based on patient statements and therefore a non theory driven scale. In other
words the scale reflects the experience of inner change in the words of the patients
themselves.
In this study we examined the changes over time of the LPS and we related these
changes with symptom and personality changes during the therapy. The outcome data were
first reduced to two component scores, a Global Symptom Score (GSS) and a Global
Personality Score (GPS).
Method
Participants
The study group is the same as in study 1.
220

221
Measures
For the clinical symptom change and personality related outcome change, we used the
same measures and the reduction of these data in a Global Symptom Score (GSS) and a
Global Personality Score (GPS), as described in study 1.
The construction of the Leuven Psychotherapy Scale (LPS), its reliability and congruent
and divergent validities are discussed in chapter 4. The LPS self-report scale is composed of
112 items that are ordered, following factor analysis, into 7 subscales: S1, 'General positive
well-being' ; S2, 'Openness to authentic relationships'; S3, 'Capacity to deal with unexpected
events'; S4, 'Capacity to deal with negative emotions'; S5, 'Capacity for interiorisation'; S6,
'Positive experience of the setting'; S7, 'Control over substance use'.
Procedure
After obtaining written informed consent, the LPS and clinical symptom measures were
assessed on admission, every three months, at discharge, three months after discharge and one
year after discharge. Personality characteristics were assessed on admission, at discharge,
three months after discharge and one year after discharge.
221

222
Data Analysis
Changes over time of the LPS and of the clinical symptom measures and the personality
related measures are studied with analysis of variance, repeated measures analysis.
Then separate random coefficient models with patient-specific intercepts and time
moments as a fixed predictor variable were estimated to test for different trends (linear,
quadratic, cubic trends) between time moments and the LPS scales.
To examine whether changes in clinical outcome were related to the intrapsychic
changes we used the results of the data reduction of Study 1, by Principal Component
Analysis of the clinical symptom measures (resulting in the Global Symptom Score -GSS)
and of the personality related measures (Global Personality Score - GPS).
Then, for each of the scales of the LPS, a model with patient-specific intercepts and
fixed slopes for 7 time moments of the Global Symptom Score were estimated to predict the
LPS Scales. Similar analyses were performed to predict LPS scales on the basis of the Global
Personality Score for 4 time moments.
Results
Process
We found significant improvement over time for six of the seven scales of the LPS. For S1 '
Positive well-being' this was F (6, 246)=21.25, p = < .0001, for S2 'Openness to authentic
relationships' F (6, 246)=10.19, p = < .0001, for S3 'Capacity to deal with unexpected and
stressful events' F (6, 246)=2.97, p = .008, for S4 'Capacity to deal with negative emotions'
222

223
F (6, 246)=16.67, p = < .0001, for S5 'Capacity for interiorization' F (6, 246)=16.41, p = <
.0001, for S6 'Positive experience of the setting' F (6, 246)=5.99, p = < .0001, for S7
'Control over substance use' F (6, 246)=1.71, p = .11. The mean scores for each subscale,
scored every three months are plotted in figures 7-13. The trends of this change over time
are given in table 3 and indicated in Figures 7 through 13 by dotted lines.
Figure 7. Mean scores of the ' Positive Well Being' Scale of the LPS over 7 moments. The
cubic trend is presented by a dotted line
223

224
Table 3
Trends of Change Over Time for the LPS Scales (df=1,246)
Subscale Trend F p
Positive Well Being Linear 121.66 < .0001
Quadratic .18 .67
Cubic 4.16 .04
Openness to Authentic Relationships Linear 52.81 < .0001
Quadratic 1.92 .17
Cubic 5.21 .02
Capacity to deal with unexpected and stressful events Linear 9.02 .003
Quadratic 3.07 .08
Cubic 4.52 .03
Capacity to deal with negative emotions Linear 88.22 < .0001
Quadratic 6.91 .009
Cubic 4.26 .04
Capacity for interiorization Linear 89.32 < .0001
Quadratic .02 .90
Cubic 5.80 .02
Positive experience of the setting Linear 25.62 < .0001
Quadratic .93 .33
Cubic 2.91 .09
Control of substance use Linear 6.48 .01
Quadratic .05 .82
Cubic 1.90 .17
224

225
Figure 8. Mean scores of the ' Capacity to Authentic Relationships' Scale of the LPS over 7
moments. The cubic trend is presented by a dotted line.
Figure 9. Mean scores of the ' Capacity to deal with unexpected and stressful events' Scale of
the LPS over 7 moments. The cubic trend is presented by a dotted line.
225

226
Figure 10. Mean scores of the ' Capacity to deal with negative emotions' Scale of the LPS
over 7 moments. The cubic trend is presented by a dotted line.
Figure 11. Mean scores of the ' Capacity for interiorization' Scale of the LPS over 7 moments.
The cubic trend is presented by a dotted line.
226

227
Figure 12. Mean scores of the 'Positive experience of the setting' Scale of the LPS over 7
moments. The cubic trend is presented by a dotted line.
Figure 13. Mean scores of the ' Control of substance use' Scale of the LPS over 7 moments.
The linear trend is presented by a dotted line.
227

228
Process Outcome
To test the relation between the LPS scales and the GSS, the significance of the fixed
slope was tested. Changes of the GSS (Global Symptom Score) were related to changes in
each of the LPS scales. This relation was very strong for the S 1 ‘Positive well being’: F(1,
250) = 736.03, p < .0001. The scales S2 ‘Openness to authentic relationships’: F(1,250) =
178.78, p < .0001) S3 ‘Capacity to deal with negative emotions’: F(1, 250) = 201.11, p <
.0001, S5 ‘Capacity for interiorization’: F(1, 250) = 130.73, p < .0001, S3 ‘Capacity to deal
with unexpected and stressful events’: F(1, 250) = 88.94, p < .0001, S7 'Substance control':
F(1, 250) = 43.06, p < .0001, and S6 'Positive experience of the setting': F(1, 250) = 43.06, p
< .0001 were significant as well.
To test the relation between the LPS scales and the GPS, the significance of the fixed
slope was tested. Changes of the GPS (Global Personality Score) were significantly related to
changes in each of the scales. This relation was again very strong for the S1 ' Positive well
being': F(1,129) = 135.30, p < .0001. The relation with S 4 ‘Capacity to deal with negative
emotions’: F(1,129) = 98.94, p < .0001, S5 ‘Capacity for interiorization’: F(1,129) = 63.35, p
< .0001), S2 ‘Openness to authentic relationships’: F(1,129) = 56.48, p < .0001, S3
‘Capacity to deal with unexpected and stressful events’: F(1,129) = 37.52, p < .0001, S7
'Substance control': F(1,129) = 1.91, p < .0001 and S6 'Positive experience of the setting':
F(1,129) = 8.08, p < .0001 were all significant as well.
228

229
Discussion
All scales of the LPS show a significant positive improvement except for the 'Control of
substance use' scale. In the six scales showing an improvement, the major improvement is
after six months of therapy and goes on till three months after discharge, to remain at a high
level a year after discharge. We see also a major improvement after six months in our own
outcome results (see Chapter 5) and in the process changes as measured by independent
researchers (Study 1 of this Chapter). In the findings of Bateman and Fonagy’s (1999, 2001)
RCT study of borderline patients, most outcome differences in clinical symptoms became
manifest after six to nine months in a comparison to a treatment control group.
Looking at the process in more detail, three scales of the LPS start with a small
improvement in the first six months, with the major improvement starting afterwards: S1
'Positive well being', S2 'Capacity to engage in authentic relationships' and S5 ' Capacity for
interiorization'. The concepts which were revealed in these three scales seem to be related
with the three dimensions of our treatment model (see Chapter 1): felt safety, mental
representations of objectrelations and mentalization. In our study of the validity of the LPS
(see Chapter 4), we found in fact a significant correlation of LPS-S1 'Positive well being' and
the ORI-Felt Safety Scale. We suppose that this inner experience of 'Positive Well Being' is
enhanced by offering a secure, coherent environment and cohesive program to the patients.
The many moments of distress during the treatment going with the possibility to rely on
fellow-patients, therapists and nurses and the experience of finding new ways to deal with
these states of mind, undoubtedly create such a basic feeling to which one can fall back in
later moments of emotional difficulty. S2 about the 'Capacity to engage in authentic
relationships', with intimacy and emotional vulnerability, is connected to changing mental
229

230
representations of self and others, the second dimension of our treatment model. The S3
‘Capacity for interiorization’ (‘Verinnerliching’) subscale is about a self-reflective attitude
accompanied by not being afraid to try new experiences, and is close to the concept of
mentalization which is the third dimension focused on during the treatment.
It is striking that the mean scores of three other scales show a marked decrease in the
first six months before the improvement starts. These scales are S3 'Capacity to deal with
unexpected and stressful events', S4 ' Capacity to deal with negative emotions' and S6 '
Positive experience of the setting'. S3 and S4 seem to show that people have to release their
usual maladaptive coping in response to stress and unexpected events, and this results at first
in a decrease in their capacity to cope. This lowering of defences is known in psychoanalytic
theory as regression. This regression is typical in a psychoanalytically informed treatment,
where it used to be called a regression in function of the ego (Meisnner, 1988). People have to
release their usual maladaptive coping in response to stress and unexpected events, and this
results at first in a decrease in their capacity to cope. It is in accordance with psychoanalytic
theory that this can happen at a moment when there is enough safety. S3, S4 are at their
deepest point, while S1 ' Positive well being' is increasing sharply. The same regression is
manifest in the change over time of the S6 scores ‘Positive experience of the setting'. Clinical
experience shows that only when patients feel safe enough they can start criticizing and de-
idealizing the setting and experience that nevertheless their inner safety can remain. Figures 7
and 12 show that after a ‘honeymoon‘ feeling of being contained within the setting, the '
Positive experience of the setting' (S6) score drops to its lowest point at 6 months, while the
'Positive feeling of well being' (S1) is increasing at that moment. After this decrease at six
months, the S6 'Positive experience of the setting' increases again sharply.
For the scale 'Control over substance use’, there is no significant improvement in this
two year study. There is an improvement in control during therapy and in the first months
230

231
after treatment. This decrease after discharge, as there are at once more occasions to use than
during the hospitalization period.
Given this image of the process, with its marked positive change, we wondered about
the relation of this inner change with outcome. The change, manifest in the scores of the LPS,
relates for all scales significantly with outcome. This for outcome as measured with symptom
scales, as with scales related to personality characteristics. The relation is stronger than we
expected, and stronger than with the measures assessed by independent researchers (study 1)
and the measures assessed by the therapists themselves (study 3).
This is remarkable because the LPS is an atheoretical, purely empirically constructed
scale, drawn from patient statements while some of the personality related measures, such as
the IPO and the IIP, are highly theory driven. It may be that the factors on which the LPS is
constructed touch the mutative factors in the treatment, and in this respect we found it
stimulating that the three dimensions of the proposed model revealed themselves in the LPS.
For a part the strong correlation can be explained, at least for the clinical symptom scales by
the fact that these scales and the LPS are both self-report scales, while the method of
assessing the process in Study 1 and 3 differs from the method of assessing the outcome by
self-report scales. A major reason may be that the LPS is constructed for personality
disordered patients in hospital treatment, which is not the case for the process measures of
study 1 and 3.
Study 3 The relation between process and outcome, with the outcome seen from the
perspective of the therapists.
Having studied the therapeutic process of personality disordered patients in a
psychoanalytically informed hospitalization-based treatment, from the perspective of
231

232
independent researchers with FSS, DRS, RFS and BGS (Study 1) and from the perspective of
the patients themselves with the LPS (Study 2), we wanted to examine whether the therapists
saw the same dimensions of change at similar moments as patients and external researchers.
Reliable measures to study the psychoanalytic process as a whole from a therapist's
point of view and which are fit to assess on a larger number of patients are rare, the
Psychoanalytic Process Rating Scale (PPRS) was the only scale which we came across. It is a
newly developed scale, which we discussed in Chapter 5. The scale focuses on the features of
contact development in a session and is based on theory driven items. For this reason it is of
particular interest to compare the results of the PPRS with those of the LPS, as the latter is
based on statements of the patients.
In this study we examined the changes over time of the PPRS and we related these
changes with symptom and personality changes during the therapy. The outcome data were
first reduced to two component scores, a Global Symptom Score (GSS) and a Global
Personality Score (GPS).
Method
Participants
The participants are the same as in the study 1 and 2.
Measures
The outcome measures are the same as in study 1 and 2. The PPRS is discussed in Chapter 5,
it is a 47 item scale with 5 subscales: Co-operation (close co-operation), Tension (negative
232

233
feelngs), Exploration (explicatation of behaviour,thoughts and feelings), Disillusion (feelings
of being rejected), Contentment (enjoying the treatment). For the clinical symptom change
and personality related outcome change, we used the same measures and the reduction of
these data in a Global Symptom Score (GSS) and a Global Personality Score (GPS), as
described in study 1
Procedure
After obtaining written informed consent, clinical symptom measures were assessed
on admission, every three months and at discharge. As the treatment ended as this moment,
therapists could not score in the post treatment period. Personality characteristics were
assessed on admission and at discharge. The PPRS was assessed around the same moments
as the assessment of the clinical symptom measures. As the psychotherapy was a group
psychoanalysis (1.5 hours, three times a week), we decided to rate each patient on the
sessions of the last two weeks before the moment of assessment. The PPRS was assessed
by the therapist of each patient, there are three senior group therapists who are treating two
groups each.
Data Analysis
Changes over time of the PPRS and of the clinical symptom measures and the
personality related measures are studied with analysis of variance, repeated measures analysis.
Then separate random coefficient models with patient specific intercepts and time
moments as a fixed predictor variable were estimated to test for different trends (linear,
quadratic, cubic trends) between time moments and the PPRS scales.
233

234
To examine whether changes in clinical outcome were related to the intrapsychic
changes we used the results of the data reduction of Study 1, by Principal Component
Analysis of the clinical symptom measures (resulting in the Global Symptom Score -GSS)
and the personality related measures (Global Personality Score - GPS).
Then, for each of the scales of the PPRS, a model with patient specific intercept and
fixed slopes for the 5 time moments of the Global Symptom Score were estimated to predict
the PPRS Scales. Similar analyses were performed to predict PPRS scales on the basis of the
Global Personality Score (3 time moments).
Results
Process
We found significant changes over time for the five subscales of the PPRS. For S1
'Cooperation' F (4, 162)=20.56, p < .0001, for S2 'Tension' F (4, 162)=2.97, p = .02, for S3
'Exploration' F (4, 162)= 18.9, p < .0001, for S4 'Disillusion' F (4, 162)= 5.08, p = .0007, for
S5 'Contentment' F (4, 162)= 14.17, p < .0001.
Significant trends of these changes are given in Table 4. The mean scores of the PPRS scales
and the trends of the change over time of each of the PPRS scales are plotted in Fig 14-18.
234

235
Table 4
Trends in change over time of the five scales of the PPRS (df=1,162)
Trend F p
Cooperation Linear
Quadratic
Cubic
62.06
9.48
6.64
< .0001
.002
.01
Tension Linear
Quadratic
Cubic
.02
9.12
2.32
.88
.003
.13
Exploration Linear
Quadratic
Cubic
36.91
26.94
6.43
< .0001
< .0001
.01
Disillusion Linear 13.23 .0004
Quadratic
Cubic
3.9
1.57
.05
.21
Contentment Linear
Quadratic
Cubic
53.67
.06
1.84
< .0001
.81
.18
235

236
Figure 14. Mean scores of the 'Cooperation' Scale of the PPRS over 5 moments. The cubic
trend is presented by a dotted line.
Figure 15. Mean scores of the 'Tension' Scale of the PPRS over 5 moments. The cubic trend
is presented by a dotted line. The quadratic trend is presented by a dotted line.
236

237
Figure 16. Mean scores of the 'Exploration' Scale of the PPRS over 5 moments. The cubic
trend is presented by a dotted line.
Figure 17. Mean scores of the 'Disillusion' Scale of the PPRS over 5 moments. The quadratic
trend is presented by a dotted line.
237

238
Figure 18. Mean scores of the 'Contentment' Scale of the PPRS over 5 moments. The linear
trend is presented by a dotted line.
Process-outcome
To test the relation between the PPRS scales and the GSS, the significance of the fixed
slope was tested. Changes over time of the Global Symptom Score were related with changes
in the S5 'Contentment' scale: F(1,161)=15.86, p = .0001, in the S1 'Cooperation' scale:
F(1,161)=10.95, p = .0012 and in the S2 'Tension' scale: F(1,161)= 4.48, p = .04. We found
no significant relationship to the S3 'Exploration' scale: F(1,161)= .06, p = .45 nor to the S4
'Disillusionment' scale: F(1,161)=2.35, p = .13.
To test the relation between the PPRS scales and the GPS, the significance of the fixed
slope was tested. Changes in the Global Personality Score were not significantly related to
change in any of the scales of the PPRS.
238

239
Discussion
When studying the process with the PPRS, we saw that all the scales showed a
significant improvement over time, with quadratic and cubic trends, as illustrated in Figures
1-6, except for the S5 'Contentment', which shows a linear improvement.
In our study group of patients with personality disorders in psychoanalytically informed
hospitalization, the therapists rated a marked increase during the first three months for the S1
‘Cooperation’ and S3 ‘Exploration’ scale going with an increase in S2 ‘Tension’. As we
discussed in the previous studies, this complex phenomenon of an increased cooperation and
exploration going with an increased tension, which occurs within the holding environment of
the setting, is known as regression. We find it back from another angle in this study.
It is remarkable that in study 3, where the patients rated their process with the Leuven
Psychotherapy Scale at the same moments in time, they reported positive changes to occur
later than the therapists do. This discrepancy may be explained by the fact that therapists rate
this regression as positive at a moment that the patient only feels that he or she gets worse and
can not see that this phase might be an essential part of the treatment. What therapists evaluate
as positive change in the way patients deal with their feelings and thoughts is not experienced
in the same way by the patients. For instance, when a patient relies less on his usual coping
like denial or splitting, he will feel more distressed at first and he will evaluate this in a
negative way while the therapist will judge this as a progress.
After this initial period of marked increase of cooperation, exploration and tension as
well, we see that cooperation and exploration remain at the same level while tension decreases
and contentment improves. This would be an ideal evolution, with a continuing exploration
during therapy, but which is better supported by the patients. We may wonder whether there is
239

240
not a bit wishful thinking in the rating of the therapists in this regard. Our outcome study
(Chapter 6) suggests that a number of patients are treated in an explorative way, at a moment
that they are probably not at that level yet. We may wonder perhaps, whether the therapists
are not overestimating some of the patients, at least in the first period. As the PPRS in
particular is a mixture of statements about the patient and about the therapist in the session,
the attitude of the therapist is well reflected in this scale. This is for instance the case in the
'Exploration' scale where five of the six items are therapist related. On the contrary in the
'Disillusion' scale five of the seven items are patient related. In this regard, it is striking that
the 'Disillusion' scale scores are high at six months, the same moment as the Leuven
Psychotherapy Scale, S6 'Positive experience of the setting' is at its deepest point, as shown in
Study 3. This is probably because the 'Disillusion ' scale is closer to the experience of the
patients themselves, while the other scales reflect the mixture of the attitude and experience of
therapist and patient.
The results of this process-outcome study show the strongest relation between the
Global Symptom Score and the S5 'Contentment' Scale and the S1 'Cooperation' Scale of the
PPRS, in which half of the items express a feeling of feeling fine in the session such as item
P-07, ‘I feel careful- friendly- full of love’; P-12, ‘I feel calm - satisfied- relaxed’; P-17, ‘I
feel strong- fine- powerful’.
It was remarkable that there was no relationship between Global Symptom Score
change and the PPRS S3 ‘Exploration’ Scale, with exploration being a specific aspect of a
treatment on analytic lines. This finding gives support to our conclusion of the Outcome study
(Chapter 5),that an explorative approach is probably not suited for a number of the patients,
especially not in the beginning of the treatment.
It is unexpected that none of the PPRS scales was related to the change in the Global
Personality Score, in contrast with the relations of most other scales such as the FSS, DRS,
240

241
LPS with the GPS. This is unexpected as the PPRS is an instrument to measure the
psychoanalytic process, which is theoretically more related to personality characteristics than
to volatile symptoms. Too, the PPRS is derived from a theory-driven scale developed for this
psychoanalytic process and personality related scales are for a large part theory driven as well
(such as the IPO). It may also be that the items about the feelings and interactions during
psychotherapy as expressed in the PPRS, are more expressing a current state, while the LPS
for instance contains more items about change in relationship with other people and in dealing
with emotional difficulties, which are more related to personality characteristics. In this
regard, when examining the convergent validity of the PPRS (Chapter 5), we found a poor
correlation of the PPRS with the SCID II and with the IPO and this may be a reason.
In general we may conclude that the PPRS offered another perspective on the same
process as described in Study 1 and 2, in which we see the phenomenon of regression,
revealed from another angle and interpreted differently than by the patients. Furthermore, the
results obtained with the PPRS revealed that non-specific factors of the psychoanalytic
relation as rated by the therapist, the contentment scale and the cooperation scale, are
significantly related to symptom change.
General Discussion
We discussed that the outcome results of this study in Chapter 6. These outcome results
are reflected in the change over time of the component scores, the global symptom score and
the global personality score.These outcome results of this study corroborate the findings of
the Bateman and Fonagy RCT study; with a comparable treatment we also found an
improvement with a continuation after discharge of clinical symptom measures. Furthermore
we found the same improvement for more stable personality related measures.
241

242
It proved to be possible to measure the psychoanalytic process from three perspectives.
The change over time showed a coherent pattern. In nearly all scales we see a marked
improvement in the second half of the treatment, just like in the symptom scores.
During the first months, we see a coherent pattern of regression which is revealed from
the three perspectives, followed by an improvement of most parameters after 6 months. The
ORI- measures show a decrease of mentalization with an increase of felt safety in the
beginning of the treatment. The same phenomenon is shown in the patients' self-report
approach: there is a decrease in defence mechanisms while the feeling of positive well-being
is slowly increasing. Therapists rate this period as positive, in contrast with the patients who
only feel the distress at this moment.This phenomenon of regression was much debated
(Meissner, 1988) and was often seen as a contra-indication for the hospital treatment of
personality disorders. Within a specialised and structured therapeutic setting it seems possible
to control this regression; without such a setting a regression in psychoanalytic therapy of
patients with a borderline personality organisation may be malignant (Meissner, 1988) as
shown in the DPV outcome study (Leuzinger- Bohleber, 2002).
We hypothesize that this regression is a 'new beginning' (Balint, 1968) for most of the
patients but it may as well be a handicap for others, this will be further examined taking
trajectories of change and client variables into account (Chapter 9). This phenomenon of
regression is shown in other studies of the psychanalytic proces as well (Dahl,1988).Weiss,
1995 found the same parabolic curve when measuring the evolution of insight during
tretatment as we found with the RF state measured on the ORI, which he also interprets as a
regression being an essential part of the treatment process. Verheul (2004) points at a
destabilisation as a necessary condition for change. Anyhow, the evolution of the various
process and outcome scores show that such a regression can be controlled within the setting.
242

243
This specific dose response relation with a stagnation and regression in the beginning,
suggests that a treatment which is too short (3-6 months) might cause an adverse effect, while
the major gain of the treatment is clearly after 6 months.
The most important finding of this study is that symptom and personality related
changes are related with the changes targeted at with the treatment: the enhancement of an
inner experience of safety by offering a structured environment and a reflective attitude of the
staff members (Fonagy et al., 2002) and the increase of the integration of self-other mental
representations by working with transferences in the here and now and by offering new
experiences both show a strong relation with symptom and personality changes. The relation
between change in outcome measures and mentalization is probably more complex and not
similar for all kinds of patients.
The strong relation of outcome measures with 'Felt Safety' from the researcher's point of
view, with the ' Positive Well-Being' from the patient's point of view and with 'Contentment'
from the therapist's perspective, indicate that this is an important dimension in the treatment.
This is in vein with Sandell's (2005) recent finding that when comparing patients who had a
successful psychoanalytic treatment with a control group, differences were not found in an
increased self-reflection, or a greater insight, but in the presence of 'inner soothing objects'
wich are helpful in moments of distress.
Finally the relation of psychic change with changes in personality related outcome
measures and the continuation of improvement after treatment of the process-variables are
arguments for a structural change.
243

244
244

245
PART 5
PATTERNS OF PSYCHIC CHANGE
245

246
246

247
Chapter 9
Patterns of Inner Change and Their Relation with Patient
Characteristics and Outcome in a Psychoanalytically Informed
Treatment of Patients with Personality Disorders
A hospitalization premised on analytic lines for personality disordered patients was
shown to be effective in an RCT study conducted by Bateman and Fonagy (1999, 2001). With
a comparable treatment for patients in the borderline spectrum we also found a significant
decrease in clinical symptom measures and in personality-related measures during treatment
and in follow-up. Furthermore these outcome changes were shown to be related to the inner
changes at which the treatment aims. We conceptualised these inner changes to happen along
three dimensions: felt safety, mental representations of self-object relations and mentalization.
These inner changes were measured with four psychoanalytic model-specific measures scored
on Object Relational Interviews (ORI). This way a psychoanalytic process could be revealed
in a group of personality disordered patients in a psychoanalytically informed hospitalization-
based treatment (Chapter 8).
Inner changes during psychotherapy, however, are not homogeneous. In the present
study, we want to examine whether we can delineate patterns of psychic change and whether
these patterns are related to patient variables and clinical outcome. Therefore we cluster the
patients according to their scores on the four ORI- measures of psychic change at six
moments of assessment. A better understanding of the differences in intrapsychic process and
their relation with patient's pre-treatment characteristics and with outcome may help
modulating the treatment to the needs of patients.
247

248
Treatment
The psychoanalytically informed hospitalization based treatment of patients with
personality disorders at the University Center St.-Joseph at Kortenberg, Belgium treats four
groups of 8 patients on a hospitalization-base and two groups of 8 patients in day hospital
(Vermote & Van Sina 1998, Pieters & Vermote, 2002). This treatment is similar to the
treatment provided in the randomised controlled trial of Bateman & Fonagy (1999, 2001)
described in Bateman & Fonagy (2004).
Method
Participants
We defined the study group to forty seven consecutively admitted patients, which
stayed long enough to have at least the measures at 0-3-6-9 and three months after discharge,
this to have enough moments to enable a process study and to have a length of time of
treatment comparable with the Bateman and Fonagy study (99, 01) where patients were
treated for 18 months. Of these 47 patients, three patients were not included in the statistical
analysis because of missing data at discharge or follow-up: one refused further assessments,
the self-reports of another got lost by mail delivery service, one patient was acutely
discharged because of selling drugs.
Thus the patients in the study consisted of 44 admitted patients (13 males, 31 females).
31 were in-patients, 13 in a day-treatment setting. The patients were between the ages of 17
and 58 (M = 28.0, SD = 9.24). 17 were single, 17 lived with their parents, 7 were married or
248

249
living together with a partner, 3 were divorced. Sixteen patients completed secondary
education and 28 higher education or university. Using the SCID-II we assigned all patients to
one of the three clusters of personality disorders according to DSM-IV; cluster B: 32 (23
borderline, 7 narcissistic, 1 narcissistic and borderline, 1 histrionic and borderline); cluster C:
7 (1 dependent, 1 avoidant and dependent, 5 not otherwise specified); clusters B+C: 5. On
DSM-IV Axis I, we diagnosed 20 mood disorders, 2 anxiety disorders, 1 eating disorder, 5
adjustment disorders, 5 substance related disorders.
Measures
Clinical symptom measures.
The Self Harm Inventory (SHI) (Sansone et al., 1998) about the last two weeks, the
Symptom Checklist 90 (SCL-90) (Derogatis,1977), translated by Arrindell & Ettema
(1993), the Spielberger State-Trait Anxiety Inventory translated by van der Ploeg et al.
(1980), the Spielberger State-Trait Anger Inventory translated by van der Ploeg et al.
(1982), the Beck Depression Inventory translated by Bouman et al. (1985).
Personality characteristics.
The questionnaire of the Structured Clinical Interview for DSM-III-R Axis II disorders
SCID-II (Spitzer & Williams,1985) has been translated in Dutch and adapted for DSM-IV, by
Weertman, Arntz & Kerkhofs (2000). The Inventory of Personality Organization
(Lenzenweger, Kernberg, Clarkin, Foelsch, 2001; Normandin et al., 2002) is a self-report
questionnaire used to assess a structural diagnosis according to the model of Kernberg on
249

250
borderline personality organization. The reliability and validity of the IPO translated in Dutch
was demonstrated by Vermote, Maes, Vertommen, Corveleyn, Verhaest & Peuskens (2003)
and Vermote, Smits, Claes & Vertommen (2005). The Inventory of Interpersonal Problems,
circumflex version (Alden et al., 1990) measures interpersonal functioning. All these
measures have a well established reliability and validity.
Intrapsychic process measures
Intrapsychic Process Measures were measured with the Object Relation Inventory (Blatt
et al., 1996), a semi- structured interview with open questions about parents, self, therapist
and significant other.
The Felt Safety Scale (FSS) is a five point scale to rate on the ORI ranging from 1.No
felt safety in therapy to 5. Strong feeling of safety. (Vermote, Vertommen,Verhaest, Franssen,
Geenen, Corveleyn & Peuskens, 2004; Vermote, Vertommen, Corveleyn & Verhaest, 2004;
Chapter 3).
The Differentiation Relatedness Scale (DRS (Blatt et al., 1996) is a ten point scale which is
rated on the ORI, measuring the degree of self-other differentiation and relation ranging from
1-2:. Lack of basic differentiation between self-other to 10:. Reflectively constructed
integrated representations of self and others in reciprocal and mutual relationships.
The Reflective Functioning Scale (RFS) is a 11 point scale about the capacity to
perceive ones own and others' actions in terms of mental states (Fonagy et al, 1998) ranging
from 1: Lacking RF to 7:. Marked RF. The Reflective Functioning Scale is used in the Adult
Attachment Interview (Main et al., 1991) where it measures the capacity for reflective
functioning (Fonagy et al., 1998), while when used on the ORI, it measures more the state
reflective functioning (Vermote, Vertommen, Verhaest & Peuskens, 2004; Chapter 3).
250

251
The Bion Grid Scale (BGS) is a 6 point scale in which answers on the ORI can be put in 6
categories according to their level of mentalization, openness to explore and degree of
switching between levels. The scale reflects the degree of creative exploration and mental
transformation of experiences (Vermote, Vertommen , Verhaest et al., 2004; Vermote,
Vertommen, Corveleyn et al., 2004; Chapter 3).
Procedure
The Object Relation Inventories were conducted by a psychologist blind to all other
patient information and who had no contact with the therapeutic staff. The interviews were
conducted upon admission, every three months during treatment, at discharge and finally
three months after discharge. Anonymized interview transcripts were distributed in random
order to raters blind to all patient information and having had no contact with them. The ORI
interviews were evaluated by a group of 12 trained raters, three raters per scale, for whom
satisfactory reliability was established with each of the following scales: the Felt Safety Scale,
the Bion Grid Scale, the Reflective Functioning Scale, and the Differentiation Relatedness
Scale.
Clinical symptom measures were assessed at the same moments as the ORI. Personality
characteristics were assessed on admission, at discharge, and three months after discharge,
with the exception of the lifetime history of trauma and self-harm which were assessed only at
admission.
251

252
Data Analysis
To examine whether there were different patterns of therapeutic process, a k-means
clustering was used to cluster the 4 measures of therapeutic change measured at six moments.
We preferred a cluster analysis over a trajectory analysis, because we were interested in the
differences in relation between the measures at given moments of the process. A trajectory
analysis shows individual trajects for the four measures. The characteristics (SHI, SCL, STAI,
STAXI, BDI, SCID, IPO) of two clusters of patients were compared with the t-test.
To examine whether changes in clinical outcome were related to the clusters of
intrapsychic process, multiple clinical symptom measures were reduced to a single symptom
component score, the Global Symptom Score (GSS) by means of a Principal Component
Analysis (PCA) on the seven symptom-scores of the patients at the first measurement of a
group of seventy patients of which the study group is a part (see Chapter 6). The PCA yielded
a first component accounting for 50 % of the total variance. The loadings on this component
were the following: .53 for the Self Harm Inventory (SHI 2w), .86 for the Symptom Check
List-psychoneuroticism scale, .85 for the Beck Depression Inventory (BDI), .73 for the
Spielberger State, .89 for the Trait Anxiety Inventory Scores (STAI), .44 for the Spielberger
State and .49 for the Trait Anger Inventory Scores (STAXI). These loadings were used to
calculate GSS component scores at each of the seven time moments.
A similar PCA was performed on the total scores of the different personality
characteristics (IPO, SCID, IIP) resulting in a component, the Global Personality Score (GPS)
explaining 74.6 % of the variance. The loadings on this component, were .88 for the IPO, .87
for the SCID and .81 for the IIP. Also in this case, the loadings of the scales at the first time
moment of a larger group of 70 patients were used to calculate component scores at each of
the three time moments.
252

253
The clinical change over the six moments of assessment (SHI, SCL, BDI, STAI,
STAXI) was compared for the two clusters with Analysis of variance, repeated measures with
one between groups effect.
Results
Patterns of Intrapsychic Change
A two cluster solution was the most parsimonous and meaningful solution, Table 1
shows the non-standardized means of the four ORI scales at the six moments of assessment
for the two clusters Figures 1 and 2 show the plots of the 4-ORI measures of the two clusters
with measures standardized at Assessment 1. In one cluster the pattern over time of the
process scores shows a stable increase (therefore we call it the stable cluster), in the other
cluster the process measures fluctuate (therefore we call it the fluctuating cluster).
The scores of the four scales were each higher at similar times of assessment for the
stable cluster compared to the fluctuating cluster. In both clusters, FS showed an increase over
time with continuation after discharge, but in the fluctuating cluster FS decreased between 2
and 4 months, which is not the case for the stable cluster. The increase in DRS was higher in
the stable cluster compared to the fluctuating cluster. In the stable cluster, the measures of
mentalization, RFS and APP increased steadily (with only a minor decrease in RFS at three
months), while in the fluctuating cluster both measures initially decrease, begin to increase
after 6 months, but fall again at the moment of discharge and regain after discharge.
253

254
Table 1
Non standardized measures of intrapychic change (Cluster cl.1,n=27, cl.2,n=17)
FSS global DRS global RFS global BGS global
Time M SD M SD M SD M SD
Admission
fluctuating Cl 2.2 0.7 4.5 1.2 4 1.3 3.8 1
stable Cl. 2.5 0.6 5.4 1.3 4.8 1 4.2 1
3 months
fluctuating Cl 2.7 1 4.9 1 3.5 1 3.7 0.9
stable Cl 3 0.7 6.2 1.2 4.6 1.4 4.2 0.9
6 months
fluctuating cl 2.3 1 4.8 1.2 3.4 1 3 0.9
stabe cl 3.1 0.9 6 0.8 4.4 1.4 4.5 0.9
9 months
fluctuating cl 3.1 0.9 4.9 0.9 4.1 1.2 3.7 1.2
stable cl 3.7 1 6.2 0.9 4.8 1 4.8 0.9
12 months
fluctuating cl 3.3 1.1 5.4 0.9 3.5 0.9 3.4 0.9
stable cl 3.9 0.7 6.9 1 5.1 1.3 4.8 0.7
Discharge + 3months
fluctuating cl 3.5 1 5.3 0.9 3.6 0.9 3.6 0.9
stable cl 4 0.7 6.9 1.1 5.2 1.3 5.1 1
254

255
Figure 1. Traject of the mean global Felt Safety measure and of the mean global
Differentiation Relation measure for the fluctuating-anaclitic (1) and the stable-introjective
(2) cluster of patients
Figure 2. Traject of the mean global Reflective Functioning Scale and of the mean global
Bion Grid Scale for the fluctuating-anaclitic (1) and the stable-introjective (2) cluster of
patients
255

256
Patient Characteristics Associated with the two Clusters of Intrapsychic Change
The difference between the stable and the fluctuating cluster in severity of borderline
personality organization, as measured on the IPO, was not significant M = 137.4 (SD = 24.82)
versus M= 130.6 (SD = 27.28) with t(42) = .80, p = .43. There was no significant difference in
the degree of past traumatic experience, M = 7.33 (SD= 3.01) versus M = 5.14 (SD = 4.15)
(t(42) = 1.29, p = .20) and no difference in self harm, the SHI being M=7.37 (SD= 4.77) for
the fluctuating cluster and M=7.36 (SD = 3.86) for the stable cluster (t(42) = .01, p = .99).
We found significant differences in clinical symptoms as measured on the SCL-90-R,
with the fluctuating cluster having more hostile symptoms compared to the stable cluster
group M= 9.67 (SD =2.79) versus M=7.07 (SD =1.92) (t(42) = 2.08, p = .04) and the
depressed range M= 50.07 (SD = 12.89) versus M= 42.41 (SD = 15.78) (t(42) = 1.90, p = .06).
We also found a significant difference in the descriptive pattern on the SCID-II, the
fluctuating cluster having more avoidant: M= 4.81 (SD = 1.73) versus M=1.87 (SD = 1.88)
(t(42) = 3.21, p = .002), more borderline: 8.81 (SD = 3.10) versus 7 (SD = 2.97) (t(42) =1.92,
p = .061), more dependent 3.74 (SD = 2.09) versus 2.52 (SD = 2.10) (t(42) =1.87, p = .068)
and more depressive features 5.78 (SD = 1.40) versus 4.53 (SD = 1.84) (t(42) =2.55,, p =
.014) on the SCID-II. The stable cluster shows significantly more narcissistic features than the
fluctuating cluster M= 4.24 (SD = 2.95) versus M= 2.48 (SD = 1.89) (t(42) = .02, p = .02).
Outcome Comparisons for Intrapsychic Change Clusters
The Global Symptom Score shows a significant decrease for the whole group across
the moments of assessment (F (5,198) = 14.90, p < .0001), with a difference in clinical
symptom change between the two clusters (F (1,42)=5.77, p = .02). The fluctuating cluster
256

257
was associated with more symptoms at the start that decreased more quickly during therapy,
reaching the level of the stable group at discharge. The stable group had fewer symptoms at
the start, and those symptoms decreased more slowly, but after discharge this global symptom
score tended to decrease further.
When we looked at the separate clinical symptom scales, each measured at six
moments, we saw a significant difference in outcome for the SCL anger subscale (F(5, 201) =
2.17, p = .058). Fluctuating cluster patients being more angry at the start (SCL ang 27.78 SD
9.43) than the stable cluster (SCL ang 23.59 SD 10.25), but this anger diminished in a way
that the fluctuating cluster reached the same level of anger at moment six as the stable cluster
(st.cl.: 19.93 SD 9.86; fl.cl. 2 19.58 SD 7.35). We found a similar difference for the BDI (F(5,
200) = 2.31, p = .046): fluctuating cluster patients being more depressed at moment 1 (BDI M
= 28.59, SD = 8.70) than stable cluster patients (M = 20.29, SD = 11.17) and showing a faster
decrease in BDI scores but remaining more depressed at moment 6 (BDI M = 15.90, SD =
12.00) than stable cluster patients (BDI M = 11.8, SD = 10.48).
Analysis of changes in personality characteristics, measured with the Global
Personality Score at three moments, showed a significant decrease for the two groups with no
difference between the clusters.
When looking at the separate scales we see that all SCID-II categories score higher for
the cluster 1 group than for the cluster 2 group, except for the narcissistic SCID-II features
which are higher for the cluster 2 group and show a continuous pronounced decrease with a
tendency for a significant difference in change between the two clusters (F(2, 84) = 2.81, p =
.065).
For every other score the greatest change is seen during the therapy phase for the first
group and after discharge for the second group. For the SCID-II Paranoid features (F(2, 84) =
3.47, p = .035), we found the cluster 1 starting at a higher score and showing a continuing
257

258
decrease, while cluster 2 patients start at a lower level, show an increase during therapy but a
large decrease after dismissal. There is a difference in the change of the borderline SCID-II
features (F(2, 84) = 2.77, p = .068): cluster 1 starts at a higher level, there is a decrease during
treatment and an increase after discharge in comparison with the scores at discharge while
cluster 2 shows a large and continuing decrease. This is similar for the SCID-II schizotypal
features (F(2, 84) = 2.82, p = .065), the major decrease also happens after treatment for
cluster 2.
Discussion
The clinical impression that there are different patterns of inner change during a
psychoanalytically-informed hospital treatment for personality disorders is supported by our
findings. A cluster analysis of the scores of the four measures of intrapsychic change at the six
moments of assessment revealed two clusters of patients.
In a first cluster, the pattern of inner change shows a fast increase in felt safety once the
patients are in the setting. After this so-called ‘honeymoon’ phase, their felt safety decreases
sharply, to increase again gradually during further treatment. The scores of the two
mentalization measures show a disintegration associated with this period of decrease in felt
safety. After this period, the mentalization process recuperated slowly. There was, however, a
new decrease in the mentalization parameters at the intensive emotional moment of separation
from the setting, with mentalization increasing again after discharge. The changes in mental
representations of self-other relationships were minimal in this fluctuating cluster.
A second cluster of patients showed a more stable pattern of inner change. In
comparison with the fluctuating cluster of patients, at six months there is only an asymptote of
the growth of Felt Safety and no disintegration of mentalization. After this moment at six
258

259
months, the Felt Safety and the two measures of mentalization (RFS and BGS) steadily
increased. There was also a major increase in the quality of mental representations of self and
other, as measured with the DRS. The positive changes of all measures continued after
discharge.
The critical moment at six months is present in both clusters and seems to correspond
with the regression described in Chapter 8. In general the stable cluster of patients seemed to
profit by the treatment as it was offered, with a post treatement increase of their explorative
and associative capacities, a development of their reflective attitude toward the inner world,
and a substantial change in the inner representations of self and other, while this is not the
case for the fluctuating cluster.
We further found that the two discussed patterns of inner change were related to some of the
patient characteristics. As far as the severity and nature of their personality disorders and
degree of psychic suffering were concerned, we found very few differences between the
clusters: we found that they were as severely disordered regarding borderline organization,
had similar histories of suicidal threat and self harm, had no significant differences in past
traumatic experiences and could not be reliably distinguished on the sum score of the SCID-II
personality scales. However, the two clusters showed a different clinical picture at the outset,
which is of great value in the prediction of the kind of therapeutic process. The fluctuating
group, manifesting substantially greater symptomatology, reported more anger and depressive
symptoms than did the stable group, which showed less symptoms and a better impulse
control. Furthermore, from a descriptive DSM point of view as measured on the SCID-II, we
see that the fluctuating cluster of patients tends to show significantly more avoidant and
borderline features, while the second shows significantly more narcissistic features.
259

260
The distinction of these two groups seems to correspond to the differentiation between
the borderlines as described from a psychodynamic point of view by Kernberg (1975), and the
group of the narcissistic borderlines as described by Kohut (1971).
Our results, based on an analysis of the therapeutic process with theory-specific
measures for the psychoanalytic model of inner change, fit very well with the findings of
Blatt. When reanalyzing the results of the Menninger study (Blatt, 1992), Blatt found that the
results of the treatment were meaningfully related to patient characteristics. He distinguished
a group of so-called anaclitic patients, who were more relation-oriented and benefited from
supportive therapy, from a group of introjective patients, who were more concerned with self
definition and benefited from explorative or insight-focused therapies. In the present study,
we found the fluctuating cluster of patients showing more borderline and avoidant
characteristics, which corresponds to the anaclitic group of Blatt. The stable cluster was
associated with more narcissistic characteristics which is typical for Blatt’s introjective group.
The pattern of the psychoanalytic process which we found in the fluctuating - anaclitic
group corresponds well with the mechanism of change that Blatt hypothesises for this group.
He states that the process of this group is dependent on the therapeutic relationship. In the
process pattern of this group we see that changes in mentalization and diffrentiation
relatedness correspond with changes in felt safety, and drop when this is low as in th
ebeginning of the treatment and at the moment of separation at discharge. Furthermore
changes are not consolidated in the post treatment phase, when the environmental support is
no longer present.
For the introjective patients, Blatt supposes an introjective mechanism of change
through interpretation and insight. In our stable-introjective cluster, we see that changes are
less vulnerable to felt safety and occur late in the treatment and continue after discharge. This
is compatible with the hypothesized change by relection and interiorization..
260

261
Fertuck et al. (2004) found that anaclitic patients benefited more from a structuring
attitude toward their mental processes, while the narcissistic group clearly profited from an
increase in their associative functioning. This corresponds with our results: the anaclitic
cluster showed less improvement on the mentalization dimensions, their Reflective
Functioning being vulnerable to emotional moments such as the separation on discharge. The
introjective cluster clearly showed an improvement on the mentalization parameters.
The outcome for both groups was good, but there were differences. The fluctuating -
anaclitic group with strong borderline features and a more irregular process decreased more
quickly in depressive and anger symptoms but patients from this cluster showed a difficulty in
maintaining these improvements after dismissal. Their major inner psychic changes were in
the domain of Felt Safety and were therefore more relational and support-based. As they had a
poorer change in their representations of self and other and less change in their mentalization,
one might recommend that a continuous follow-up and prolonged ambulatory treatment are
needed. In contrast, the stable- introjective cluster, exhibiting a more controlled process,
showed therapeutic gains continuing to increase after dismissal. We argue that the reason for
this is a structural inner change caused by the explorative side of the treatment to which these
patients seemed to respond much more than the anaclitic group.
In summary, we may conclude that specific patterns of change could be related to types
of patients. We found a difference between an anaclitic group that probably profits more by a
supportive and structuring approach and an introjective group that probably profits more by
an explorative approach. These groups proved not to differ in the severity of the personality
disorder, but in personality style. This may indicate that it is possible from the outset of the
treatment to discern the group that would benefit from more support and structure in the initial
phases of the treatment and on dismissal, from the group that profits from the explorative
treatment as it is provided now.
261

262
Blatt suggests that the anaclitic group needs enough safety and relational support before
they can start to explore in a second stage of the treatment, while the needs of the introjective
group must first be addressed by offering insight and understanding before they can gain form
the relational aspect of the treatment.
These findings based on the analysis of the process, need to be confirmed by studying
the process, starting from two distinctive groups in anaclitic and introjective characteristics. In
this sense it is important to remark that contrary to their expectations Blatt & Shahar, 2005
found that a group with mixed anaclitic and introjective features evidenced significantly
greater gain over the course of long-term intensive, inpatient treatment. They hypothesize that
this group has less consolidated modes of adaptation and is therefore more accessible to
treatment.
262

263
Conclusion
Pointers in Hospitalization-Based Psychoanalytic Therapy of
Personality Disorders
1. Hospitalization-based psychotherapy for Personality Disorders works.
The prevalence of patients with personality disorders (PD) is high. In a large U.S. NIH
epidemiological survey (Grant et al., 2004), 14.8 % of the population meets the standard
DSM-IV criteria for a personality disorder. Borderline personality disorder occurs in 1 to 2%
of the population and 14-20 % of the psychiatric population (Zannarini et al., 2003). This is a
serious condition: borderline personality disordered patients have a 10% chance of suicide,
and a combination of BPD, mood disorder, and alcoholism is associated with a life
expectancy of only 50% (Stone, 1990). Two recent meta-analyses showed the effectiveness of
psychodynamic therapy and cognitive behavior therapy in the treatment of personality
disorders, with large overall effect sizes (Perry et al., 1999 ; Leichsenring & Leibing, 2003).
Both treatments ware of a comparable intensity (Migone, 2003). Psychoanalytically informed
hospitalization-based treatment is an intensive form of psychotherapy for PD, being offered in
specialised centres. The outcome data in Chapter 6 corroborate the findings of the Bateman
and Fonagy study.
2. There is a continuation of the effect and an improvement in the post-treatment phase.
One of the most intriguing findings in the Bateman-Fonagy (1999, 2001) study is that
the improvement continues in the post-treatment phase. We found the same phenomenon on
most measures for the whole group. As seen in the outcome trajectories, the worst group,
defined by clinical presentation, even changes dramatically in this post treatment period
263

264
(Chapter 6). The late and the post-treatment symptom outcome improvement may be due to
an indirect effect of psychic changes that occur during treatment. Studies about symptom-
oriented treatments were inconclusive about post-treatment effects (Linehan, Heard &
Armestrong, 1993;Verheul et al., 2003, Bohus et al., 2004), certainly with regard to
hopelessness and depression.
However, the post-treatment effect can be attributed to the fact that in our sample 50%
continued with ambulatory treatment outside the setting and 75% in the Bateman & Fonagy
(1999) sample. Before the hospitalisation treatment, in contrast, these treatments failed, which
led to the hospitalisation treatment in a majority of cases. The hospitalisation that offered a
sound base for a continuing ambulant psychotherapy seems to be a good result. Based on
these findings, we decided to focus on the process of inner change, while also organising less
intensive post-treatment formulas by which the process can continue. In this regard Chiesa
(2000) and Chiesa & Fonagy (2003) showed better results in a step down programme where
inpatient therapy was followed by day-hospital and ambulatory treatment versus a prolonged
inpatient treatment.
3. Patient drop-out can be low, paying attention to a first phase in the treatment; some
predictors for drop-out were established.
Drop-out rates in the therapy programs for personality disorders are high, ranging
from 32-66 % ( Chiesa, 2000). Our study showed that the drop-out group differed
significantly in vindictiveness and hostility. This corroborates the findings of Smith et al.
(1994). The judgement of the nurses predicted drop-out; this was not the case, however, for
the judgement of the psychotherapists or for the quality of the therapeutic process in the
beginning. These findings help us to better delineate the group that is vulnerable for drop-out
(Chapter 7).
264

265
4. It is possible to measure the treatment process.
The psychonalytically informed treatment focuses on inner psychic change, which is
one of the definitions of the psychoanalytic process. This process is typically imperceptible
and hard to define. There exists however a wealth of theoretical concepts and clinical
knowledge about this process, but it is in the soft clinical tissue of individual therapists
without much empirical bones. One of the reasons is that there are few instruments to measure
the process as a whole in personality disorders. Although highly inferential, the different
process measures that we examined and/or developed showed a sufficient reliability and
validity (Chapters 3, 4, 5).
5. The psychoanalytic process was revealed when measured from different perspectives.
Measuring the psychotherapeutic process from different perspectives in a larger group
of patients revealed a coherent pattern. The elephant becomes obvious for the blind man
(Bucci, 1996). The pattern is peculiar: nearly all scales show a stagnation or decrease in the
first half of the treatment and an increase in the second half, with a continuing improvement
after discharge.
When looking in detail at this phenomenon in the first half of the treatment, we see
that while the patients cooperate with the treatment, they feel worse and show less adaptive
coping. From the independent researcher's point of view as measured with the ORI measures,
we see an increase in 'felt safety' and a decrease in 'state reflective functioning'. The patients'
self-reports show a decrease in the 'capacity to deal with unexpected and stressful events' and
in the 'capacity to deal with negative emotions'. The therapists rate an increased 'tension' with
an increased 'cooperation' and 'exploration' in this first period.
265

266
In psychoanalysis this phenomenon is known as regression (Meissner, 1988 ). It was
often considered to be a contra-indication to hospitalise personality disorders because of the
danger of malignant regression, but we see that in a psychoanalytically informed
hospitalisation-based treatment, this regression may be well managed and is part of the
process. One way of interpreting it is that patients let their old and ineffective ways of coping
go and thus suffer more distress in the beginning. This is described from several theoretical
points of view ( Balint ,1968; Dahl,1988; Weiss, 1995; Verheul, 2005).
6. The three dimensional model of psychic change in PD was not falsified.
As a complement to the classic model of the psychoanalytic process (resistance,
interpretation, working through), we suggested a three dimensional model: felt safety,
mentalization, and object relations. In patients with a neurotic personality there is no failing in
felt safety and mentalization, and the classic process of interpreting object relations in the
transference can take place. In personality disorders, significant development in the three
dimensions must take place (this being what we called the basic psychoanalytic process,
Chapter 1).
When measuring the process in the three interrelated dimensions, we indeed saw
different and meaningful changes taking place during the treatment (Chapter 8). Fonagy &
Target (2005) recently proposed a model of psychic change that shows some resemblance
with the three dimensional model that we studied. They outlined three modes of psychic
change: first, a mode of intersubjective shifts that allows the patient to externalise alien parts
within an atmosphere of human generosity (this corresponds for a part to the background
dimension of felt safety); second, a mode about the revitalisation of mental processes, going
with a regression in the beginning (this corresponds with our dimension of mentalization); and
266

267
third, a mode of representational change (this is similar to the dimension regarding mental
representations of object relations).
7. Measured from three different perspectives (independent researchers, patient self-
report, and therapists), treatment outcome is meaningfully related to the psychoanalytic
process.
Relations between treatment process and outcome were low or non-existent in prior
investigations (e.g., Fonagy et al., 2002; Vaughan & Roose, 1995). We found significant
relations between outcome change and psychic change from each of the three perspectives we
took as part of the process: the independent researcher's point of view (ORI-measures), the
patient's point of view (LPS) and the therapist's point of view ( PPRS).
8. Of all the process variables, the variables related to a background experience of safety
showed the strongest relation with outcome change in PD.
The same factor reveals itself as having the strongest relation with symptom change
from the three perspectives: 'Felt Safety' in ORI, 'Positive well being' in LPS and
'Contentment' in PPRS. This relation is not only for symptom change but for personality
change as well.
This background experience of safety is a subjective feeling about the human
environment. It is not just a vague notion of being loved. In psychoanalytically informed
hospitalisation, it corresponds to a culture on the ward where discipline, care, authenticity,
respect, and non-judgmental attitudes are of high value (Chapter 1). Although mentalization is
perhaps not the primary agent of change, a mentalizing attitude on the part of the caregivers
seems important to guarantee this background of safety. Patients with a low level borderline
organisation attack this frame and the attitude of the staff. We hypothesize that keeping this
267

268
frame and attitude while being confronted with patients being in emotional storms and
enacting violent traumas of the past and aggressive internal objects makes it possible for
patients to introject this experience and permit new internal objects develop. This finding adds
further empirical value to the psychoanalytical ideas of treating so called 'primitive mental
disorders' by the Hungarian School of Ferenczi and Balint, the attachment theory of Bowlby,
the theories of Fonagy and Bion, and the approach of the Sandlers (see Chapter 1).
Recent findings from Sandell (2005) are in the same vein: comparing patients with a control
group after a successful analysis revealed that differences were not in insight or self
reflection, but in having soothing inner objects to turn to in difficult moments.
9. Changes in mental representations of self and others are related to changes in
symptom and personality related outcome.
Blatt & Auerbach (2001) and Fonagy (1999) defined the mental representations of
object relations as implicit, procedural patterns (see Chapter 1). We suppose that felt safety
and the mental processing of interpersonal experiences may lead to alterations in these mental
representations. In psychoanalytically informed hospitalisation treatment, this is enhanced by
experiencing new types of interactions and the interpreting in the here and now of the multiple
transferences which are happening in the group psychotherapies. The Differentiation
Relatedness Scale (DRS) measures the mental representations of object relations. Blatt found
a relation between changes in the DRS and outcome measured with the Global Adjustment
Scale in severe psychopathology (Blatt et al., 1996). Likewise, we find a significant relation
between changes in the DRS and changes in symptoms and in personality related measures.
In line with Blatt & Auerbach (2001) and Fonagy (1999), we may hypothesize that it
might be the changes in these deeply ingrained patterns that provoke long lasting changes. It
is probably not a coincidence that the group in which we see changes in the DRS during
268

269
therapy is also the group which shows a continuing improvement of symptom and personality
related measures in the post-treatment phase. This is not the case in the group without
substantial changes in the DRS.
10. The role of mentalization seems to depend on the phase of the treatment and the
personality type of the patient.
The relation of mentalization with outcome change is not as unequivocal as for the
other two dimensions of the model. We studied two types of mentalization, the reflective
functioning according to Fonagy and the psychic processing according to Bion (see Chapter
2). Changes in reflective functioning are related with changes in personality related measures,
but changes in the Bion type of mentalization are not. The regression in the first half of the
treatment is strongly reflected in the decrease of the 'state' reflective functioning, as measured
on the ORI (Chapter 8). There is a difference in the change over time of the mentalization
parameters according to the personality type of the patient (Chapter 9).
These findings support the hypothesis of Blatt & Shahar (2004) that the importance of
focusing on mentalization in the patient seems to depend on the phase of the therapy and of
the personality type of the patient. In introjective patients, this dimension is very important
from the start and a necessary condition before they can start to experience and change object
relational patterns. In anaclitic patients, a preceding phase of attention to the felt safety with
environmental support and structure seems necessary before they can adapt a mentalizing
attitude.
269

270
11. We found different patterns of the psychoanalytical process, probably pointing to
different mechanisms of change for different types of patients.
We found a rather unstable pattern of change in the three dimensions with an anaclitic
type of patient. This may be explained by the findings of Blatt, that the mechanisms of change
in this kind of patient depends on the support and the structure from the environment. We
found a stable pattern of change in the three dimensions, which continues in the post-
treatment phase to be related with an introjective type of patients. This supports the findings
of Blatt & Shahar (2004) that the mechanism of change in this kind of patients is probably
due to introjective mechanisms of change.
12. The treatment programme can be adapted to the needs of the patients.
At the moment, the psychoanalytically informed hospitalisation-based treatment offers
a uniform treatment for the different types of personality disorders. Differences in what is
offered to the patients depend on the severity of the personality disorder. We found, however,
that the groups with an improving trajectory of outcome and the group with a trajectory of
poor outcome differ not in severity of the personality disorder but in personality style. The
findings of Chapters 6, 8, and 9 show that the group of anaclitic patients have other needs than
does the group of introjective patients. The introjective patients seemingly benefit from the
treatment as it is. This corresponds to Karterud et al. (1992) finding that patients from the
anxious cluster C improved more in a day hospital therapeutic community treatment for
personality disorders than patients with borderline personality disorder.The anaclitic type of
patients probably needs more environmental support in the beginning of the treatment and at
discharge and follow-up. The distinction between the two kind of patients can be made from
the outset. These findings were unexpected and based on comparing the patient variables of
trajectories of outcome and of patterns of psychoanalytic process. The hypothesis needs to be
270

271
confirmed by starting with a pure anaclitic group and an introjective group and studying the
process.
13 The criteria of health economy tend to be based on short term studies with symptom
outcome measures in selected groups of patients. This threatens the most effective
therapeutic agents of change in an effective treatment setting.
It is obvious that the therapeutic effects which we found to be related with outcome
changes, such as the increase of felt safety and the changing of deeply ingrained patterns of
self-other relationships, cannot be changed in three or four weeks. Based on meta reviews,
Perry et al.(1999) and Leichsenring & Leibing (2003) point at the fact that the effects of
psychotherapy in PD are clearly related to the length of treatment. They warn against
interpreting short term RCT based on symptom self-report measures which may reflect short
term honeymoon effects but not long term effects.
In this study we found that there is a probably necessary regression in the first months
of the treatment while there is continuing improvement in the post-treatment phase. This
corroborates the findings of the Bateman & Fonagy (1999, 2001) RCT effectiveness study.
Therefore, it seems illogical to discharge patients early in the treatment or to compare this
form of effective treatment in its early phases with other forms of treatment. For a long time,
one saw PD as a fairly untreatable condition, and it is true that these patients are very difficult
to handle in private practices. The mentioned meta reviews show that there is no longer
reason to hold this defeatist opinion when one treats these patients long enough. Given the
high prevalence of this psychopathology and its psychological and sociofamilial burden, it is
important to avoid creating revolving door patients by keeping unrealistically hopeful or,
contrarily, defeatist expectancies about their treatment. Both are at the base of short term
treatments. Perry et al. (1999) showed that an effective treatment of personality disorders may
271

272
be associated with a sevenfold faster rate of recovery in comparison with the natural history
and that treatments of less than a year may better be characterised as treating crises and
concurrent axis I disorders but not the personality disorder. Gabbard , Lazar & Hornberger
(1997) and Bateman & Fonagy (2003) show the cost-effectiveness of long-term treatments.
14. Doing psychotherapy research is not harmful for therapeutic skills and intuition. A
personal experience.
Psychotherapists and psychoanalysts show an ambivalent attitude towards research. It is a
widespread idea that the generalisations and simplifications, necessary to answer research
questions in a statistically valid way, do not respect the wealth and inifinite varieties of what
happens in a psychotherapy and may even be harmful to the discipline. The caricature being
that of a researcher who does not pay attention to the very indivual feelings of his patient.
Others like Gabbard (2003,p. 837) state 'that it makes no sense to write about therapeutic
action as if somehow the question of what is therapeutic and how best to help our patients is
one that can be settled by logical argument and debate. It is in fact an empirical question ,
which can no more be answered by logic and debate than the question of whether one and
another treatment for heart disease is more effective'. I would not be so radical as Gabbard.
Many of our findings confirm existing clinical experience and theoretical knowledge. But
because of research the psychoanalytically oriented treatment under study is no longer based
on a belief, but grounded and that makes a world of difference. Such a base offers even more
room for open questions and for the creativity of the individual therapists confronted with the
very individual problems of the patients. Furthermore, I experienced that the modes of
functioning as a researcher and as an analyst are not contradictory. I learned that research is a
highly creative and emotional process with moments of despair when nothing seems to fit and
moments of esthetic beauty when after complex analyses something simple and true or new
272

273
reveals itself. This way, the basic attitude of research is not so different from a genuine
psychoanalytic attitude in the sessions. I experienced research as a refreshing antidote against
an authoritative form of knowledge in psychotherapy, which is inherent in learning
psychotherapy as a skill from masters and in the necessary combination of personal therapy
and teaching. Furthermore I enjoyed the solidarity within the research community. Although I
am convinced that research is of vital importance for our profession, I do not think that it is
necessary that psychotherapists need to do the research themselves. But I have the impression
that being a therapist helps to ask the right research questions about psychotherapy and that in
this sense, a close connection between the two disciplines will always be necessary.
273

274
274

275
References
Adler, G. (1994). Borderline Psychopathology and its treatment. New York: Jason Aronson.
Abrams, S. (1987). The Psychoanalytic Process: A Schematic Model. International Journal
Psycho-Analysis., 68, 441-452.
Akhtar, S. (1992). Severe Personality Disorders and their Treatment. New York: Jason
Aronson.
Akhtar, S. (1995). Treating The Borderline Patient: A Contract-Based Approach. Journal
American Psychoanalytical Association., 43, 270-275.
Alden, L. A., Wiggins, J. S., Pincus, & A. L. (1990). Construction of Circumplex Scales for
the Inventory of Interpersonal Problems. Journal of Personality Assessment, 55, 521-
536.
American Psychiatric Association (APA). (1980). Diagnostic and Statistical Manual of
Mental disorders DSM-III. Washington DC: American Psychiatric Association.
American Psychiatric Association (APA). (1994). Diagnostic and Statistical Manual of
Mental disorders DSM-IV. Washington DC: American Psychiatric Association.
Anzieu, D. & Monjauze, M. (1993). Francis Bacon. Adda: Archimbaud.
Arrindell, W. A. & Ettema, J. H. M. (2003). SCL-90: handleiding bij een multidimensionele
pathologie-indicator [SCL-90: manual of a multidimensional pathology indicator].
Lisse, The Netherlands: Swets Test Publishers.
275

276
Auerbach, J. S. & Blatt, S. J. (1996). Self-representation in severe psychopathology: the role
of reflexive self-awareness. Psychoanalytical Psychology, 13, 297-34.
Balint, M. (1960). Primary Narcissism and Primary Love. Psychoanalytic Quarterly, 29, 6-
43.
Balint, M. (1968). The Basic fault. Therapeutic Aspects of Regression. London: Tavistock
Publications.
Barrett-Lennard, G. T. (1966). Technical note on the 64-items revision of the Relationship
Inventory. Waterloo: Unpublished Manuscript, University of Ontario.
Bateman, A. W. (1998). Thick- and Thin-Skinned Organizations and Enactment in Borderline
and Narcissistic Disorders. International Journal of Psychoanalysis, 79,13-25.
Bateman, A. & Fonagy, P. (1999). Effectiveness of partial hospitalization in the treatment of
borderline personality disorder: a randomised controlled trial. American Journal of
Psychiatry, 156, 1563-1569.
Bateman, A. & Fonagy, P. (2001). Treatment of borderline personality disorder with
psychoanalytically oriented partial hospitalization. 18 month follow-up. American
Journal of Psychiatry, 158, 36-42.
Bateman, A. & Fonagy, P. (2003). Health service utilisation costs for borderline personality
disorder patients treated with psychoanalytically oriented partial hospitalisation versus
general psychiatric care. American Journal of Psychiatry, 160, 169-170.
Bateman, A.& Fonagy, P. (2004). Psychotherapy for borderline personality disorder. Oxford:
Oxford press
Beck, A. T., & Steer, R. A. (1993). Beck Depression Inventory: Manual. San Antonio: The
Psychological Corporation.
Beenen, F., & Stoker, J. (2001). Psychoanalytic Process Visualised. Unpublished Manuscript,
Dutch Psychoanalytic Institute.
276

277
Berkeley, G. (1993). A Treatise Concerning the Principles of Human Knowledge. In G.
Berkeley, Philosophical Works. London: Everyman.
Bion, W. R. (1959). Attacks on linking. In W.R. Bion, Second thoughts (pp. 93-109). New
York: Jason Aronson.
Bion, W. R. (1963). Elements of Psychoanalysis. London: H. Karnac Books.
Bion, W. R. (1965). Transformations. London: H. Karnac Books.
Bion, W.R. (1984). A theory of thinking. In W. R. Bion, Second thoughts (pp. 110-119). New
York: Jason Aronson. (Original work published 1962)
Bion, W.R. (1984). Differentiation of the psychotic form the non-psychotic personalities. In
W. R. Bion, Second thoughts (pp. 43-64). New York: Jason Aronson. (Original work
published 1957)
Bion, W.R. (1984). Elements of Psychoanalysis. London: H. Karnac Books. (Original work
published 1963)
Bion, W.R. (1984). Learning from experience. London: Maresfield Reprints. (Original work
published 1962)
Bion, W.R. (1984). Transformations. London: H. Karnac Books. (Original work published
1965)
Bion, W.R. (1986). Attention and Interpretation. London: H. Karnac Books. (Original work
published 1970)
Bion, W.R. (1989). The Grid. In W. R. Bion, Two papers (pp.2 -33). London: H. Karnac
Books. (Original work published 1977)
Bion, W.R. (1997). The Grid. In F. Bion (Ed.), Taming Wild Thoughts. London: H. Karnac
Books. (Original work published 1963)
277

278
Blatt, S. J. (1992). The differential effect of psychotherapy and psychoanalysis on anaclitic
and introjective patients: the Menninger Psychotherapy Research Project revisited.
Journal of the American Psychoanalytic Association, 40, 691-724.
Blatt, S. J. & Auerbach, J. S. (2001). Mental representations, severe psychopathology, and the
therapeutic process. Journal of the American Psychoanalytical Association, 49, 113-
159.
Blatt, S. J., & Auerbach, J. S. (2003). Psychodynamic measures of therapeutic change.
Psychoanalytic Inquiry, 23 (2), 268-307
Blatt, S. J., & Behrends, R. S. (1987). Internalization, Separation-Individuation, and the
Nature of Therapeutic Action. International Journal of Psychoanalysis, 68, 279-297.
Blatt, S.J., & Shahar, G. (2004). Psychoanalysis: with whom, for what, and how ?
Comparisons with psychotherapy. Journal of the American Psychoanalytical
Association, 52, 393-447.
Blatt, S.J., & Shahar, G. (2005). A dialectic mode of personality development and
psychopathology: recent contributions to understanding and treating depression. In:
Corveleyn, J., Luyten, P.& Blatt, S.J. (Eds.), The theory and treatment of depression (p.
137-162). Leuven: Leuven University Press.
Blatt, S. J., Stayner, D., Auerbach, J., & Behrends, R. S. (1996). Change in object and self
representations in long-term, intensive, inpatient treatment of seriously disturbed
adolescents and young adults. Psychiatry, 59, 82-107.
Blount, C., King,J., & Menzies, D. (2002). Experiences of joining and leaving a therapeutic
community and how preparation may help reduce drop-outs. Therapeutic Communities,
23, 271-284.
Bohus, M., Haaf, B., Simms, T., Limberger, M.F., Schmahl, C., Unckel, C., Lieb, K.,
Linehan, M.M. (2004). Effectiveness of inpatient dialectical behavioral therapy for
278

279
borderline personality disorder: a controlled trial.Behavioral Research and Therapy, 42,
5, 487-499.
Bollas, C. (1987). The Shadow of the Object. New York: Columbia University Press.
Botella, C. & S.(2001). Figurabilité et régrédience [Representation and regression]. Revue
française de psychanalyse, 4, 1150-1239.
Bouman, T., Lutijn, F., Albersnagel, F., & Van der Ploeg, F. (1985). Enige ervaringen met de
Beck Depression Inventory [Some experience with the Beck Depression Inventory].
Gedrag, 13, 13-24.
Bowlby, J. (1969). Attachment. New York: Basic Books.
Bram, A., & Gabbard, G.O. (2001). Potential space and reflective functioning: towards a
conceptual clarification and preliminary clinical implications. International Journal of
Psychoanalysis, 82, 685-700.
Bucci, W. (1996). Research in Psychoanalysis: Process, Development, Outcome.
International Journal of Psycho-Analysis., 77:827-833.
Chiesa, M., Drahorad, C., & Longo, S. (2000). Early termination of treatment in personality
disorders treated in a psychotherapy hospital. Quantitative and qualitative studies.
British Journal of Psychiatry, 177, 107-111.
Chiesa, M. & Fonagy, P. (2000). Cassel personality disorder study. British Journal of
Psychiatry, 176, 485-491.
Chiesa, M., & Fonagy, P. (2003). Psychosocial treatment of personality disorder study. 36-
months follow-up,. British Journal of Psychiatry, 183.
Clarkin, J. & Levy, N. (2004). The influence of client variables on psychotherapy. In: M.
Lambert (Ed.), Bergin and Garfield’s Handbook of Psychotherapy and Behavior
Change (pp. 194-226). New York: Wiley.
279

280
Clarkin, J., Yeomans, F., & Kernberg, O. (1999). Psychotherapy for borderline personality.
New York: John Wiley & Sons.
Costa, P. T., Jr. & Widiger, T. A.(1994). Introduction. In P.T. Costa, Jr. P& T.A,Widiger
(Eds.), Personality disorders and the five-factor model of personality (pp. 1-8).
Washington, DC: American Psychological Association.
Dahl, H. (1988). Frames of mind. In: Dahl, H., Kächele, H. & Thomä H. (eds.),
Psychoanalytic Process Research Strategies (pp. 51-66). New York: Springer.
Damasio, A. (2000). The feeling of what happens. London: William Heineman.
De M’Uzan, M. (1989). Pendant la séance [During the session]. Nouvelle Revue de
Psychanalyse, 40, 147-163.
Derogatis, L. R. (1977). SCL-90: administration, scoring and procedures manual-I for the
R(evised) version. Baltimore: John Hopkins University School of Medicine, Clinical
Psychometrics Research Unit.
Diamond, D., Blatt, S. J., Stayner, D., & Kaslow, N. (1991). Self-other differentiation of
object representations. Unpublished research manual, Yale University, Department of
Psychiatry, New Haven.
Diamond, D., Blatt, S.J., Stayner, D., & Kaslow,N. (1993). Differentiation, cohesion, and
relatedness of self and other representations: a developmental scale. Unpublished
manuscript. Yale University, New Haven.
Doi, T. (1989). The Concept of Amae and its Psychoanalytic Implications. International
Review of Psychoanalysis, 16, 349-354.
Ekman, P. & Rosenberg, S.E. (1997). What the face reveals: basic and applied studies of
spontaneous expression using the Facial Coding System (FACS). New York: Oxford
University Press.
280

281
Ferenczi, S. (1949). Confusion of the Tongues Between the Adults and the Child—(The
Language of Tenderness and of Passion). International Journal of Psychoanalysis, 30,
225-230.
Ferro, A. (2000). La Psychanalyse comme oeuvre ouverte [Psychoanalysis as an open work].
Ramonville Sainte Agne: Erès.
Fertuck, E., Bucci, W., Blatt, S., & Ford, R. (2004).Verbal representation and therapeutic
change in anaclitic and introjective in-patients. Psychotherapy: Theory, Research,
Practice, Training, 41, 13-25.
Fonagy, P. (1999). The process of change and the change of the processes: what can change in
a ‘ good analysis’. Keynote address to the Spring meeting of the American
Psychological Association. www. dspp. com/papers/fonagy. htm
Fonagy, P. (Ed.). (2002). An open door review of outcome studies in psychoanalysis. London:
IPA.
Fonagy, P., & Target, M. (1998). Mentalization and the Changing Aims of Child.
Psychoanalysis Psychoanalytical Dialogues, 8, 87-114.
Fonagy, P., & Target, M. (2000). Playing with reality: III. The persistence of dual psychic
reality in borderline patients. International Journal of Psychoanalysis, 81, 853-873.
Fonagy, P. & Target, M. (2005). Some reflections on the therapeutic action of psychoanalytic
therapy. In : Auerbach, J.S., Levy, K.N. & Schaffer, C.E (Eds.).Relatedness, Self-
Definition and Mental Representation. Essays in honor of S.J.Blatt (pp. 192-212).
London , New York: Routledge.
Fonagy, P., Gergely, G., Jurist, E. L., & Target, M. (2002). Affect regulation, mentalization,
and the development of the self. New York: Other Press.
281

282
Fonagy, P., Leigh, T., Steele, M., Steele, H., Kennedy, R., Mattoon, G., Target, M., & Gerber,
A. (1996). The relation of attachment status, psychiatric classification, and response to
psychotherapy. Journal of Abnormal Psychology, 64, 22-31.
Fonagy, P., Moran, G. S., Edgcumbe, R., Kennedy, H., & Target, M. (1993). The Roles of
Mental Representations and Mental Processes in Therapeutic Action. Psychoanalytical
Study Child, 48, 9-48.
Fonagy, P., Steele, H., Moran, G., Steele, M., & Higgitt, A. (1991). The capacity for
understanding mental states: the reflective self in parent and child and its significance
for security of attachment. Infant Mental health Journal, 13, 200-217.
Fonagy, P., Steele, M., Moran, G. S., Steele, H., & Higgitt, A. C. (1993). Measuring the
ghosts in the nursery: An empirical study of the relation between parent’s mental
representations of childhood experiences and their infant’s security of attachment.
Journal of the American Psychoanalytic Association, 41, 957-989.
Fonagy, P., Steele, M., Steele, H., Higgitt, A., & Target, M. (1994). The theory and practice
of resilience. Journal of Child Psychology and Psychatry, 15, 231-257.
Fonagy, P., Steele, M., Steele, H., Leigh, T., Kennedy, R., Mattoon, G., & Target, M. (1995).
Attachment, the reflective self and borderline states: The predictive specificity of the
Adult Attachment Interview and pathological emotional development. In S. Goldberg,
R. Muir, & J. Kerr (Eds.), Attachment Theory: Social, Developmental and Clinical
Perspectives (pp. 233-278). Hillsdale, NJ: Analytic Press.
Fonagy, P., Target, M., Steele, H., & Steele, M. (1998). Reflective Functioning Manual
Version 5. Unpublished manual, University College London Psychoanalysis Unit,
London.
282

283
Freud, S. (1957). Formulations on the Two Principles of Mental Functioning. In Strachey
(Ed. & Trans.), The standard edition of the complete psychological works of Sigmund
Freud (Vol. 12, pp. 213-226). London: Hogarth Press. (Original work published 1911)
Freud, S. (1957). On narcissism. In Strachey (Ed. & Trans.), The standard edition of the
complete psychological works of Sigmund Freud (Vol. 14, pp. 67-102). London:
Hogarth Press. (Original work published 1914)
Freud, S. (1961). The ego and the id. In Strachey (Ed. & Trans.), The standard edition of the
complete psychological works of Sigmund Freud (Vol. 19, pp. 3-66). London: Hogarth
Press. (Original work published 1923)
Gabbard, G. O. (1989). Two subtypes of narcissistic Personality disorder. Bulletin of the
Menninger Clinic, 53, 527-532.
Gabbard, G. O. (1994). Psychodynamic Psychiatry in Clinical Practice. Washington:
American Psychiatric Press.
Gabbard, G.O. (2003). Rethinking therapeutic action. International Journal of Psychonalysis,
84, 823-841.
Gabbard, G.O., Lazar, S.G., Hornberger, J. (1997). The economic impact of psychotherapy: a
review. American Journal of Psychiatry, 154, 147-155.
Gerber, A.J., Fonagy, P., Bateman, A. & Higgitt, A. (2004). Structural ad symptomatic
change in psychoanalysis and psychodynamic psychotherapy of young adults: a
quantitative study of process and outcome. Journal of Psychoanalysis, 52(4), 1235-
1236.
George, C., Kaplan, N. & Main, M. (1985). The Berkeley Adult Attachment Interview.
Unpublished manuscript, Department of Psychology, University of California, Berkeley.
Godfrind, J. (1993). Les deux courants du transfer [The two streams of transference]. Paris:
Presses Universitaires de France.
283

284
Green, A. (1973). Le discours vivant [The living speech]. Paris: Presses Universitaires de
France.
Green, A. (2000). The central phobic position: a new formulation of the free association
method. International Journal of Psychoanalysis 81, 429-51.
Green, A. (2001, July). Meeting the author. Paper presented at the 42nd Congress of the
International Psychoanalytical Association, Nice, France.
Grotstein, J. S. (2000). Who is the dreamer who dreams the dream? Hillsdale: The Analytic
Press.
Gunderson, J.G. & Zanarini, M.C. (1987). Current overview of Borderline Diagnosis. Journal
of Clinical Psychiatry, 48 , (suppl. 8) , 5 - 14.
Gunderson, J.G., Frank, A.F., & Ronningstam, E.F. (1989). Early discontinuation of
borderline patients from psychotherapy. Journal of Nervous and Mental Disease, 177,
38-42.
Hoekstra, H., Ornel, J., & Fruyt, F. (1996). Persoonlijkheidsvragenlijsten NEO-PI-R en NEO-
FFI: Handleiding. Lisse, The Netherlands: Swets & Zeitlinger.
Holmes, J. (2001). The Search for the Secure Base. East Sussex: Brunner-Routledge.
Horvath, A. O., & Greenberg, L. S (1989). Development and validation of the Working
Alliance Inventory. Journal of Counselling Psychology, 36, 222-233.
Hume, D. (1985). A Treatise of Human Nature. London: Penguin Classics. (Original work
published 1739)
Jones, B. L., Nagin, D. S., & Roeder, K. (2001). A SAS procedure based on mixture models
for estimating developmental trajectories. Sociological Methods and Research, 29, 374-
393.
284

285
Jones, E. E. & Windholz, M. (1990). The psychoanalytic case study: toward a method for
systematic inquiry. Journal of the American Psychoanalytical Association., 38, 985-
1015.
Jones, E. E. (2000). Therapeutic action: A guide to psychoanalytic research. Northvale, N.J.:
Jason Aronson.
Karterud, S., Vaglum, S., Friis, S., Irion, T., Johns, S., Vaglum, P. (1992). Day hospital
therapeutic community treatment for patients with personality disorders. Journal of
Nervous and Mental Disease, 180, 238-243.
Kandel, E. R. (1999). Biology and the future of psychoanalysis: a new intellectual framework
for psychiatry revised. American Journal of Psychiatry, 156, 505-524.
Kant, I. (1929). Kritik der reinem vernunft [Critique of Pure Reason, tr. Kemp Smith]
London: Mc Millan. (Original work published 1781)
Kelly, T., Soloff, P.H., & Cornelius, P.H (1992). Can we study (treat) borderline patients?
Attrition from research and open treatment. Journal of Personality Disorders, 6, 417-
433.
Kernberg, O. F. (1975). Borderline Conditions and Pathological Narcissism. New York:
Jason Aronson.
Kernberg, O. F. (1980). Some implications of object relations theory for psychoanalytic
technique. In H. P. Blum (Ed.), Psychoanalytic Explorations of Technique (pp. 207-
239). New York: International University. Press.
Kernberg, O. F. (1984). Severe Personality Disorders. New Haven: Yale University Press.
Kernberg, O. F. (1996). A psychoanalytic theory of personality disorders. In J. F. Clarkin, M.
F. Lenzenweger, Major theories of personality disorders (pp. 106–137). New York:
Guilford Press.
285

286
Kernberg, O. F., & Clarkin, J. F. (1995). The inventory of personality organization.
Unpublished Manuscript.White Plains, NY: The New York Hospital-Cornell Medical
Center
Kinet, M. & Vermote, R. ( 2005). Mentalisatie. Leuven, Apeldoorn: Garant. In Press.
Klein, M. (1932). The Psychoanalysis of Children. London: Hogarth Press.
Klein, M. (1935). A contribution to the psychogenesis of manic-depressive states. In M.
Klein, The Writings of Melanie Klein. Volume I: Love, Guilt and Reparation (pp. 262-
289). London: Hogarth Press.
Klein, M. (1940). Mourning and its Relation to Manic-Depressive States. International
Journal of Psychoanalysis, 21, 125- 153.
Kohut, H. (1971). The Analysis of the Self. New York:International University Press.
Lecours, S., & Bouchard, M. (1997). Dimensions of mentalization: outlining levels of psychic
transformation. International Journal of Psychoanalysis, 78, 855-875.
Ledoux, J. (1998). The emotional brain. London: Phoenix.
Leichsenring, F., & Leibing, E. (2003). The effectiveness of psychodynamic therapy and
cognitive behavior therapy in the treatment of personality disorders: a meta-analysis.
American Journal of Psychiatry, 160, 1223-1232
Leichsenring, F., Kunst, H., Hoyer, J. (2003). Borderline personality organization in violent
offenders: Correlations of identity diffusion and primitive defense mechanisms with
antisocial features, neuroticism, and interpersonal problems. Bulletin of the Menninger
Clinic, 67, 314-327.
Lenzenweger, M. F., Clarkin, J. F., Kernberg, O. F., & Foelsch, P. A. (2001). The inventory
of personality organization: psychometric properties, factorial composition, and criterion
relations with affect, aggressive dyscontrol, psychosis proneness, and self domains in a
non-clinical sample. Psychological Assessment, 13, 577-591.
286

287
Leuzinger-Bohleber, M. (2002). DPV: a representative multicentre study of long-term
psychoanalytic therapies. In: Leuzinger-Bohleber,M. & Target, M. (red.). Outcomes of
Psychoanalytic Treatment. London: Whurr Publishers.
Linehan, M. M., Armstrong, H. E., Suarez A., Allmon D., & Heard, H. (1991). Cognitive-
behavioral treatment of chronically parasuicidal borderline patients. Archives General
Psychiatry, 48, 1060-1064.
Linehan, M. M., Heard, H. L., & Armstrong, H. E. (1993). Naturalistic follow-up of a
behavioral treatment for chronically parasuicidal borderline patients. Archives General
Psychiatry, 50, 971-974.
Locke, J. (1975). An Essay concerning human understanding. Oxford: Oxford University
Press. (Original work published 1689)
Luborgsky, L. (1975). Clinicians’ judgement judgement of mental health: specimen case
descriptions and forms for the Health Sickness Rating Scale. Bulletin Menninger
Clinic, 39, 448-480.
Luborsky, L. & Crits-Cristoph, P. (1988). The assessment of transference by the CCRT
method. In: ed. H. Dahl, H. Kächele & H. Thomä. Psychoanalytic Process Research
Strategies,. New York: Springer-Verlag, pp. 99-108.
Luborgsky, L & Crits-Christoph, P. (1990).Understanding transference: the CCRT method.
New York: Basic Books.
Lucas, R. (1993). The psychotic wavelength. Psychoanalytic psychotherapy, 7, 1, 15-24.
Luquet, P. (1981). Le changement dans la mentalization [Changes in mentalization]. Revue
Française de Psychanalyse, 45, 1023-1028.
Luquet, P. (2002). Les Niveaux de Penser [The levels of thinking]. Paris: Presses Univ.
France.
287

288
Mahler, M. S. (1968). On Human Symbiosis and the Vicissitudes of Individuation, Infantile
Psychosis. New York: International University Press.
Main, M. (1991). Metacognitive knowledge, metacognitive monitoring, and singular
(coherent)vs. multiple (incoherent)models of attachment: Findings and directions for
future research. In: C. Parkes, J. Stevenson- Hinde, & P. Marris (Eds.). Attachment
across the life cycle (p. 127-195). London: Routledge & Kegan Paul.
Marty, P. (1991). Mentalization et Psychosomatique [Mentalization and Psychosomatics].
Paris: Laboratoire Delagrange.
Marty, P. & De M’Uzan, M. (1963). La ‘pensée opératoire’ [Operational Thinking]. Revue
Française de Psychanalyse, 27(Suppl.), 1345- 1356.
Masterson, J. F. (1993). The Emerging Self. New York: Brunner Mazel Inc.
Matte-Blanco, I. (1988). Thinking, Feeling and Being. London and New York: Routledge and
the Institute of Psycho-Analysis.
Meissner, W. W. (1988). Treatment of patients in the borderline spectrum. New York: Jason
Aronson.
Meissner, W. W. (1996). The Therapeutic Alliance. New Haven: Yale University Press.
Meltzer, D. (1967). The psycho-analytical process. Perthshire: Clunie Press
Meltzer, D. (1975). Adhesive identification. Contemporary Psychoanalysis, 11, 289-310.
Mergenthaler, E. & Bucci, W. (1999). Linking verbal and non-verbal communications:
computer analysis of referential activity. British Journal Medical Psyhcology, 72, 339-
354.
Migone, P. (2003, October). Kernberg's Transference Focused Therapy (TFP) and Linehan's
Dialectical Behavior Therapy (DBT) treatments for borderline personality disorder:
what do they have in common? Paper presented at the VIII International ISSPD
Congress. Firenze, Italy.
288

289
Nagin, D. S. (1999). Analysing developmental trajectories: a semi-parametric group-based
approach. Psychological Methods, 4, 139-1777.
Nijenhuis, E. R. S., Van der Hart, O., & Kruger, K. (2002). The psychometric characteristics
of the Traumatic Experiences Questionnaire (TEC): First findings among psychiatric
outpatients. Clinical Psychology and Psychotherapy, 9, 200-210.
Normandin, L., & Foelsch, P. (1999). Transference Counter Transference Analysis (TCA): A
system to assess structural change in severe personality disorders thoughout treatment.
Paper presented at the annual conference for psychotherapy research, Braga, Portugal.
Normandin, L., Sabourin, S., Diguer, L., Poitras, K., Foelsch, P., & Clarkin, J. (2002).
Evaluation de la validité théorique de l’inventaire de l’organization de la personnalité
[Evaluation of the theoretical validity of the inventory of personality organization].
Canadian Journal of Behavioral Science, 34, 59-65.
O’Malley, S. S., Suh, C. S.& Strupp, H. H. (1983). The Vanderbilt Psychotherapy Process
Scale: a report on the scale development and a process-outcome study. Journal of
Consulting and Clinical Psychology, 51, 581-586.
Ogden, T. (1997). Reverie and Interpretation. Northvale: Jason Aronson.
Perry, J., Banon, E., Ianni, F. (1999). Effectiveness of psychotherapy for personality
disorders. American Journal of Psychiatry, 156(9), 1312-1321.
Pieters G., & Vermote R. (2002). In-patient treatment of borderline personality disorder:
convergences and divergences between a psychoanalytical and a cognitive-behavioral
approach. Acta Neuropsychiatrica 14(2), 81-85.
Ramachandran, V. S. (1998). Phantoms in the Brain. London: Fourth Estate.
Rosenfeld, H. (1987). Impasse and Interpretation. Londen: Tavistock publications.
Sandell, R. (2005, March). Learning form the patients through research. Paper presented at
the 18th Conference of the European Psychoanalytical Federation. Vilamoura, Portugal.
289

290
Sandell, R., Blomberg, J., Lazar, A., Carlsson, J., Broberg, J., & Rand, H. (2000). Varieties of
long-term outcome among patients in psychoanalysis and long-term psychotherapy: a
review of findings in the Stockholm outcome of psychoanalysis and psychotherapy
project (STOPP). International Journal of Psychoanalysis, 81(5), 921-943.
Sanderson, C., Swenson, C., Bohus, M. (2002). A critique of the American psychiatric
practice guideline for the treatment of patients with borderline personality disorder.
Journal of Personality Disorders, 16(2),122-129.
Sandler J. (1960). The background of safety. International Journal of Psychoanalysis 41, 352-
356.
Sandler, A. M. (1993). Introduction to the one-day conferenceon the work of the Anna Freud
Centre’s Young Adults Scheme: development issues in psychoanalytic work with young
adults. Bulletin of the Anna Freud Centre, 16, 3.
Sansone, R. A., Wiederman, M. W., & Sansone, L. A. (1998). The Self-Harm Inventory
(SHI): Development of a scale for identifying self-destructive behaviors and borderline
personality disorder. Journal of Clinical Psychology, 54, 973-983.
Skodol, A., Buckley, P., & Charles, E. (1983). Is there a characteristic pattern to the treatment
history of clinic outpatients with borderline personality? Journal of Nervous and Mental
Disease, 71, 405-410.
Smith, T.E., Koenigsberg, H.W., Yeomans, F.E., Clarkin, J.F., & Selzer, M.A. (1995).
Predictors of dropout in psychodynamic psychotherapy of borderline personality
disorder. Journal of Psychotherapy Practice and Research, 4, 205-213.
Spence, D. P. (1993). Discussion: new understandings of the psychoanalytic process. Journal
American Psychoanalytical Association, 41(S),131-141.
Spielberger, C. (1988). Manual for the State-Trait Anger Expression Inventory (STAXI).
Odessa, FL: Psychological Assessment Resources.
290

291
Spielberger, C., Gorsuch, R., Lushene, R. (1970). Manual for the State-Trait Anxietyr
Inventory. Palo Alto: Consulting Psychologists Press.
Spitzer, R. & Williams, J. (1985). Structured clinical interview of DSM-III-r Personality
Disorders (SCID-II). New York State Psychiatric Institute, Biometrics Research
Department.
Sroufe, L. A. (1996). Emotional Development: The Organization of Emotional Life in the
Early Years. New York: Cambridge University Press.
Steiner, J. (1993). Psychic Retreats. London: Routledge.
Stern, D. N. (1985). The Interpersonal World of the Infant. New York: Basic Books.
Stoker, J., & Zevalkink, J. (2002). What is going on during treatment? Visualising the
psychoanalytic process. Poster presented at the J. Sandler Conference, London, United
Kingdom.
Stoker, J., & Zevalkink, J. (2003). Handleiding voor de.invuller en beoordelaar van de
Psychoanalytische Periodieke Beoordelingsschaal (PPBS) en Psychoanalytische
Periodieke Rapportage (PPR). Versie 1.2. Unpublished document Dutch
Psychonalytical Institute.
Stoker, J., & Zevalkink, J. (2005). Handleiding voor de Psychoanalytische Proces Rating
Schaal (PPRS) en Psychoanalytische Rapportage (PPR). Versie 1.2. Unpublished
Manuscript, Dutch Psychoanalytic Institute.
Stone, M. H. (1990). The Fate of Borderline Patients: Successful Outcome and Psychiatric
Practice. New York: Guilford.
Target, M., & Blatt, S. (2001). Workshop on RF on AAI and ORI. Preconference meeting of
the J.Sandler Research Conference, March UCL London
Target, M., & Fonagy, P. (1996). Playing with reality: 2. The development of psychic reality
from a theoretical perspective. International Journal of Psychoanalysis, 77, 459-479.
291

292
Thunissen, M., Remans, Y., & Trijsbrug, R. (2004). Vroegtijdige beëindiging van
kortdurende klinische psychotherapie [Early drop out from short hospitalization base
psychotherapy].Tijdschrift voor Psychiatrie, 46, 739-743.
Thys, M. en Vermote, R. (1995). Trauma en taboe [Trauma and Tabou]. Leuven-Apeldoorn:
Garant..
Tustin, F. (1986). Autistic barriers in neurotic patients. London: Karnac.
Vanden Berghe B &Vermote R. (2000). Onbewuste aspecten van het werk als
verpleegkundige in een psychoanalytisch dagcentrum. Tijdschrift voor hulpvrleners in
de geestelijke gezondheidszorg , 6,337-348.
Van der Ploeg, H., Defares, P., & Spielberger,C. (1980).Handleiding bij de Zelf-
beoordelingsvragenlijst, ZBV.[Manual on the self-report anxiety inventory] Lisse, The
Netherlands: Swets & Zeitlinger.
Van der Ploeg, H., Defares, P., & Spielberger,C. (1982).Handleiding bij de Zelf-
beoordelingsvragenlijst boosheid en woede,ZAV.[Manual on the self-report anger
inventory] Lisse, The Netherlands: Swets & Zeitlinger.
Van Lysebeth-Ledent, M. (2001). Du réel au rêve [From the Real to the dream]. Paris:
L’Harmattan.
Van Sina, M.J. (1993). Per-agir. Revue Belge de Psychanalyse, 23, 13-32.
Vaughan, S. C.,& Roose, S. (1995). The analytic process: clinical and research definitions.
International Journal of Psychoanalysis, 76, 343-356.
Vaughan, S. C., Spitzer, R., Davies, M., & Roose, S. (1997). The Definition And Assessment
Of Analytic Process: Can Analysts Agree? International Journal of Psychoanalysis., 78,
959-973.
Verhaest, S. (1982). Comprehensive assessment of two diffrenet types of therapuetic
communities. International Journal of Social Psychiatry, 24, 46-54.
292

293
Verhaest, S. (1983a). The assessment of the development of a therapeutic community.
International Journal of Therapeutic Communities, 4, 183-195.
Verhaest, S. (1983b). The assessment of the development of the psychotherapeutic milieu in a
therapeutic community. In: W.R.Minsel & W. Heiff (Eds.). Methodology in
Psychotherapy Research. Peter Lang, Frankfurt a. Main, Bern p. 172-181.
Verheul, R., Van den Bosch,L.M., Koeter, M.W., De Ridder, M.A., Stijnen, T., Van den
Brink,W. (2003). Dialectical behaviour therapy for women with borderline personality
disorder. British Journal of Psychiatry, 102, 135-140.
Verheul, R. (2004). Niets is veranderlijker dan de mens. Inaugurale rede. Amsterdam:
Vossiuspers UvA.
Vermote, R & Callens, K.(1994, June). Long term residential psychoanalytic psychotherapy.
Paper presented at the First European Congress on Personality Disorders. Nijmegen,
The Netherlands.
Vermote, R. (1994). Le mythe d'Oedipe à la lumière du mythe du Sphinx. Revue Belge de
Psychanalyse , 24, 29-43.
Vermote, R. (1995). Réalités psychiques et formes [Psychic reality and Forms]. Revue Belge
de Psychanalyse, 27, 39-55.
Vermote, R. (1996a), Een setting voor behandeling van persoonlijkheidsstoornissen,
Tijdschrift voor Cliëntgerichte Psychotherapie, 34, 3, 39-49
Vermote R.(1996b), Op zoek naar de grondslagen van het psychisch functioneren. Tijdschrift
voor Psychoanalyse, 2, 122-124.
Vermote, R. (1996c). Op zoek naar de bron. In: J.Baneke en R.Pierloot (Eds.). Psychoanalyse
en antropologie (p.73-81). Amsterdam: Thela Thesis
Vermote, R. (1997a). Helende factoren in de behandeling van pre-oedipale stoornissen. In: M.
Hebbrecht (Ed.) Bouwen aan een basis. St. Truiden: Smets & Ruppol.
293

294
Vermote, R. (1997b). The dark side of the moon. Psychische pijn als element van het
psychisch functioneren. In: Dehing J. & Pieters G. (Eds.). Van rêverie tot ideologie
(pp.213-224). Leuven, Apeldoorn: Garant.
Vermote, R. (1997, July) The Grid: that two-dimensional crutch.Paper presented at the Bion
Centenary Conference. Turin, Italy.
Vermote, R. (1998a). Les transformations psychiques et la grille de Bion [Psychic
transformations and Bion's Grid]. Revue Belge de Psychanalyse, 32, 49-65.
Vermote, R. (1998b). De Grid en de Caesura in het werk van Bion. In: Dehing J. (red.) Een
bundel intense duisternis ( p.53-73). Leuven-Apeldoorn: Garant.
Vermote, R. (1999c). Listening to the unknown, often unborn person behind the curtain of
mental retardation.In: Aldridge, D. Institute for Music Therapy, collected papers on
Info CD Rom II, University Witten Herdecke
Vermote, R. (2000a). Psychoanalytische en psychiatrische diagnostiek bij
persoonlijkheidsstoornissen. Tijdschrift voor Psychiatrie, 9,667-674.
Vermote, R. (2000b, December). Inpatient Treatment of Borderline Personality Disorder.
Paper presented at the Tryptich Congress. Maastricht, The Netherlands.
Vermote, R. (2000c). Psychose et souffrance. In: De Hert M., Peuskens J., Thys E., Vidon G.
(Eds.). Raisonner la déraison.Bruxelles: Epo en Frison-Roche.
Vermote R. (2000d) Les transformations psychiques et la ‘ Grille ‘ de Bion. In : J. Dehing,
editor. Autour de W.R. Bion. Essais Psychanalytiques(p. 65-94).Paris, L’Harmattan 65-
94.
Vermote, R. (2000e, October). Un modèle psychanalytique des troubles de personalité et son
schéma de recherche [A psychoanalytic model of perosnlity disorders and its research].
Paper presented at the Belgian Psychoanalytical Society. Brussels.
294

295
Vermote, R. (2001, November). Bions theorie over het denken [Bion's theory of thinking].
Paper presented at the Faculty of Philosophy of the Catholic University of Leuven.
Vermote, R (2002a) The nature of the problems of psychoanalysis with psychotic and so-
called “difficult” patients. Panelreport IPA. International Journal of Psychoanalysis.
83,3,689-694.
Vermote, R. (2002b, March) Discussion of Clarkin, J.,The development of a transference
focused psychotherapy for borderline personality disorder. Paper presented at the 3th
Joseph Sandler Research Conference. London, United Kingdom.
Vermote, R. (2002c, March). Implicaties van Bions model voor de psychoanalytische praktijk.
Lezing voor de Nederlandse Vereniging voor psychoanalyse en het Nederlands
gezelschap voor Psychoanalyse.
Vermote, R. (2002d, May). La recherche en Psychanalyse [Research in Psychoabanalysis].
Paper presented at the Congres Européen des langues romanes de l’IPA. Brussels.
Vermote, R. (2002e, Juin). Een mantel van gedachten. Psychopathologie vanuit het model van
Bion. Paper presented at the Vlaamse Vereniging Psychoanalytische Psychotherapie
Annual Conference. Kortenberg.
Vermote, R.(2002f). Mentalisation et psychopathologie. Une approche clinique de l’oeuvre de
Bion. Communications Ecole Belge de Psychanalyse, 33, 2, 20-26
Vermote, R. (2003a) The analyst at work. Two sessions with Catherine. With commentaries
by A. Goldberg and R. Roussillon. International Journal of Psychoanalysis, 84, 1415-
1429.
Vermote R, (2003b). De cirkel van de herhaling. In : De Buyne.A, Heuves,
(Eds.) De Ezel en de Steen. Amsterdam: Boom.
Vermote, R. (2004a, October). Examining shifts of level during a first interview with Bion's
Grid. Paper presented at the Initiating Psychoanalyis Workshop. Vienna, Austria.
295

296
Vermote, R. (2004b). Het meten van een intrapsychische veranderingen. Bulletin van de
vereniging voor Klinische Psychotherapie, 6, 3 , 21-23.
Vermote, R. (2005) A dictioniary of the work of Bion by Lopez-Corvo, Tijdschrift van
Psychiatrie.
Vermote, R. & F. Auwerkerken ( 1999). Een psychodynamisch-structurele benadering van
de DSM IV-as II persoonlijkheidsstoornissen. Diagnostiek-Wijzer, 33,4, 171-184.
Vermote, R., Corveleyn, J., & Vertommen, H. (2002). The Kortenberg-Leuven Study on
inpatient psychoanalytically oriented hospitalization for personality disorders. In P.
Fonagy (Ed.). An open door review of outcome studies in psychoanalysis. London: IPA.
Vermote, R., Smits, D., Claes, L. & Vertommen, H. (2005). The Inventory of Personality
Organization-Revised: a short version, its validity and discriminative power.
Unpublished paper.
Vermote, R., Vertommen, H., Corveleyn, J., & Peuskens, J. (2002, August). An integrative
psychoanalytical model on personality disorders. Poster session presented at the 12
World Congress of Psychiatry, Yokohama, Japan.
Vermote R., Van sina, M.J. (1998). A psychoanalytic hospital unit for people with severe
personality disorders. In: Pestalozzi J, Frisch S, Hinshelwood RD, et al.( Eds.)
Psychoanalytic Psychotherapy in Institutional Settings (p. 75-93). London: Karnac
Books.
Vermote, R., Maes, E., Vertommen, H., Corveleyn, J., Verhaest, Y., Peuskens, J. (2003,
October). Properties of the Dutch version of the Inventory of Personality Organisation.
Poster session at the VIII International ISSPD Congress, Controversial Issues in
Personality Disorders, Abstract book, p. 124, Firenze, Italy.
Vermote, R., Vertommen, H., Corveleyn, J., Verhaest, Y., Peuskens, J. (2003, October).
Assessing intrapsychic changes in patients with personality disorders in an in-patient
296

297
treatment on analytic lines. Poster session presented at the VIII International ISSPD
Congress, Controversial Issues in Personality Disorders.Firenze, Italy.
Vermote R., Vertommen H., Verhaest, Y., Franssen, M., Geenen, G., Corveleyn, J., Peuskens
J. (2004, March). Measuring dimensions of the psychoanalytic process on the ORI.
Poster session presented at theWorking at the Frontiers, 43RD International Conference
of the International Psychoanalytic Association. New Orleans, USA
Vermote, R., Vertommen, H., Verhaest, Y., Stroobants, R., Corveleyn, J., Peuskens, J. (2004,
March). The Kortenberg-Leuven Process-Outcome Study. Poster session presented at
theWorking at the Frontiers, 43RD International Conference of the International
Psychoanalytic Association. New Orleans, USA.
Vermote, R, Vertommen, H., Verhaest, Y, Corveleyn, J. & Peuskens, J.( 2004, March).
Preliminary results of the Kortenberg-Louvain Process Outcome Study, Paper presented
at the meeting of the research Fellows IPA, 43RD International Conference of the
International Psychoanalytic Association. New Orleans, USA.
Vermote, R., Vertommen, H. , Corveleyn, J. , Verhaest, Y, Peuskens, J. (2004, April).
Clinical Implications of The Kortenberg-Louvain Process Outcome Study .Paper
presented at the Third New Style European Psychoanalytical Federation Annual
Conference. Helsinki, Finland.
Waldron, S. (1996). The Analytic Process Scales Coding Manual, unpublished manuscript.
Wallerstein, R. S. (1988). Assessment Of Structural Change In Psychoanalytic Therapy And
Research. Journal of the American Psychoanalytical Association., 36(S):241-261.
Wallerstein, R. S. (1991). Assessment of the structural change in Psychoanalytic theory and
research. In T. Shapiro, T. (Ed.), The concept of Structure in Psychoanalysis. Madison:
International Universities Press.
297

298
Wallerstein, R.S. (1994) Psychotherapy Research and its Implications for a theory of
Therapeutic Change. A forty year overview. Psychoanalytic Study of the Child, 49, 120-
141.
Weertman, A., Arnzt, & Kerhof, (2000). SCID II persoonlijkheidsvragenlijst. Lisse: Swets &
Zeitlinger.
Weinshel, E. M. (1988). The Many Borders of Borderline: On the Virtues of Modesty in
Psychoanalytic Diagnosis. Psychoanalytical Inquiry, 08, 333-352
Weinshel, E. M. (1984) Some observations on the psychoanalytic process Psychoanalytic
Quarterly 53, 63-92.
Weiss, E (1995). Empirical studies of the psychoanalytic process. In T.
Shapiro & R.N. Emde (Eds.). Research in psychoanalysis: process,
development, outcome (pp. 7-29). Madison: International University
Press.
Wilgowicz, P. (1991). Le Vampirisme [Vampyrism]. Lyon: Césura.
Winnicott, D. W. (1958). The observation of infants in a set situation. In D. W. Winnicott,
Collected Papers: Through Paediatrics to Psychoanalysis (pp. 52-69). New York: Basic
Books.
Winnicott, D. W. (1960). Ego distortion in terms of true and false self. In D.
W. Winnicott, The Maturational Processes and the Facilitating
Environment (pp. 140-152). London: Hogarth Press.
Winnicott, D. W. (1960). The theory of the parent-infant relationship. In D. W.Winnicott, The
Maturational Processes and the Facilitating Environment (pp. 37-55). London:
Hogarth Press.
Winnicott, D. W. (1965). Communicating and not communicating leading to a
study of certain opposites. In D. W. Winnicott, The Maturational
298

299
Processes and the Facilitating Environment (pp. 179-192). New York:
Int. Univ. Press.
Winnicott, D. W. (1967). Mirror-role of Mother and Family in Child
Development. In P. Lomas (ed.), The Predicament of the Family: A
Psycho-analytical Symposium. London: Hogarth Press and the Institute of
Psycho-Analysis.
Winnicott, D.W. (1967), Mirror-role of Mother and family in Child Development. In:
D.W.Winnicott, Playing and Reality. Penguin Books, Hammersworth, p.130-139.
Winnicott, D.W. (1969). The Use of an Object. International Journal of Psychoanalysis, 50,
711-716.
Winnicott, D. W. (1971). Playing and Reality.New York: Basic Books.
Zanarinini, M. C., Frankenburg, F. R., Henne, J., & Silk, K. (2003). The longitudinal course
of borderline psychopathology: 6 year prospective follow-up of the phenomenology of
borderline personality disorder. The American Journal of Psychiatry, 160, 274-83.
299

300
300
