Differential effects of atypical versus typical antipsychotic medication on earnings of schizophrenia patients: Estimates from a prospective naturalistic study

Background: Rising public and private expenditure on antipsychotic medications is concentrated on the cost of second generation or 'atypical' medications, which are more expensive than first generation medications and make up a rapidly growing share of all antipsychotic prescriptions. Previous studies have examined whether the higher acquisition costs of atypicals are offset by other cost and/or utilisation benefits. This paper extends this literature by examining possible effects of atypicals on earnings and related measures of labour supply in a large naturalistic study with a long-term follow-up period. Methods: We analysed data on earnings and other characteristics from the Schizophrenia Care and Assessment Program (SCAP), a 3-year longitudinal study (with data collection during the years 1997-2003) of 2327 adults with schizophrenia (including schizoaffective and schizophreniform disorders) recruited from behavioural healthcare provider systems in six areas of the US. We used empirical criteria and data from the SCAP database to identify 336 patients aged <50 years who were in the stable or 'maintenance' phase of their antipsychotic treatment during the 6 months prior to baseline. Effects of atypicals compared with typicals were estimated from Tobit regression models that included additional covariates and the baseline-dependent variable values. Regression-dependent variables were reported earnings per month, hours worked per month, days worked per month and a binary indicator of employment. To control for the effect of selection bias in choice of type of atypical, we employed an instrumental variables (IV) estimation procedure. Results: For all dependent variables, our IV Tobit regressions yielded consistently positive coefficient estimates for atypical use that were either marginally significant (p < 0.1) or significant (p < 0.05) for earnings, significant for hours and days of work and not as consistently significant for employment status. Results from these regressions imply a positive effect of atypical use on monthly earnings in the range of $US107-122. In regressions that did not control for selection bias by using IVs, coefficients for atypical use were often negative and never statistically significant. Conclusions: Our results indicate that higher drug costs of atypicals for maintenance-phase treatment are at least partially offset by higher earnings among patients. These effects represent benefits to consumers as well as savings to taxpayer-supported income transfer programmes. Future studies should seek to determine if treatment with atypicals increases patients' earnings via better control over negative symptoms and/or improved patient cognition. Both appear to be connected with employment and labour supply in patients with schizophrenia, and both may be improved through use of atypicals.