Beginning therapy: clinical outcomes in brief treatments by psychiatric trainees

Abstract

Aims and Method

Psychotherapy provided by inexperienced psychiatric trainees was assessed through comparison of post-treatment outcomes with a brief psychodynamic therapy and a brief integrative psychotherapy. A retrospective case–control design was applied to all patients seen by nine senior house officers (SHOs) during 6-month placements who provided follow-up data. Matching linked each patient offered simple cognitive–analytic therapy to one who had received brief psychodynamic therapy from the same SHO. Post-treatment changes were analysed by treatment received and differences between trainees were explored.

Results

Seventeen cases in each group met entry criteria. Patients receiving the integrative treatment had more severe pathology and outcomes were more variable when this model was used. Eight of the nine SHO therapists performed better with the psychodynamic intervention.

Clinical Implications

Significant clinical improvements can be achieved by inexperienced psychiatric trainees providing brief psychotherapy under supervision. Patient selection is undoubtedly important. Fundamental psychotherapy competencies need to be mastered before structured hybrid models are attempted.

The College recommends that, during general professional training, senior house officers (SHOs) in psychiatry see more cases in brief, time-limited therapy than in longer-term work, and that experience is obtained in more than one model of brief therapy (Royal College of Psychiatrists, 2002). Despite their medical proficiency, trainees may be highly inexperienced as therapists. Existing evidence of the efficacy of a therapeutic model is not necessarily a guide to its clinical impact in this learning situation. However, some models of therapy have been so widely used among trainee psychiatrists that their work has contributed to evidence of a model's effectiveness (e.g. Brockman et al, 1987; Guthrie et al, 1999).

The recommendations that trainees treat at least one patient for 6 months using psychodynamic, cognitive-behavioural and integrative models say little about the order of training experiences that makes the most educational sense. Although such early cases may well be discounted from formal efficacy studies, their outcome remains of critical interest to the patients concerned and to the service in which training is taking place. Trainees need to learn a number of general therapeutic skills, such as establishing and maintaining the external frame of therapy, building up a working alliance with a patient, encouraging disclosure of warded off thoughts and feelings, and developing links and hypotheses as their understanding progresses. The best vehicle for laying this foundation remains unclear.

Of the brief therapy methods in common use, cognitive-analytic therapy (CAT; Ryle, 1991) has a number of features to recommend it to anyone approaching psychotherapy from psychiatry (Rees, 2000). For instance, CAT maintains an implicit diagnostic model (through identification of ‘target problem procedures’) for which detailed history-taking and questioning are required. Two written summaries have to be produced in the course of the treatment and shared with the patient. Treatment follows a relatively clear plan, and the therapist is expected to be active in devising and proposing tasks for the patient to undertake between sessions. These structural features may reduce some patients’ anxiety and promote their engagement in therapy. Their relative familiarity for trainee psychiatrists means they could also have a containing effect when the therapist may be no more familiar with the situation, nor less anxious about it, than their patient. Evidence for this could include patients fulfilling their contract to complete treatment, as well as making positive and clinically significant changes when these are independently assessed.

Comparison of the impact of therapy conducted on traditional psychodynamic lines, v. therapy following the protocols of CAT, is a first step in examining the relative value and timing of different training experiences. The working hypothesis was that inexperienced psychiatrist therapists would obtain better outcomes, individually and collectively, following a structured CAT model, than in time-limited psychodynamic therapy.

Method

Therapists (five men; four women) were general psychiatry trainees, undertaking a 6-month training attachment at the same specialist psychotherapy unit. None had undertaken psychodynamic psychotherapy with more than one patient previously, and none subsequently went on to train as a specialist in psychotherapy. All had received at least 6 h of class teaching in psychodynamic and cognitive-behavioural therapies prior to starting, and further personal teaching in psychodynamic and CAT models was provided through the attachment. Only two senior house officers (SHOs) were known to have received personal therapy at some point during the attachment. Each had a personal case-load of patients being treated either by CAT or by brief psychodynamic therapy (BPT) over the same 6-month period. All patients had previously been assessed by an experienced psychotherapist and, on the basis of a provisional formulation, were thought likely to benefit from brief individual psychotherapy. All were told at the outset of treatment this was likely to continue for around 5 months of weekly individual sessions. Patients treated with CAT always completed a ‘psychotherapy file’, undertook some self-monitoring of target problems and other homework, received a written ‘reformulation’ summarising problematic procedures around 4 weeks into their therapy, and received a final ‘goodbye letter’ from the therapist. None of these features was present in the brief psychodynamic therapies. Instead, treatment was also planned to end at a set date, but without phasing of the treatment, use of written tools or prescription of tasks between sessions. A focal dynamic theme would be agreed from the outset of therapy, usually based on the initial assessment, and trainees were encouraged to pay close attention to patients’ feelings and how they felt themselves during sessions, as well as to the quality of their relationship with their patients. Use of interpretation would vary from one trainee to another. All therapies of either kind were discussed with the same supervisor (C.M.) in weekly one-to-one meetings for the duration of treatment. Although C.M. was very committed to the CAT model, and expected trainees to do well with it, his prior experience of working psychodynamically with patients in psychiatric settings was 10 years longer than his experience of using CAT.

At assessment, all patients completed ratings of their affective distress on the 12-item version of the General Health Questionnaire (GHQ-12; Goldberg & Williams, 1991) and impact of the presenting problem on each of five 10-point locally devised analogue scales, namely: disability; distress; quality of relationships, ability to use leisure time (‘play’) and capacity to work (‘work’). Three months after the conclusion of therapy, all patients are sent a brief follow-up questionnaire in which all these measures are repeated as part of standard audit procedures. Patients having CAT receive this independently of an invitation to have a follow-up interview. In-service audit shows around 60% of all treated patients return this questionnaire, the prospects of return being predicted neither by pre-treatment patient factors nor therapists’ subjective assessment of outcome. (This attrition is consistent with a published figure of 41% for routine 6-month postal follow-up of a large UK sample sent a longer battery to complete; Lucock et al, 2003.)

A list was compiled of the 46 brief cases seen by the nine SHOs for whom this follow-up information was available. These were subdivided into those receiving CAT and BPT. Where numbers receiving either treatment were the same for a given doctor, all cases were entered into the analysis. Where there was an imbalance, case-matching was used to pair CAT and BPT cases across four domains. Four clinical features were weighted according to a formula determined in advance of any examination of the data: diagnosis (50%), duration of patient's difficulties (20%), perceived disability (20%) and number of previous treatments (10%). The number of paired cases entered for a given trainee was therefore equal to the subgroup with fewest outcome reports. Each trainee is represented by between one and three patient pairs.

The two matched groups were then compared for initial demographic and clinical differences. Clinical change was expressed as the difference between outcome and initial measures of symptomatology and functioning. Group differences between the two therapy models were compared before the potential significance of confounding factors across the sample was crosschecked using non-parametric statistics.

Results

The patients whose therapy was included for analysis on this basis are shown in Table 1. Despite the initial decision to match cases on diagnosis before other factors when a selection needed to be made, this was the most evident, but non-significant, difference between the two groups: χ²=3.29 (2 d.f.); P=0.19. Therapies were intended to last around 16 sessions on average in either group because of the length of the training post. Most resulted in fewer attended sessions than this: only two therapies in either group were recorded as exceeding 16 sessions. At first glance, therapy was comparably successful in the two groups. Only one patient from each set was recommended for further treatment at the conclusion of these brief therapies. Specific changes are summarised in Table 2.

Table 1. Patient demographics¹

	BPT (n=17)	CAT (n=17)
Male gender	8	10
Age, years: mean	39.9	42.2
Duration of presenting problem, years: median Primary diagnosis, n	72	102
Affective	7	5
Personality	5	10
Other	5	2
Other therapy sessions, n: mean	13.1	13.3
Completing fewer than six sessions, n	2	1
Experience of previous therapy, n	10	10

Table 2. Overall clinical outcome following BPT and CAT

	BPT	CAT
Change in GHQ—12 (mean initial GHQ*)	6.18 (7.35)	3.89 (9.18)
Change in distress (mean initial score)	3.18 (6.47)	3.42 (7.88)
Change in disability (mean initial score)	2.76 (5.76)	2.12 (6.29)
Change in work (mean initial score)1	0.53 (6.47)	1.25 (6.18)
Change in quality of relationships (mean initial score**)1	1.25 (6.31)	2.59 (3.41)
Change in play (mean initial score*)1	1.87 (5.25)	2.71 (2.82)

The inter-group differences on these initial measures were greatest for quality of relationships (W=207.5; two-tailed P=0.003); play (W=222.5; P=0.015); and initial GHQ score (W=239.5, P=0.043).

The GHQ-12 is distinct from the other measures in having a commonly observed threshold of clinical caseness at 2-3 (Goldberg et al, 1997), scores tending to be bimodally distributed to either side of this. This permits a robust classification of clinical outcome according to the starting and finishing points (see Table 3).

Table 3. Classification of clinical outcome according to GHQ—12

	Initial GHQ—12 less than 3	Initial GHQ—12 3 or more
Final GHQ—12 less than 3	Subclinical	Good
Final GHQ—12 3 or more	Poor	Unchanged

On this classification, the 34 cases in the study showed outcomes as follows: good=19 (56%); unchanged=11 (32%); subclinical=2 (6%); poor=2 (6%). Two patients in each treatment group started below the clinical threshold. Both of those in the CAT group became cases by follow-up (‘poor’), whereas both of those in the BPT group remained non-cases (‘subclinical’). Of the 11 who were effectively unchanged because they maintained scores in the clinical range, 2 were in the BPT group and 9 received CAT. Comparisons between the outcomes of patients with individual trainees are complicated by the distribution of the 17 patient pairs: 3 trainees are represented by a single pair each; 4 by two pairs and 2 by three pairs of patients. However, the 9 ‘unchanged’ CAT cases were distributed evenly among 8 of the 9 trainees, all but one trainee having greater clinical success with BPT across their patient pairs.

Discussion

This study, which began prior to routine use of the CORE outcome measure (Evans et al, 2002), is constrained by the reduction of data necessitated by attrition of follow-up data and the measures employed to assess outcome. It is not a valid comparison of the efficacy of the two treatments here, given the underlying clinical differences between the two samples. At the outset, patients receiving CAT were more severely disabled than those receiving BPT. These initial differences between the groups, which persist despite case-matching, suggest a bias in assessors’ recommendations: treatment with CAT was preferred for patients with more severe difficulties and those with more evidence of personality disorder.

The study does confirm that, in a majority of cases seen under ordinary conditions who provide follow-up data, brief psychotherapy by relatively inexperienced psychiatric trainees is associated with significant positive clinical change. When working with a model that emphasises basic therapy skills (such as establishing a good working relationship in a stable setting; sustained attention to affect and meaning) to the exclusion of additional structured tasks (such as sharing written formulations), every trainee in this group of inexperienced therapists was leaving nearly every patient with clear gains. The hypothesis that trainees might do better with a model whose characteristic tasks seemed superficially closer to the routines of psychiatric assessment was not supported. Instead, first efforts in an approach placing greater relative emphasis on listening and feeling over questioning and summarising were nearly always more successful. It is possible that BPT might have offered relatively more containment through trainees’ relatively greater familiarity with its methods and rationale, whereas the tasks and style of CAT may have weakened this containment when the two methods were practised simultaneously by these beginner therapists. Unfortunately, no systematic data concerning the SHOs’ own observations and opinions were collected: it will be important for future comparative studies to include such qualitative data when evaluating hypotheses of this kind. The importance of mastering generic therapeutic skills in order to undertake successful CAT (and other structured therapies) has been recognised in recent work on therapist competencies (Bennett & Parry, 2004) and future studies will also benefit from assessment of trainees’ developing competencies independent of measurement of clinical outcomes.

In practice, trainees may acquire experience in integrative therapies such as CAT under the SHO guidelines in several ways and these will vary in their demands. Our findings suggest that training experiences that pay most attention to common, transferable psychotherapy skills, such as the BPT described here, are best provided before work with more derivative models is undertaken.

Declaration of interest

None.