A prospective study of risk factors for nonadherence with antipsychotic medication in the treatment of schizophrenia

Abstract

Objectives: This study aimed to prospectively identify the best single predictor and the best set of predictors of risk for nonadherence with anti-psychotic medication in the treatment of patients with schizophrenia.

Method: We used data from 1579 patients in a 3-year, prospective, naturalistic, nonrandomized, multisite study of schizophrenia patients conducted from July 1997 to September 2003 (U.S. Schizophrenia Care and Assessment Program). Adherence with any oral antipsychotic medication was assessed using patient-reported medication adherence and an indirect adherence measure based on medical record prescription information. Patients who reported poor medication adherence or had a medication possession ratio < or = 80% (percentage of days with prescriptions for any oral antipsychotic) during the first year after enrollment were defined as nonadherent (N = 296, 18.8%). Thirty-nine previously reported potential risk factors of nonadherence with antipsychotic medication were assessed at enrollment with valid and reliable measures. Risk factors represented patient-, environment-, and treatment-related domains, including sociodemographics, symptom severity, substance use, threat to safety of self and others, other illness-related factors, need for supervision, medication-related adverse events, and prior medication-utilization patterns.

Results: The best single predictor of future nonadherence was nonadherence during the 6 months prior to enrollment (odds ratio = 4.1, 95% confidence interval = 3.1 to 5.6, p < .001). The best set of predictors of nonadherence, ordered by strength of association, included prior non-adherence, recent illicit drug use, recent alcohol use, prior treatment with antidepressants, and greater patient-reported, medication-related cognitive impairment.

Conclusion: Nonadherence with antipsychotic medication is associated with a well-defined set of risk factors that can be used to identify patients who are predisposed to poor adherence.