Treatment of depression in patients with alcohol or other drug dependence: a meta-analysis

Depression and substance abuse are common and costly disorders that frequently co-occur, but controversy about effective treatment for patients with both disorders persists.

To conduct a systematic review and meta-analysis to quantify the efficacy of antidepressant medications for treatment of combined depression and substance use disorders.

PubMed, MEDLINE, and Cochrane database search (1970-2003), using the keywords antidepressant treatment or treatment depressed in conjunction with each of the following alcohol dependence, benzodiazepine dependence, opiate dependence, cocaine dependence, marijuana dependence, and methadone; a search of bibliographies; and consultation with experts in the field.

Among inclusion criteria used for study selection were prospective, parallel group, double-blind, controlled clinical trials with random assignment to an antidepressant medication or placebo for which trial patients met standard diagnostic criteria for current alcohol or other drug use and a current unipolar depressive disorder. Of the more than 300 citations extracted, 44 were placebo-controlled clinical trials, 14 of which were selected for this analysis and included 848 patients: 5 studies of tricyclic antidepressants, 7 of selective serotonin re-uptake inhibitors, and 2 from other classes

We independently screened the titles and abstracts of each citation, identified placebo-controlled trials of patients with both substance dependence and depression, applied the inclusion criteria, and reached consensus. Data on study methods, sample characteristics, and depression and substance use outcomes were extracted. The principal measure of effect size was the standardized difference between means on the Hamilton Depression Scale (HDS).

For the HDS score, the pooled effect size from the random-effects model was 0.38 (95% confidence interval, 0.18-0.58). Heterogeneity of effect on HDS across studies was significant (P <.02), and studies with low placebo response showed larger effects. Moderator analysis suggested that diagnostic methods and concurrent psychosocial interventions influenced outcome. Studies with larger depression effect sizes (>0.5) demonstrated favorable effects of medication on measures of quantity of substance use, but rates of sustained abstinence were low.

Antidepressant medication exerts a modest beneficial effect for patients with combined depressive- and substance-use disorders. It is not a stand-alone treatment, and concurrent therapy directly targeting the addiction is also indicated. More research is needed to understand variations in the strength of the effect, but the data suggest that care be exercised in the diagnosis of depression-either by observing depression to persist during at least a brief period of abstinence or through efforts by clinical history to screen out substance-related depressive symptoms.