The authors concluded that long-term adjunctive antidepressant treatment was not more effective than mood stabilisers in bipolar disorder and this encourages reliance upon mood stabilisers for maintenance. This was generally a well-conducted review that highlighted the limited generalisability of the findings due to lack of evaluation of modern drugs in the discussion, but not the conclusions.

To evaluate the benefits and risks of long-term antidepressant treatment in patients with bipolar disorder.

MEDLINE, HealthStar, Current Contents, PsycINFO, CINAHL, EMBASE, National Library of Medicine, DARE and The Cochrane Library were searched from 1960 to May 2008. Search terms were reported. Searches of reference lists of identified studies, reviews, abstracts and poster presentations from regular meetings of six specified meetings (2000 to 2007) were conducted. Experts and major funding agencies were contacted for unpublished studies. English and non-English language reports were eligible.

Randomised controlled trials (RCTs) were eligible if they compared any type of antidepressant drug with a non-antidepressant drug or placebo with or without co-treatment with mood stabilising drugs in patients diagnosed with type I or II bipolar disorder using modern criteria. Treatment duration had to be six or more months. Studies had to assess mood states, recurrences or latency to either recurrence or worsening of major depressive and other episodes.

The review assessed depressive or manic relapses and recurrence. Included studies compared: antidepressant versus placebo or mood stabiliser; and antidepressant with or without mood stabiliser versus mood stabiliser or placebo. All but two studies used tricyclics (imipramine or amitriptyline) as the antidepressant and lithium as the mood stabiliser; two studies apparently included serotonin reuptake inhibitors as the antidepressant. Mean age of patients was 40.6 years. Mean duration of illness was 13 years. Percentages of males ranged from 23% to 77%. In all but one of the studies, patients had to show benefit of antidepressants or mood stabiliser treatment in acute bipolar episodes before enrolment. The authors stated that most studies were in bipolar type-I patients. Duration of included studies ranged from eight to 36 months.

Two reviewers independently selected studies.

Two blinded reviewers independently assessed validity using the Jadad criteria (randomisation, blinding and reporting of withdrawals).

For each study, rate ratios (RR) and 95% confidence intervals (CI) were calculated. In studies with multiple treatment arms, each comparison was treated as a separate contrast.

Two reviewers independently extracted data. Discrepancies were resolved by consensus among all four authors.

Pooled weighted rate ratios and 95% confidence intervals were calculated. Pooled risk differences were calculated. The number needed to treat (NNT) to prevent one case of depression and the number needed to harm (NNH) were calculated based on random-effects risk differences. The influence of each study was assessed by omitting each study in turn and repeating the analysis. Subgroup analysis was used to compare the following regimens: antidepressants versus placebo; antidepressant with or without mood stabilisers versus mood stabilisers; antidepressant with or without mood stabilisers versus mood stabilisers alone; and antidepressants versus mood stabilisers. Publication bias was assessed using funnel plots.

Seven RCTs were included (n=350). The studies were judged to have moderately high quality (mean Jadad score 3.1 out of 5). One study was not blinded. Drop-out rates ranged from 0% to 69%.

Long-term antidepressant treatment was associated with a statistically significant and moderate reduction in the recurrence of depression compared to control (RR 0.73, 95% CI 0.55 to 0.97, p=0.03, NNT=11.1; 12 comparisons, seven studies) and a significant increased risk of mania (RR 1.72, 95% CI 1.23 to 2.41, p=0.002, NNH=7; nine comparisons, five studies).

The risk of new depression was significantly reduced for antidepressants compared with placebo (RR 0.46, 95% CI 0.27 to 0.80; three comparisons, n=50). There was no significant difference in new depression for: antidepressant with or without mood stabilisers versus mood stabilisers (eight comparisons, n=346); antidepressant with mood stabilisers versus mood stabilisers alone (five comparisons, n=246); and antidepressant versus mood stabilisers (three comparisons, n=118).

The risk of new mania was significantly increased for antidepressants with or without mood stabilisers compared with mood stabilisers (RR 1.80, 95% CI 1.22 to 2.65; eight comparisons, n=364) and antidepressants compared with mood stabilisers (RR 2.37, 95% CI 1.38 to 4.05; three comparisons, n=118). There was no significant difference in new mania for antidepressants versus placebo (three comparisons, n=50) or antidepressants plus mood stabilisers versus mood stabilisers alone (five comparisons, n=246).

Long-term adjunctive antidepressant treatment was no more effective than mood stabilisers alone in patients with bipolar disorder and this encouraged reliance upon mood stabilisers for maintenance. Further research was required to evaluate currently used drugs.

The review question was clearly stated and appropriate inclusion criteria were defined. An extensive search was conducted and attempts were made to minimise publication and language biases. Appropriate methods were used to minimise reviewer error and bias during the review process. Only RCTs were included. Validity was assessed, but results were not reported in full. Studies were pooled using meta-analysis and although heterogeneity was not assessed, forest plots were presented for the main analyses thus allowing examination of differences in results among studies. Where multiple comparison groups shared a control group, no adjustment was made for statistical dependency. This limitation was acknowledged by the authors, but its influence on meta-analysis findings was not examined. Some limitations of the review were discussed. This was generally a well-conducted review that highlighted the limited generalisability of the findings due to lack of evaluation of modern drugs in the discussion, but not the conclusions.

Practice: The authors advised caution in the use of long-term antidepressants in patients with bipolar disorder and recommend relying on mood stabilisers for maintenance.

Research: The authors stated that there is an urgent need for research to evaluate the long term effects of currently widely used antidepressants and mood stabilisers alone and in various combinations in patients with type I and II bipolar disorder.

National Institute of Health (grant no. MH-64198 (SNG)); Bruce J Anderson Foundation; and McLean Private Donors Neuropsychopharmacology Research Fund.

Two authors are consultants or have received funding from pharmaceutical companies.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.