Creating a case formulation is an important and basic clinical skill. Based on the patient’s narrative, it is a tool for the therapist to structure ideas about what has caused the patient’s problems, why the problems persist and how to make the patient feel better (1). The case formulation can fill the gap between diagnosis and treatment and can be perceived as lying at the intersection of aetiology and description, theory and practice, and science and art (2). When developed, the case formulation is based on the psychotherapist’s knowledge and preconceptions, what the patient has chosen to reveal, how a patient is presenting himself and interacts with the therapist. In this study a simplified method to describe patients is tested.
The primary function of a case formulation is to guide treatment. A psychodynamic formulation includes motivational components, aims to identify central dynamic patterns that repeat themselves over and across situations resulting in distress or psychological limitations, and guides clinicians in minimizing inferences from observable clinical data (3). Many therapists have the opinion that case formulation is a time- consuming and difficult process, therefore, a simple mental construction would be sufficient.
Bowden (4) reported that 90% of educators in a study ranked case formulation skills as very important, even essential. However, case formulation skills are difficult to acquire. Kuyken and colleagues (5) studied the quality of case formulations constructed by 115 health professionals, 44 % of the formulations were considered good enough. Eells and colleagues (6) evaluated 56 intake formulations, of which less than half described predisposing life events and/or inferred psychological mechanisms, as necessary in a proper case formulation.
Seitz (7) summoned a group of psychoanalysts in an effort to create reliable case formulations but they never succeeded. He concluded that they applied different levels of inference to the clinical data and hence never agreed on what was centrally important. A basic prerequisite in formulating cases is a certain level of agreement among raters. In their review, Barber and Crits-Christoph (8) found that structured case formulations can attain reliability. This is confirmed in a study by Garb (9) with clinicians sharing the same theoretical background. A recent publication also shows that very experienced clinicians with similar theoretical stance produced reliable, and thus clinically relevant formulations without elaborate instructions about how to structure the case formulations (10).
Psychotherapy is interpersonal in nature. Several methods have been developed over the last decades to identify and describe interpersonal patterns in psychotherapy, often as part of a case formulation. The most well-known is the Core Conflictual Relationship Theme (CCRT) method (11).
The CCRT method constructed by Luborsky and Crits-Christoph is based on the assumption that humans display a central relationship theme or transference pattern (12). That is, people in general display the same patterns with different people, in various relationships, including the patient-therapist relationship (13). In several studies this method has shown high reliability in identifying peoples wishes, responses from others and responses of self (11)(14). There is also evidence for the validity of the CCRT (15, 16) (14). The CCRT has strong convergent validity with other transference- related measures such as the circumplex model of human reactions in relation to others, the Structural Analysis of Social Behavior (SASB) (17) (18).
The CCRT scoring is based on narratives presented by the patient (19). The narratives should describe interactions with others, even the therapist. These descriptions are defined as relational episodes (REs), which includes three components: the patient’s expectations or wishes (W) in meeting another person, how the other is considered or expected to react (RO; Response Other) and how the patient responded (RS; Response Self). REs depict real, imagined or dreamt episodes from the patient’s point of view. An RE told by the patient, can be more or less complete. The REs can be collected from therapy sessions or directly from an interview focusing on relationships with central others, as in the Relational Anecdote Paradigm (RAP) (20) (pp 109-120) developed by Crits-Christoph and Luborsky.
In the CCRT method, trained scorers identify and demarcate REs in the transcribed material from therapy sessions or the RAP interview. Other trained independent scorers, not involved in the identification of REs in the transcriptions, identify and categorise W, RO, and RS from the demarcated REs. The most frequent pattern is considered the most useful description of a patient’s CCRT. The goal is to avoid interpreting the patient’s narrative. The chosen categories should reflect literally what the patient told, and inferences should be avoided.
An example of an RE and a tailor-made description of W, RO and RS can illustrate the process. Here is an example of an identified RE:
“He came over, unannounced to have a coffee with me. I pretended to enjoy his visit, since he is a friend since long…but actually I really wanted to sit down and read my book, or rather this was keeping me from reading and that hassled me. I really resented it a lot. With a guy like this, he has helped me a lot before for which I’m grateful. He’s just in his own world… insensitive to others’ needs. And you know he wouldn’t understand if I told him. He would be so sad; you know it was kind of a hassle”. The episode is complete. The tailor-made descriptions can be described as follows: W- to be free of unwanted visitor. RO - “He wouldn’t understand, he would be so sad” and RS - I feel hassled, resentful, guilty and compelled to suffer his presence.
The CCRT standard categories have been empirically chosen from the most frequently used ones, resulting in a standard category list. The third edition is the most widely used. Cluster analysis of the lists of categories resulted in the creation of “clustered standard categories” consisting of eight each for the different components, W, RO and RS (20) (pp. 43–54). From the cited quotation above the following clustered standard categories can be chosen: W: To be distant and avoid conflicts, RO: Upset, RS: Oppose and hurt others (see Table 2).
In an effort to fit in a CCRT pattern into a circumplex model of interpersonal patterns, Crits Christoph et al. (21) developed Quantitative Assessment of Interpersonal Themes (QUAINT). They tried to organise the clusters according to Benjamin’s (22) SASB. A study (23) ended up in 30 W, 31 RO and 40 RS. Most categories of interpersonal behaviour could be rated reliably. However, the study showed poor interrater reliability in some items, mostly among negative wishes.
The original CCRT scoring method is labour intensive and time consuming. First, researchers transcribe therapy sessions or RAP interviews, then independent judges identify REs, and other judges identify the categories in each component and count them to determine the most frequent categories. The categories are ranked according to frequency.
Luborsky and Crits-Christoph’s (11) original work showed an agreement between two raters, scoring 35 cases and reported weighted kappa values ranging from 0.61 to 0.70 in pair comparisons. Since then many studies have been performed with varying levels of reliability. See some examples in Table 1.
Zander et al (25) scored CCRT directly from a videotaped interview. The results were presented as unweighted kappa, with the following components: W = 0.35, RO = 0,41, RS = 0.46; The study has been criticised by Luborsky and Diguer (24), partly because of its use of kappa instead of weighted kappa which is used in most of the reliability studies on CCRT.
The results indicate that the original CCRT method can be used in research, but it is probably too labour intensive to be used in ordinary clinical practice, supervision or education.
The traditional CCRT method has shown itself useful in developing case formulations. Though, it is elaborate and time consuming. The present study aimed to test a simplified method. We investigated rater agreement in a method to establish a CCRT pattern directly from a Dynamic Interview (DI) without transcribing the material.