In an effort to evaluate risk in a sufficiently large adolescent population, an observational study accessed the managed care claim records compiled in the PharMetrics Integrated Outcomes database.8 This database represents 58 million insured patients and 74 managed care plans in the United States. About 24,000 adolescents were identified to have received a diagnosis of major depression and/or a prescription for an antidepressant between January 1998 and March 2003. Even with this large sample, however, only 45 patients received TCA monotherapy. About 5000 were treated with an SSRI, 2000 received an antidepressant combination or other type of medication, and 17,000 received no medication for the diagnosed major depression. The investigators calculated propensity scores to adjust for bias in treatment selection and then applied the Cox proportional hazards approach to estimate the effect of antidepressant treatment on the probability of a suicide attempt. Their analysis indicated that the SSRIs had no significant independent effect on suicide risk. Risk was decreased in patients receiving an antidepressant for at least 6 months.8 The possibility remains that antidepressants can provoke suicide but that it occurs too infrequently to be detected by such observational studies. In Klein's2 view, the detection of this phenomenon may require a sophisticated postmarketing surveillance program to supplement or supplant the current FDA MedWatch program of volunteer adverse drug reaction reporting.
The creation of effective postmarketing surveillance must be brought to the forefront of public discussion, Klein argued, to deal with the confusing barrage of horror stories that spark regulations, warnings, and changes in medical practice, with unclear net effects.