Duration of therapy and health care costs of fluoxetine, paroxetine, and sertraline in 6 health plans

Background: Previous studies comparing fluoxetine, paroxetine, and sertraline, the 3 most common selective serotonin reuptake inhibitors (SSRIs), in naturalistic settings have produced conflicting results. With this study, we provide new evidence as to the similarities and differences among these SSRI therapies with respect to the duration of use and health care costs. Method: Data from 6 health maintenance organizations were used to identify patients with new-onset major depression, number of days with filled prescriptions, and total health care and depression-related costs. The sample consisted of 1771 patients given initial prescriptions for sertraline (N = 386), fluoxetine (N = 840), or paroxetine (N = 545) in the period from July 1, 1994, to March 31, 1997. Analyses included Cox proportional hazards models (for duration of initial therapy) and ordinary least squares regression (for cost). Results: Patients who initiated therapy with fluoxetine were more likely to have a later interruption of therapy than patients who initiated therapy with sertraline (p = .03) and paroxetine (p = .001). Total 1-year costs did not differ statistically between the treatment groups, but 1-year depression-related costs were significantly lower for patients who initiated therapy with sertraline or paroxetine than for those who initiated therapy with fluoxetine ($332 less for sertraline, 95% confidence interval [CI] = $125 to $562; $339 less for paroxetine, 95% CI = $144 to $416). Limitations: A limitation of this observational study, as well as of observational studies in general, is that unobserved characteristics of the patients may lead to biased estimates of the impact of treatment on adherence or cost, even with controls for observed characteristics. Conclusion: We found no significant differences in total health care costs among the 3 SSRIs, but noted significant differences in depression-related costs (the costs of fluoxetine are greater than those of sertraline and paroxetine). Importantly, there was no relationship between treatment interruption and increased health care or depression-related costs, in contrast to the findings of some, but not all, prior studies.