OBJECTIVES : Antipsychotic polypharmacy (APP), concurrent use of two or more antipsychotic medications, is considered low value care. Our objective was to estimate the prevalence and factors associated with APP among non-institutionalized civilian adults with any antipsychotic use.

METHODS : This study had a retrospective cross-sectional design and used the data from 2008 – 2015 Medical Expenditure Panel Survey (MEPS). The sample consisted 3130 adults over age 21 years. APP was defined as concurrent use of two or more medications within a round consisting of nearly 180 days. The Andersen's model for health service use was used to examine the association between predisposing, enabling, need, personal health behavior and environmental factors with APP use. Logistic regression models were conducted examine the factors associated with APP. Analysis accounted for the complex survey design of the MEPS.

RESULTS : Overall, 9.6% of the study sample reported APP. APP declined from 15.0% in 2008 to 7.0% in 2015 (p =.06). In adjusted logistic regression model, increasing age (Adjusted Odds Ratio (AOR) =0.98, 95% Confidence Interval (CI) = [0.97, 1.00]) significantly decreased odds of receiving APP. In terms of physical health status, individuals with poor or fair status had a significantly higher odds of APP (AOR =1.68, 95% CI = [1.00, 3.80])) compared to those with excellent or very good status. Moreover, those who were unemployed (AOR =1.95, 95% CI = [1.02, 3.72]), with depression (AOR =2.12, 95% CI = [1.37, 3.30]), and with schizophrenia (AOR =3.44, 95% CI = [2.23, 5.23]) had significantly higher odds of APP use.

CONCLUSIONS : Approximately, one in 10 adults reported APP between 2008 and 2015. Policies, and interventions outside the health sector such as improving (example: employment) may be considered to reduce APP. Future well-designed studies are may need to adopt a population-health framework to identity patient level and social determinants of APP.