No evidence that CBT is less effective than antidepressants in moderate to severe depression

ABSTRACT FROM: Weitz ES, Hollon SD, Twisk J, et al. Baseline depression severity as moderator of depression outcomes between cognitive behavioral therapy vs pharmacotherapy: an individual patient data meta-analysis. JAMA Psychiatry 2015;72:1102–9.

What is already known on this topic

Some guidelines for the treatment of severe depression recommend that antidepressant medication be used instead of cognitive–behavioural therapy (CBT). This is inconsistent with evidence collating individual patient data (IPD) from multiple randomised controlled trials (RCTs).¹ IPD meta analyses are one way of increasing statistical power and analysing depressive symptoms of varying severity. A previous IPD meta analysis included only four studies, so may still have lacked statistical power.

Methods of the study

Weitz and colleagues requested IPD from previous RCTs comparing psychotherapy and antidepressants. RCTs were identified via database searches (PubMed, PsycINFO, EMBASE and Cochrane Registry of Controlled Trials). The authors selected studies that had compared CBT alone with antidepressants alone, in adults diagnosed with depression. Relapse prevention or maintenance treatment studies and RCTs with inpatients were excluded. Twenty-four RCTs met inclusion criteria and raw data were obtained from 16 (67%), with 1700 outpatients (794 who had received CBT and 906 who had received a range of antidepressants). Post-treatment scores on the Hamilton Rating Scale for Depression (HAM-D-17) or Beck Depression Inventory (BDI I and II) were used as the main outcomes (10 studies contributed HAM-D and BDI scores, 2 contributed only the BDI and 4 contributed only the HAM-D). Models were also run using binary outcomes, response (50% reduction in HAM-D scores post-treatment) and remission (score of <7 on the HAM-D post-treatment). Multilevel statistical models that treated each study as a separate cluster were implemented to control for unobserved heterogeneity between studies.

What this paper adds

• This is the most highly powered study to compare CBT and antidepressants for depression of varying severity. The use of IPD is a major advance. Such an investigation would not be possible with study-level data which would include only the mean depressive symptom scores at baseline with no variation in severity.

• There was no evidence that CBT was less effective than antidepressants for patients with more severe depressive symptoms (regression coefficient 0.00 HAM-D points, 95% CI −0.16 to 0.16) and similar for the BDI (regression coefficient 0.00, 95% CI −0.14 to 0.14). Results were similar when response (OR 0.99, 95% CI 0.93 to 1.05) and remission (1.00, 0.94 to 1.06) were outcome variables. Similarly, when only the more severely depressed groups were examined (defined as >23 on the HAM-D or >29 on the BDI), differences after treatment were small (0.10, 95% CI −0.35 to 0.15 and 0.14, −0.30 to 0.03 for HAM-D and BDI, respectively). The authors do not mention if their thresholds for severe depression were specified a priori. The results are not consistent with the recommendation that more severe depression be treated with antidepressant medication rather than CBT.

Limitations

• Even for this large study, the CIs for the most severe group were not statistically significant: −0.35 to 0.15 on the HAMD and −0.30 to 0.03 on the BDI. The results are therefore compatible with a benefit for antidepressants of around 0.3 standard deviations, that could be clinically important.

• The data did not contain many people with very severe depressive symptoms. They defined severe depression as a HAM-D score of >23 and a BDI score of >29. In many UK trials, the mean BDI score has been over 30, for example, 33.7.² In this IPD meta analysis, mean scores at baseline were 19.2 on the HAM-D and 30.9 on the BDI. They also excluded trials of inpatients with the most severe depressions.

What next in research

More studies of the relative effectiveness of CBT versus antidepressant medication are required to increase statistical power and investigate more severe depressions. This might include inpatients or patients with depression and psychotic symptoms. Alternatively, studies could recruit people with high scores on depressive symptom questionnaires. Future studies could also further examine whether antidepressant medication combined with psychotherapy is more effective for severe depression than either treatment alone.

Do these results change your practices and why?

Yes. The findings by Weitz and colleagues highlight that some guidelines for the treatment of moderate to severe depression are not supported by current evidence. Clinicians might have been deterred from using CBT for severe depression based on current guidelines. These results from the largest study to date provide no evidence that CBT is less effective than antidepressants in the more severe depressions. This is important given that CBT may be better at reducing relapse.³ Some patients may also prefer CBT to medication which could increase the likelihood that they will persist with treatment.