Vol.
1,
núm.
1Marzo
2002
Revista Internacional On-line / An International On-line Journal
Reliability of the Spanish version of the Karolinska Psychodinamic profile.
Dr. J. Pastor
Centro de Salud Mental Erandio
Erandio (Bizkaia)
jpastor@coiib.es
Elizabeth Basaguren
psicóloga, psicoterapeuta, práctica privada
E. de la Sierra
Prof. Dr. José Guimón Ugartechea
jose.guimon@hcuge.ch
Resumen
Se ha realizado la traducción y adaptación al español del KAPP estudiando su validez concuirrente y su
fiabilidad entre examinadores. Los resultados muestran una fiabilidad aceptable en muchos de los items
aunque menor que la que obtuvieron los autores del instrumento.
Abstract
The Spanish version of the KAPP has been made in conformity with presently accepted methodology: The
fact that we have obtained a translation which is semantically close to the original as possible has to a
great extent eliminated the possible transcultural differences inherent in differences in idiom, and this has
allowed us to apply the same concurrent validity criteria to the Spanish version which we applied to the
original version. Reliability between examiners was also analyzed.
The results obtained in our study, although acceptable, show lower correlation coefficients than those
obtained by the Swedish authors in their original publication. Later studies carried out by persons who
were not directly involved in the creation of the method of measurement have obtained figures similar to
ours Further studies with different groups of subjects and researchers will be required to provide
information about the applicability of KAPP in research and in clinical practice.
INTRODUCTION
The construction of psychodynamic evaluation systems which are valid, reliable and inexpensive in their
application is a real challenge within the field of research in psychotherapy. The tools presently available
display varying degrees of reliability, require intensive training beforehand and take a great deal of time
to administer.
The Karolinska Psychodynamic Profile - KAPP - is a method of measurement based upon the
psychoanalytical theory which includes recent thinking about objectal relationships and permits us to
evaluate relatively steady, gradual types of change in mental function, as well as character traits.
Unlike other scales, the KAPP avoids metapsychological formulations and includes descriptions which are
closely related to clinically observed phenomena.
According to its authors, it is a reliable tool when used by professionals who have psychotherapeutic
experience and are trained in its application.
Following its translation into Spanish, we now present the results of the reliability study of this version.
MATERIALS AND METHODS
Assessors
6 researchers were involved. Two were interviewers: one was a psychiatrist (PQ1) and the other a
psychologist (PS1), both with extensive dynamic and clinical training. Three were assessors: one was a
psychiatrist (PQ2) - highly trained, like the two interviewers - and two psychologists. One of these (PS3)
had a lower level of training and psychodynamic experience and the other (PS2) was of an intermediate
level. PQ1 and PS2 were involved with the translation of the Spanish version.
Subjects
We worked with a sample of 24 subjects. Of these, half were undergoing psychiatric treatment and were
diagnosed as DSM-IV on Axis I. The other half had no previous history.
The distribution of diagnoses in the group of patients was as follows: four cases of schizophrenia, four
dysthymic disorders, one major depressive disorder, one generalised anxiety disorder, one alcohol
dependency and one opiate dependency.
The patients came from the acute unit of the Hospital Civil in Basurto. They were informed of the aims of
the study and included in it after giving their consent. All of them were interviewed one or two days
before being discharged. Six patients declined to take part, and two more did not appear after having
agreed to participate. The average age was 31.9 years (range from 19-42) and distribution by sex was
equal (6/6), the majority being single (9/3) and unemployed (9/3).
The other twelve subjects, who had no previous history, were working mostly in a social work centre, and
attended voluntarily. They were informed about the study and included in it after an interview with
another researcher with responsibility for selection of the sample. One candidate, who showed clear signs
of a serious personality disorder, was rejected. The average age of this group was 33.6 (range 26-43)
and distribution by sex was equal (6/6), the majority being single (8/4) and employed (8/4).
Procedure
Before beginning the interviews and their assessments, there were several group sessions of
familiarisation and training with the method of measurement. An interview was carried out and recorded
on video - this was not included in the reliability analysis - and was subsequently viewed, discussed and
evaluated in a group session by most of the researchers. Assessor PQ2 was not present at any of these
sessions.
The two interviewers (PQ1 and PS1) carried out between them a total of 24 interviews, 12 each, which
were recorded on video. The theoretical formulations and suggestions described by Kernberg (1984) then
followed, in order to form a "structured" procedure for psychodynamic interview. The average length was
one hour twenty-five minutes, all recorded in one session. The two interviewers did not know any of the
subjects, nor did they have any information about them. Each interviewer saw six patients and six
controls. After the interview they filled in the various KAPP sub-scales. In addition, each interviewer had
to view and assess the other interviewer's videos. These assessments were not discussed, so that the
learning situation was personal to each interviewer.
The rest of the researchers each assessed twelve videos, six by each interviewer, and in equal numbers of
subjects with diagnosis/subjects without diagnosis, and of men/women.
With this procedure we are attempting to:
A) Analyse reliability between examiners by studying the correlation existing between the scores awarded
independently by the researchers to the subjects in the sample.
B) Evaluate the influence of the two different levels of training by analysis of the correlation between
assessors with a different degree of experience and dynamic training.
C) Estimate the impact of pre-training in the use of the method of measurement by comparing the results
of the interviewers with an assessor of a similar level (PQ2) who did not attend the group training
sessions before the interviewers' assessment.
Statistical Analysis
An independent researcher who was not one of the assessors selected the sample and carried out the
statistical analysis. In view of the features of the measurement system, it was decided to use nonparametric methods. The Spearman coefficient was used to study the correlation between the different
administrations, and the Wilcoxon method of mean comparison to analyse the differences.
Results
Table 1 shows the correlations between the evaluations made by the different assessors. The means vary
between 0.662 and 0.360. Researcher PQ2, who did not attend the group sessions of training and video
viewing, obtained the lowest coefficients.
Pictures one and two show in graphic form the mean correlations per item (1) and the means of the
correlations of the different assessors (2).
Table 2 shows the correlations of the assessors who were present at the training sessions. The
average values per item vary between 0.80 - feeling of belonging - and 0.32 - sexual function. In
general, the items show acceptable correlation coefficients, except for sexual function (0.32), bodily
function (0.35), alexithymia (0.44) and bodily appearance (0.55), which are the sub-scales with the
lowest correlation coefficient and are not significant.
Pictures three and four show in graphic form the mean correlations per item (3) and the means of
the correlations of the different assessors (4).
Table three shows the Wilcoxon coefficients of the assessors who were present at the training
sessions.
Discussion
The results obtained in our study, although acceptable, show lower correlation coefficients than
those obtained by the Swedish authors in their original publication. Later studies carried out by
persons who were not directly involved in the creation of the method of measurement have obtained
figures similar to ours. Thus Haver et al (1), using an assessor from the original study with a sample
of 21 women who were alcohol abusers or alcohol-dependent, obtained a mean correlation of 0.56.
In this study the control sub-scales, tolerance of frustration, alexithymia and body image did not
reach statistical significance.
The correlation coefficients obtained in our study correspond to those published in other studies
which use different methods of psychoanalytical evaluation. Thus Bellak (2) in a study with the EFA
scale of evaluation obtained values of reliability between assessors of between 0.61 and 0.88. Dahl
(3), using the same scale with trained assessors, encountered lower correlations of between 0.13
and 0.81 (mean 0.47); 9 of the 12 scales had correlations lower than 0.50. Another group of
researchers obtained a mean correlation of 0.77, with intensive pre-training of assessors and
interviewers (4,5).
We ourselves did not find that there was such a clear tendency for reliability to increase in relation to
years of experience and psychotherapeutic training as in relation to group training in the use of the
measurement method. We consider that the scores obtained by PQ2 may be interpreted in this way.
The question of the amount of training and psychodynamic experience involved is not easy to
quantify.
There are several factors which might explain the differences in results from those of the original
study. Our data were based upon blind and independent evaluations of the interviews recorded on
video, using two interviewers and a sample where half of the subjects fulfilled different DSM-IV (Axis
I) diagnostic criteria. The Swedish authors listened to the tapes together, awarded scores
independently and subsequently discussed the evaluations made. This procedure, without doubt,
made the assessors' criteria more similar. The lack of experience and specific training may also
contribute, since our assessors' reliability has not been demonstrated. Until now there has been no
information about the amount of training required to produce a competent KAPP assessor.
On the other hand, the fact that a higher percentage of subjects fulfilling DSM-IV criteria were used
may also have contributed to the fact that lower correlations were obtained.
Further studies with different groups of subjects and researchers will be required to provide
information about the applicability of KAPP in research and in clinical practice.
Bibliografic
1- B.Haver, P.Svanborg, S. Lindberg . Improving the usefulness of the Karolinska Psychodinamic
Profile in research: proposals from a reliability study. Acta Psychiatr Scand 1995:92: 132-137
2- Bellak L, Hurvich M, Gediman HK. Ego functions in schizofrenics, neurotics and normals. A
systematic study of conceptual, diagnostic, and therapeutic aspect. New York: John Wiley &
Sons,1984: 160-166.
3- Dahl AA. A study of agreement among raters of Bellak#s Ego Function Assessment Test. In:
Bellak L, Goldsmith La ed. The broad scope of ego function assessment. New York: John Wiley &
Sons, 1984: 160-166.
4- Sohlberg S, Norrig C. Presentation av diagnosmetod: Bed`mning av jag-fungerande. Psykolog
Tidningen 1988: 15: 9-12.
5- Norring 11c, Sohlberg S, Rosmark B, Humble K, Holmgren S, Nordquist C. Ego functioning in
eating disorders: Description and relation to diagnostic classification. Int J Eating Disord 1989: 8:
607-621.
ASMR Revista Internacional On-line - Dep. Leg. BI-2824-01 - ISSN (en trámite)
CORE Academic, Instituto de Psicoterapia, Manuel Allende 19, 48010 Bilbao (España)
Copyright © 2002
IMPORTANTE: Algunos textos de esta ficha pueden haber sido generados partir de PDf original, puede sufrir variaciones de maquetación/interlineado, y omitir imágenes/tablas.
