آموزش روانی برای مالیخولیا: مقایسه یک رویکرد شناختی رفتاری و یک رویکرد حل مسئله

Abstract In this study, two 6-week psychoeducational courses for hypochondriasis are compared, one based on the cognitive-behavioural approach, and the other on the problem-solving approach. Effects of both courses on hypochondriacal complaints, depression, trait anxiety, and number of problems encountered in daily life, are measured pre-treatment, post-treatment, and at 1- and 6-month follow-up. Participants (N=48N=48, of whom 4 dropped out), suffering from DSM-IV hypochondriasis, were randomized into one of the two course conditions. Results showed beneficial effects of both courses. Few differential treatment effects were found: in both conditions all effect measures decreased significantly over time (p<0.01). However, between- and inter-individual variability in decrease-patterns was of considerable size, leading to large deviations from the mean pattern. Acceptability and feasibility of both courses were rated highly by their respective participants. It is concluded that both courses can be considered equally beneficial and effective over time, with the effects evident immediately after treatment and maintained over the follow-up period.

Introduction Hypochondriacal patients suffer from the fear or conviction of having a serious physical disease. This fear or conviction is based on the misinterpretation of bodily symptoms (APA, 1994). The consensus among practitioners used to be that these patients were very difficult to treat with psychological interventions. Recently, this view has changed, and several studies suggest the effectiveness of cognitive and/or behavioural interventions (Barsky & Ahern, 2004; Visser & Bouman, 2001; Warwick, Clark, Cobb, & Salkovskis, 1996; for a recent overview see Taylor & Asmundson, 2004). Cognitive-behavioural therapy (CBT) for hypochondriasis is usually based on a cognitive model (Warwick & Salkovskis, 1990), which focuses on the various concepts that seem to maintain or even bring about hypochondriacal complaints: the misinterpretation of bodily symptoms, anxiety, selective attention for bodily sensations, and checking and/or avoidance behaviour. The treatment goal is a change in hypochondriacal cognitions and behaviour. Another form of treatment with known beneficial effects is psychoeducation, which is among the most effective of the evidence-based practices that have emerged in both clinical trials and community settings (Lukens & McFarlane, 2004). Historically, psychoeducation has been described as the teaching of personal and interpersonal attitudes and skills which the individual applies to solve present and future psychological problems (Guerney, Stollak, & Guerney, 1971). It regards people who seek help as ‘participants’ rather than as ‘patients’ or ‘clients’, and ‘therapists’ as ‘teachers’. Its original goal is to move participants away from the medical model. Furthermore, psychoeducation reflects a paradigm shift to a more holistic and competence-based approach (Marsh, 1992). The psychoeducational format has often been combined with CB theory, and is usually disseminated in the form of short-term, focused courses, aimed at people who function relatively well, to teach them about their disorder. Barsky, Geringer, and Wool (1988) were the first to propose a psychoeducational course for hypochondriasis, and their suggestion was followed by several others (Avia et al. (1997) and Avia et al. (1996)). The course developed by Barsky et al. (1988) is a cognitive-educational treatment, consisting of group training on the perception and interpretation of physical symptoms. It comprises six weekly meetings, during which six to eight patients receive information about factors that can enhance or prolong somatic problems, such as cognition and symptom attribution, and dysphoric affect (Barsky et al., 1988). Stern and Fernandez (1991) found the treatment, in a group of six participants, to be successful in reducing complaints such as medical consultations and time spent thinking about disease. They did not find a significant decrease in measured anxiety and depression parameters. Avia et al. (1997) implemented the course in Spain, with modified examples, exercises and therapeutic homework. They reported beneficial effects in a group of 17 students, of whom only eight actually suffered from DSM-III-R hypochondriasis. After making considerable adaptations, Bouman applied the course in The Netherlands. This community-based course was studied in an uncontrolled trial (Bouman, 2002), and in a waiting list-controlled trial (Bouman & Polman, submitted). A total of 27 DSM IV-diagnosed hypochondriacal participants (APA, 1994) were included in the first, and 53 in the second study. The results support the notion that this programme leads to significantly reduced hypochondriacal complaints, depressive complaints, medical services utilisation, and trait anxiety. These improvements were maintained at six months follow-up. In the waitlist-controlled study (Bouman & Polman, submitted), the course also outperformed the passage of time. So far, psychoeducation seems to be successful in mitigating hypochondriacal and comorbid complaints in hypochondriacal populations. Although most studies mentioned earlier used small sample groups without control groups, a case has been made for the internal validity of psychoeducation. Moreover, psychoeducational courses have been proven to outperform mere passage of time (Bouman & Polman, submitted). However, little is known about its construct validity, i.e. the question whether the relation between the intervention and behaviour change is due to the construct given by the investigator (Kazdin, 1998). In the context of individual treatment, the question of construct validity was studied earlier by Clark et al. (1998). They compared individual CB treatment to individual behavioural stress management, finding both approaches to be equally powerful in reducing hypochondriacal complaints at 1-year follow-up of both treatments. To seek an answer to the question of construct validity, we decided to compare the CB psychoeducational group treatment with problem solving (PS), delivered in a similar format. The PS course was specifically designed for the purpose of this study. Its content was based upon the social PS approach (D’Zurilla & Nezu, 1999) that involves the application of four major PS skills: problem definition and formulation, generation of alternative solutions, decision making, and solution implementation and verification. The PS approach used in this study is model-based, structured, and directive, to ensure its format to be similar to that of the CB-course's, and have them differ only in specific content (see Method for more details). Our main reason for choosing PS is that this approach puts hypochondriacal complaints into a broader context. All aspects of life, including possible comorbid depression, anxiety, and relationship problems, can be considered in PS treatment, not just hypochondriacal complaints. These problems are thought to play a maintaining and antecedent role, and once they are reduced, this is assumed to have a positive effect on hypochondriasis. It should be noted that we did not aim to test the PS model per se, but only its approach in a psychoeducational framework. Therefore, in this study it is hypothesized that the psychoeducational approach in itself has beneficial effects over time, implying a significant improvement on effect measures for both the CB- and the PS-course. In addition, the CB-course is expected to lead to a greater reduction of hypochondriacal symptomatology, because of its more specific focus on this disorder.

Results Missing data Missing data occurred in this study: 14 (63.6%) of the 22 completers in the CB-condition returned all four assessments and 19 (81.8%) of the 22 completers in the PS-condition did so. Firstly, those who have not returned all their questionnaires were compared with those who have returned all questionnaires, with regard to their post-assessment by means of t-test. The comparisons were made for the four outcome measures: GIAS, BDI, STAI, and PAQ. These tests showed that both groups did not differ significantly at post-assessment (GIAS: t=1.4, p=0.17; BDI: t=−0.6, p=0.56; STAI: t=0.1, p=0.91; PAQ: t=0.04, p=0.97). Secondly, it was studied whether they who had not returned all questionnaires of the CB-group differed from they who had not returned all the questionnaires of the PS-group. When analysed with a Mann–Whitney test, it was found that these two groups did not differ either on any of the outcome measures (GIAS: Z=−0.2, p=0.83; BDI: Z=−1.4, p=0.2; STAI: Z=−1.3, p=0.18; PAQ: Z=−0.3, p=0.73). These results should be viewed with caution, because 8 participants had not returned all measurements in the CB-group, versus 3 in the PS-group, which makes comparison difficult. Outcome of the multilevel analyses Results of the multilevel analyses are shown in Table 2. Because preliminary analyses showed that none of the biographical variables (age, gender, and level of education) had a significant effect, they were not included in the descriptions of the multilevel analyses, or in Table 1. Table 1. Multilevel models for the development of the GIAS, the BDI, the STAI, and the PAQ over time and between conditions Fixed effects GIAS BDI STAI PAQ Estimate SE t Estimate SE t Estimate SE t Estimate SE t Intercept (mean score at t1) 92.16 4.07 14.48 1.30 54.86 1.63 79.14 2.75 Mean difference at t2 (vs. t1) −20.17 3.50 −5.76*** −4.09 0.87 −4.70*** −6.12 1.38 −4.43*** −2.20 1.72 −1.28 Mean difference at t3 (vs. t1) −25.46 3.77 −6.75*** −5.10 0.92 −5.54*** −7.23 1.47 −4.91*** −6.06 1.85 −3.28** Mean difference at t4 (vs. t1) −29.72 3.90 −7.62*** −6.00 0.96 −6.63*** −6.63 1.55 −4.28*** −9.98 2.91 −3.43*** Treatment (PS vs. CB)difference at t1 −4.96 8.15 −0.62 0.32 2.59 0.12 4.00 3.26 1.23 2.91 5.50 0.53 Treatment (PS vs. CB)difference at t2 10.75 7.00 1.54 3.27 1.75 1.87 6.59 2.75 2.40** 8.33 3.45 2.42** Treatment (PS vs. CB)difference at t3 10.71 7.54 1.42 0.76 1.85 .41 3.54 2.94 1.20 −.37 3.70 −0.10 Treatment (PS vs. CB)difference at t4 4.10 7.81 0.52 −.089 1.91 −.046 −5.14 3.10 −1.66 .076 3.90 0.019 Random effects χ2 Between individual variance 466.69 117.06 31.03 7.96 76.17 18.97 268.55 61.56 Additional variance at t1Covariance 4.91 6.25 9.1 11.96 4.91 Measurement variance 264.51 35.66 14.10 2.39 40.91 5.51 64.21 8.66 Note. CB=Cognitive-behavioural group; PS=Problem-Solving group; t1=pre-treatment assessment; t2=post-treatment assessment; t3=follow-up at 1 month; t4=follow-up at 6 months; SE=Standard Error; *=p<.05; **=p< .01; ***=p<.001. Table options Hypochondriacal complaints In the analysis, the total-score on the GIAS is implemented, because a preliminary analysis showed that the four subscales described in the method section displayed a similar pattern of decrease over time. Table 1 shows a substantial decrease in hypochondriacal complaints between assessment 1 and 2 (t=−5.8, p<0.0001). Scores on the GIAS decrease further at assessment 3 and 4. Differences between the courses are non-significant at all times of assessment. The between-individual variance of the random effects (466.69, implying a standard deviation of almost 22 points) demonstrates that the differences in mean scores of all participants are considerable, and of approximately the same size as the mean improvement. The measurement variance (indicating differences over time within participants) is smaller (264.51), with a standard deviation of approximately 14), but also considerable. Depressive complaints Table 1 indicates that, between assessments 1 and 2, the mean score of the BDI drops significantly (t=−4.7, p<0.0001). At assessments 3 and 4, the scores decrease further. For condition, no significant interaction effects were found, indicating that both courses perform equally well. The measurement variance at the first assessment is somewhat larger than at the later time points (approximately 20 instead of 14). Again, the between-individual variance is larger (31.03) than the measurement variance (14.89). Trait anxiety Table 1 shows a significant decrease in scores on the STAI between time of assessment 1 and 2 (t=−4.4, p<0.0001). A significant difference in decrease between the two groups was found at assessment 2, (t=2.4, p=0.01), indicating that the complaints in the CB-group on average have significantly decreased more, immediately after the course, than those in the PS-group. The between-individual variance (76.17) is larger than the measurement variance (40.91) and its standard deviation of more than 8 larger than the mean improvement. Number of problems experienced in daily life Table 1 shows that, between assessments 1 and 2, scores decrease slightly, but not significantly (t=−1.3, p<0.15). However, they have decreased significantly at time 3 (t=−3.3, p<0.002), with reference to time 1. This decrease has continued at time 4. A significant difference between the groups was found at assessment 2 (t=2.4, p=0.01), indicating that the PS-group participants on average experience more problems in daily life right after the course than the CB-group participants do. The between-individual variance for this measure is quite large (268.6), whereas the measurement variance (64.2) is relatively small. Inter-individual differences The between-individual variance, described above and shown in Table 1, indicates there are large differences between participants on all measurements. These differences are illustrated in Fig. 2. Individual differences between male and female participants on the four times of ... Fig. 2. Individual differences between male and female participants on the four times of assessment of hypochondriacal complaints, depression, trait anxiety, and problems experienced in daily life. Note. GIAS=Groningen Illness Attitude Scale; BDI=Beck Depression Inventory; STAI=Spielberger State-Trait Anxiety Inventory; PAQ=Problem Area Questionnaire; week 0=pre-treatment assessment; week 6=post-treatment assessment; week 10=1-month follow-up; week 36=6-month follow-up. Figure options The inter-individual differences are of such size that in spite of an average decrease over time of all measures, it is possible that an individual does not show an improvement at all or even shows an increase of complaints. Effect sizes within groups The differences within the groups (when contrasting pre-treatment with post-treatment, first follow-up and second follow-up) are further illustrated by showing their effect sizes (Cohen's d) in Table 2. Table 2. Within-group effect sizes (Cohen's d) of all outcome measures at the different times of assessment Post-treatment vs. pre-treatment One month follow-up vs. pre-treatment Six months follow-up vs. pre-treatment CB PS CB PS CB PS GIAS 1.01 0.54 1.05 0.73 1.21 1.09 BDI 0.78 0.29 0.67 0.58 0.74 0.81 STAI 0.90 0.29 0.83 0.55 0.55 0.84 PAQ 0.35 0.01 0.18 0.31 0.62 0.62 Table options All within CB-group effect sizes, except for the PAQ, are medium to large, and consistently larger than the effect sizes within the PS-group. Exceptions are the effect sizes for the BDI and STAI within the PS-group at the second follow-up. Attendance and evaluation of the courses Both programs were high in acceptability, were attended equally well, and were largely evaluated equally positive. The mean attendance rate was 90.8% (range 77–100%) for the CB-group, and 88.6% (range 68.2–100%) for the PS-group. Immediately after the sessions, CB-completers award the separate sessions a mean score of 8.3 out of 10 (range 8–10, SD=0.58), and PS-completers rate theirs with a mean of 7.8 (range 6–9, SD=0.75), ranging from 1, meaning ‘very bad’, to 10 meaning ‘excellent’. When analysed with a t-test, this is a small, but significant difference (t=2.4, p=0.02, d=0.58, 95% CI=0.076–0.89). When asked retrospectively, CB-participants awarded the entire course a mean grade of 7.9 (SD=0.99) out of 10, PS-participants rated theirs on average a 7.2 (SD=1.9) out of 10. A t-test shows that this difference is not significant (t=1.4, p=0.16, d=0.59, 95% CI=−0.27–1.6). Furthermore, the participants were asked after the first session to rate how much they expected to benefit from their particular course. On a scale of 1 (=not at all) to 10 (=very much), the CB-participants awarded a mean score of 7.3 (SD=1.4), and the PS-participants awarded a mean score of 7.5 (SD=1.2) (t=−0.50, p=0.62, d=−0.18, 95% CI=−1.01–0.61). Clinical significance analyses The RCI (see Fig. 1) is used in the present study to determine reliable change for participants in both conditions, with regard to the GIAS. Results have been computed with the use of the Cronbach's alpha of the GIAS at pre-test within this group (α=0.95). Results show that at post-assessment, 16 (72.7%) participants of the CB-group have achieved reliable change, as have 6 (28.6%) participants of the PS-group. However, at follow-up at 6 months, 8 (57.1%) of the 14 participants returning this questionnaire of the CB-group score within the range of reliable change, whereas 12 (63.2%) of the 19 participants who have returned this questionnaire of the PS-group have achieved reliable change at this point. As a second way of determining clinical significance, mean scores of participants of both groups were compared to the mean scores of both a community sample norm group, and a patient norm group. CB- and PS-group were taken together, because the groups did not differ significantly on any of the assessments. The mean scores of the norm groups on the GIAS, as reported by Visser (2000), are: 30.5 (SD=25.3) for the community sample, and 101 (SD=25.8) for the patient norm group. Results show that at pre-assessment, the participants of the study differ significantly from the community sample (t=14.3, p<0.00), and differ significantly from the patient norm group (t=−1.9, p<0.05), which places them in between both norm groups, with more resemblance to the patient norm group. At 6-month follow-up, it is clear that the participants of this study do not score within the range of the community sample (t=7.7, p<0.00), but gradually over time, they have started to differ more from the patient norm group (t=−8.3, p<0.00). In conclusion, it is clear that clinically significant improvement is achieved, in terms of reliable change for a substantial number of participants, but not in terms of the participants scoring within the range of a community sample. Furthermore, these results should be interpreted with caution because of the missing data occurring in this study.