Abstract
Depression is among the most common psychiatric disorders of adults, with approximately 10–20% of the population experiencing a major depressive episode at some point in their lifetime [1, 1, 2]]. Data from the Epidemiologic Catchment Area (ECA) program of the National Institute of Mental Health (NIMH) indicates a 1-month prevalence rate of 2.2% and a lifetime prevalence rate of 5.8% for major depressive episodes [3]. The probability of recovery within 1 year was 70%; of those who did not recover in the first year, the probability of remaining chronically ill for at least 5 years is 11.5% [4]. Depression not only poses as a serious health problem to many people, owing to its morbidity, mortality and prevalence [5, 5, 6]], but also entails a high economic cost to society, due to the lost productivity of workers [7].
The need to treat mood disorders effectively is important, and a wide range of psychological and biological interventions is available for the treatment of depression [8]. Cognitive-behavioural therapy (CBT) and pharmocotherapy are effective interventions for depression [09, 9, 10]]. In clinical settings, when depressed patients undergo therapeutic treatment, many of them are already on medication [11]. At present, therapists who use CBT to treat depressed patients do not know whether these pre-existing medications will have an effect on the therapy that they are providing. Although there has been no lack of positive outcome studies in the literature on the treatment of depression, they have been mainly laboratory-orientated, with controlled medication trials. Thus, the status of the unmanipulated pre-existing nature of these medications on CBT in a clinical setting has not been established sufficiently. The purpose of this study is to fill this gap in the research, by investigating whether the prescribed medication that depressed patients are on will have an influence on CBT. It asks whether pre-existing antidepressant medication for patients undergoing group CBT for depression enhances or attenuates the efficacy of the therapy. The answer to this question is of clinical utility to the therapist, as precautionary measures can be taken if the result proves to be negative.
Several studies have investigated and demonstrated the efficacy of CBT for depression [09, 9, 12]]. It is generally accepted that the efficacy of CBT is equal to pharmacotherapy and interpersonal therapy. However, investigations focusing on the efficacy of combined CBT-pharmacotherapy, compared to either pharmacological treatment or CBT alone, have produced mixed results. Although some researchers have found combined treatment to be superior to pharmacotherapy alone, others have not [13, 14, 15]]. At present there are some indications from the literature that combined cognitive-pharmacotherapy might be superior to either single modality alone [16, 17, 18]].
The issue of whether pharmacotherapy enhances or attenuates the efficacy of CBT still needs clarification. This is particularly so in clinical settings where patients on pre-existing medication seek CBT treatment. It is appropriate to look at what impact this pre-existing medication may have on CBT as an intervention for depression. The focus of this study has clinical relevance for therapists, as the influence of these pre-existing medications which the patients bring to the CBT is largely unknown. One study has looked into this issue directly, using patients with panic disorder and agoraphobia [19]. The authors compared groups who were on group CBT alone, on antianxiety drugs plus CBT, on antidepressants plus CBT, and on antianxiety drugs and antidepressants plus CBT. The finding was that their pre-existing medication regimes had no significant positive or negative effect on CBT, even at long-term follow-up. Whether this finding can be generalized to depressed patients is as yet unknown. The present study addresses this gap in research by focusing on depressed patients.
In clinical practice, suitable patients are accepted for CBT regardless of their medication status. Their medication regimes are under the supervision of their medical doctors, and not within the control of the non-medically qualified therapist providing the CBT. To simulate this typical clinical situation, a quasi-experimental research approach was chosen in favour of an experimental research paradigm in this study (i.e. the specific types of medications that the subjects were on, or the dosage levels, were not manipulated or controlled for). Similarly, random assignment of subjects is not used. Instead, their status with regard to medication at the point of entry, duration and exit of the group CBT program determined their assignment to either the CBT group (CBT) or the CBT plus pre-existing medication group (CBT-M). Thus, patients who were on preexisting medication and who underwent group CBT would be compared to patients who were not on medication but who underwent group CBT. This approach is similar to the earlier work by Oei et al. [19]. It was hypothesised that the addition of preexisting pharmacotherapy to group CBT would enhance the effect of group CBT in the treatment of depression.
Method
Participants
Seventy-one non-psychotic, non-bipolar depressed outpatients were recruited from a local community by a media release asking for persons suffering from depression, and from general referrals by medical practitioners. Patients who were in this study met the following criteria: a diagnosis of major depression, current episode, following a diagnostic interview using the Structured Clinical Interview for DSM-III-R (SCID); attended the group CBT program at the psychology clinic; and informed consent. Patients were excluded from the study according to the following criteria: bipolar affective disorders or other major psychiatric disorders (e.g. schizophrenia, organic brain syndrome, or antisocial personality); alcohol or drug abuse; or any person with reading difficulties or who was not fluent in English.
The patients were allocated to one of two treatment groups on the basis of their self-reported, pre-existing medication status during the assessment phase conducted prior to their participating in the CBT program, as well as assessments during and at the end of the therapy. These two treatment groups included the CBT without medication (CBT) group and the CBT plus anti-depressant medication (CBT-M) group. Forty-six subjects were allocated to the CBT group, which comprised patients who reported not taking any antidepressant medication at the beginning and at the end of the program. Twenty-five subjects were allocated to the CBT-M treatment group, which comprised patients who reported being on antidepressant medications at initial assessment prior to undergoing the group CBT program, as well as during and at the end of the program.
A mean age of 42.5 years (SD = 12.2 years) and a mean education of 12.5 years (SD = 1.8 years) was found for the patients in the CBT treatment group. The mean number of weeks of therapy attendance was 10.8 (SD = 2.5) for this group. In the CBT-M group, the mean age was 39.9 years (SD = 12.0 years), the mean education was 12.3 years (SD = 2.6 years) and the mean number of weeks of therapy attendance was 11.9 (SD = 0.7 week). Chi-squared analyses showed that there were no significant differences across the treatment conditions with regard to sex and marital status. Similarly, t-tests showed that there were no significant differences in age, education and duration of therapy attendance between the two groups.
Data collected for the purposes of the present study were gathered during the pre-treatment and post-treatment phases of the group CBT program. Persons who responded to the media release went through a telephone screening interview prior to being included for a diagnostic interview using the Structured Clinical Interview for DSM-III-R (SCID). The patients were required to complete the battery of questionnaires as described below, prior to the commencement of therapy, as well as during and upon completion of the group CBT program. This is a 12-session group psychoeducational program of cognitive therapy. The program was conducted weekly for 2 h per session for 12 weeks, and it consisted of mini-lectures, class exercises, guided reading and homework tasks. The CBT program has been shown to be efficacious in treating mood disorder patients [12].
Multiple outcome measures (e.g. Beck Depression Inventory, Zung Self-Rating Depression Scale) and process measures (e.g. Automatic Thoughts Questionnaire, Dysfunctional Attitude Scale and Hopelessness Scale) were used. Outcome measures refer to those questionnaires that assess the severity of depression. Process measures refer to those questionnaires that evaluate the cognitive processes that are believed to underlie depression.
Outcome measures
The Beck Depression Inventory (BDI) [20] is a 21 four-choice item inventory designed to assess affective, behavioural, cognitive, motivational, and vegetative aspects of depression. Each item has four options, ranked in order of severity, with scores ranging from 0 to 3 for each option. The person selects the option closest to his or her present state. The greater the score the greater the severity of depression. Spearman-Brown reliability is around 0.93 [20] and its test-retest reliability ranges from 0.48 to 0.86 for psychiatric patients and from 0.60 to 0.83 for non-psychiatric patients [21]. The BDI shows good concurrent validity when compared to psychiatric ratings of severity of depression in a clinical population.
The Zung Self-Rating Depression Scale [22] is a 20-item self-administered scale designed to measure the symptoms of depression. It involves ratings on a four-point qualitative, temporal scale. The split-half reliability is 0.73. The scale correlates 0.52–0.80 with the BDI.
The Daily Activity Rating Form is an analogue scale from 1 to 9, with 1 meaning ‘no activity’ and 9 meaning ‘extremely active’, upon which the person rates his or her level of activity. From this daily monitoring of activity, changes in pattern of involvement in activity can be seen. Empirically, involvement in activities is correlated with a non-depressed mood state.
The Daily Mood Rating Form was used as a measure of mood state. The person is to record what his or her mood is on a nine-point scale from 1 (‘very depressed') to 9 (‘very happy'). Mood-change in a positive direction can be a signal that automatic thoughts are diminishing, and such a rating scale is sensitive to small changes in mood.
Process measures
The Automatic Thoughts Questionnaire (ATQ) [23] is a 30-item self-report questionnaire that assesses the frequency with which respondents experience 30 depressotypic self-statements (e.g. ‘I'm no good') on a 1–5-point scale. Total scores can range from 30 to 150, and depressed patients tend to score in the 90–130 range. The items significantly discriminate between the depressed and non-depressed criterion groups. The split-half reliability coefficient was 0.97.
The Dysfunctional Attitude Scale (DAS) is a 40-item self-report questionnaire, with items assessing perfectionistic performance standards, rigid ideas about the world, and concern about the judgments of others. The respondents are asked to rate their level of agreement with each of the 40 statements on a seven-point Likert scale. Total scores can range from 40 to 280, and depressed patients tend to score in the 140–200 range. The DAS has good psychometric properties [12, 12, 24]].
The Hopelessness Scale (HS) [25] is a 20-item, true/false, self-report measure intended to tap the degree of the respondent's general pessimism. High scorers on this scale endorse such items as: ‘Things just won't work out the way I want them to’. Total scores range from 0 to 20, and depressed patients tend to score 8 or above. The HS has been shown to distinguish clinically depressed from non-depressed psychiatric patients.
Results
The mean scores and standard deviations for the two groups (CBT and CBT-M) on the outcome and process measures obtained at the pre-treatment and post-treatment phases are presented in Table 1, and a series of ANOVAS was carried out with each of these measures as the dependent variable. There were no significant treatment main effect, F 1,47 = 0.35, p = 0.559, or interaction effect, F 1,47 = 0.58, p = 0.449, in the BDI scores observed between the two treatment groups. However, the time main effect was significant, F 1,47 = 56.37, p < 0.0001, suggesting that the BDI scores of both the CBT and CBT-M groups dropped significantly from pre-treatment to post-treatment. There was no significant difference between the two groups in the rate of improvement.
Similarly, no significant main effect between the two groups, F 1,47 = 1.19, p = 0.280, or interaction effect, F 1,47 = 0.58, p = 0.449, was found on the Zung Self-Rating Depression Scale. The results suggest that the rate of change in the symptoms of depression was not different between the CBT and CBT-M groups. Statistically significant time main effect was found, F 1,47 = 47.16, p < 0.0001, demonstrating significant improvement on self-ratings of symptoms of depression from pre-treatment to post-treatment. The concurrent use of pre-existing medication with group CBT did not significantly enhance or detract from its post-treatment outcome on the Zung depression inventory.
Ratings of activity for each week (from daily monitoring) were totalled and divided by the number of days in order to give a weekly activity score. There was no significant treatment main effect, F 1,28 = 2.45, p = 0.129, or interaction effect, F 1,28 = 1.31, p = 0.261 for weekly activity score, but time main effect was significant, F 1,28 = 12.45, p < 0.001. Overall, the findings suggest that weekly activity score improved significantly for both the CBT and CBT-M groups from pre-treatment to post-treatment, but the rate of weekly activity improvement was not significantly different for the two groups.
Mean scores and (SD) on depression scores and cognitive process measures for the cognitive-behavioural therapy (CBT) and the cognitive behaviour therapy-medicated (CBT-M) group
Ratings of mood for each day (from daily monitoring) were totalled and divided by the number of days in order to give a weekly mood score. Neither the main effect between the two treatment groups, F 1,28 = 1.79, p = 0.192, nor the interaction effect, F 1,28 = 2.00, p = 0.168, was significant. There was, however, significant time main effect, F 1,28 = 25.86, p < 0.0001. The results suggest that there was improvement of mood from pre-treatment to post-treatment regardless of treatment groups, and the rate of improvement for the two groups did not differ significantly.
Similarly, the treatment main effect (F 1,83 = 1.19, p = 0.279) and the interaction effect (F 1,83 = 0.06, p = 0.800) were not significant for the automatic thoughts score. The time main effect was significant (F 1,83 = 113.94, p < 0.0001). The overall finding suggests that the scores significantly dropped between pre-treatment and post-treatment on the ATQ for both CBT and CBT-M groups, with no significant differences between the CBT and CBT-M groups on the rate of improvement.
As for the HS scores, neither the treatment main effect, F 1,47 = 0.83, p = 0.366, nor the interaction effect, F 1,47 = 1.68, p = 0.201, were significant. This suggests that there was no significant difference in the rate of improvement in hope found between the CBT and CBT-M groups. Main effect of time was not significant (F 1,47 = 6.66, p = 0.013) suggesting that there was no significant drop in HS scores following treatment in the two groups.
There was no significant treatment main effect (F 1,42 = 2.50, p = 0.121), or interaction effect (F 1,42 = 0.09, p = 0.121) in the dysfunctional attitude scores. However, time main effect was significant (F 1,42 = 13.27, p < 0.001), suggesting that the scores on DAS improved from pre-treatment to post-treatment for both the CBT and CBT-M groups. The rate of improvement was not statistically different between the two groups.
In summary, the analyses for both the outcome measures (BDI, Zung) and cognitive process measures (ATQ, DAS and HS) showed no statistically significant treatment effects or interaction effects. Similar results were found on self-ratings of activity level and mood. However, for all the above measures there was a significant time main effect, except for the Hopelessness scores. The conclusion drawn from these results is that both the CBT and CBT-M group demonstrated improvement with time, and the rate of improvement was quite similar. This suggests that the concurrent use of pre-existing antidepressant medication with group CBT did not significantly enhance or detract from the efficacy of group CBT. Thus, the main hypothesis of this study (i.e. that the addition of pre-existing pharmacotherapy to group CBT will enhance the effect of group CBT) is not supported. The concurrent use of drugs with group CBT did not significantly enhance or detract from the outcome of group CBT.
Discussion
The aim of the present study was to look at the impact of a pre-existing medication regime on the post-treatment outcome in the treatment of depression using a group CBT program in a community setting. The findings suggest that the pre-existing medication regime does not add or detract from the positive outcome of the group CBT program. Cognitive behaviour therapy appears to be just as efficacious in the treatment of depressed patients, whether they are on pre-existing medication or not. As the majority of referrals of depressed patients to psychologists in clinical practice in the community are already on pre-existing medication, psychologists have to contend with this medication regime when the patient entered into therapy. The implication of this finding is that they need not be excessively worried about the effect of pre-existing medication on the CBT that they are conducting. The result of this study is consistent with the finding by Oei et al. [19] in their study of patients with panic disorder with agoraphobia. They concluded that pre-existing anti-anxiety medication, or antidepressant medication, or combined anti-anxiety and antidepressant medication, did not significantly enhance or detract from the efficacy of the group CBT for these patients. These two related findings seem to suggest that psychotherapists need not be too concerned of the impact of pre-existing medication on the cognitive behaviour therapy that they are offering to their patients.
There are several possible explanations for the lack of significant impact of pre-existing medication on the post-treatment outcome of group CBT. First, it is possible that those patients allocated to the CBT-M group were not actually adhering to their medication regimes, as the problem of a patient's non-compliance with the drug therapy is not uncommon. The non-compliance would render the status of the patients in the CBT-M group comparable to the patients in the CBT group. However, this seemed unlikely, as the medication status was checked before, during and after therapy. It is also conceivable that the medication doses prescribed may have been suboptimal (i.e. their drug doses are not comparable with the recognised therapeutic dosage levels recommended in the literature). However, this can only be speculative, as data on the daily drug doses from patients in the CBT-M group were not collected in detail and thus were not available for comparison. What is relevant is that even if there is variation in medication compliance or dosage, it did not interact with CBT to produce a lesser effect. The concurrent use of medication has no discernible negative impact on CBT.
It could perhaps be argued that the patients who are in the CBT-M group are more severely depressed than the patients in the CBT group, as medicated out-patients represented a higher level of depressive impairment or symptomatology. Consequently, any additive significance of CBT plus pre-existing medication was masked, resulting in findings of equivalent efficacy in both treatment groups. However, the demonstrated lack of significant differences between the two treatment groups at pre-treatment on most of the dependent measures employed argued against the above possibility. Both the CBT and CBT-M groups were found to be associated with significant reductions in depressive symptoms in this study, with neither treatment proving superior to the other with regard to any of the several cognitive types of outcome. The results suggest that pre-existing medication plus CBT has a non-enhancement effect: combining the two effective interventions does not exceed that of a single therapeutic approach. This finding is consistent with the conclusions reached by meta-analyses of controlled outcome studies that CBT plus medication is just as effective as other treatment approaches for depression [9].
There are a few methodological limitations in the present study which may restrict its generalisability. The non-random nature in the present design does not allow for the elimination of important sources of bias. The individualised medication regime and lack of uniformity in medications and dosage could obscure the results of the study. Due to this lack of experimental rigour, findings from such quasi-experimental studies must be considered cautiously. However, it would be a mistake, to ignore or dismiss these findings on the less than rigorous methodological grounds [26]. The major aim of this study was to examine the impact of pre-existing medication in a clinical setting. Therefore, rigorous control is not central to the aim of this study. The present study should be viewed as clinically relevant and ecologically valid in its approach. The advantage of this is that it clarifies the impact of antidepressants that patients in a community setting are likely to be prescribed on the efficacy of CBT.
To conclude, although this finding supports the efficacy of CBT for depressed patients, whether they are on pre-existing antidepressant medication or not, caution is advised in making definite conclusions regarding the enhancement or attenuation of pre-existing medication on CBT for depressed patients, until further research can substantiate this finding.
Footnotes
Acknowledgements
This project was supported by a grant from the Australian Research Council. We would like to thank A. Lamberton for her help.