Objective: To determine the efficacy of antidepressant therapy for unexplained symptoms or symptom syndromes.
Search strategies: We identified original studies through searching MEDLINE, EMBASE, PsycLIT, the Federal Research in Progress database, and The Cochrane Library. We also searched the bibliographies of primary and review articles for additional studies.
Selection criteria: We excluded trials of patients with neuropathic, oncologic, or degenerative joint pain. Independent duplicate review of 392 articles identified 94 relevant reports of randomized trials involving 6595 patients across 6 symptom syndromes. Independent duplicate assessment was made for inclusion and data abstraction. Meta-analysis was performed on extractable placebo-controlled data.
Main results: Of 94 included trials, most studied either tricyclic antidepressants, antiserotonin antidepressants, selective serotonin reuptake inhibitors (SSRIs), or multiple agents for the treatment of the following syndromes: headache (50), fibromyalgia (18), functional gastrointestinal syndromes (13), idiopathic pain (11), tinnitus (2), and chronic fatigue (2). The quality of the studies was fair (mean score = 4.8 on a scale of 0 to 8). A majority of the studies (69%) demonstrated benefit for at least one outcome measure. Symptom improvement typically did not correlate with depression response in the few studies where it was assessed. Meta-analysis of all extractable data showed a substantial benefit from antidepressants: For the dichotomous outcome of improvement, the odds ratio was 3.4 (95% confidence interval [CI], 2.6 - 4.5), and for continuous outcomes, the standardized mean difference was 0.87 (95% CI, 0.59-1.14). The absolute percentage difference in improvement between the antidepressant and placebo arms was 32%, yielding a number needed to treat of 3 to improve one person's symptoms. Meta-regression indicated no differential effect across the classes of antidepressants; however, onbivariate tally tricyclic studies were associated with a greater likelihood of efficacy than SSRI studies (P = .02).
Conclusions: Antidepressants can be effective for various physical symptoms and symptom syndromes. The relation of outcome to depression and the efficacy of SSRIs needs further study.