The impact of pharmacy benefit design changes on medication adherence and generic drug utilization among commercially insured continuously enrolled patients with chronic diseases

Abstract

While it is crucial for health plans to be protected against rising prescription drug costs, increasing cost sharing too much may mean that beneficiaries can not afford medication and non-compliance will become an issue. The objective of the study was to investigate the impact of pharmacy benefit design (PBD) changes on adherence to chronic medications and generic utilization. The study samples were three cohorts of commercially insured continuously enrolled patients with pharmacy claims of diabetes, hypertension, and hyperlipidemia between Oct. 1, 2010 and Jun. 30, 2013. Pre and post quasi-experimental design with control group was applied. To understand the impact comprehensively, PBD changes were examined in two steps: Step 1: Any changes in PBD; Step 2: Changes in cost sharing strategies only. Medication adherence was measured in proportion of days covered (PDC) by at least one medication of the target disease. PDC was also dichotomized as PDC>=80% and PDC<80%. Generic utilization was measured as the generic dispensing rate (GDR) for all medications took during the study period. Two statistical models were fit: General Linear Regression model for the continuously measured variables of PDC and GDR; and Logistic Regression for dichotomized PDC. Control variables were classified into three categories based on Andersen's behavioral model: predisposing characteristics, enabling resources, and need factors, including beneficiaries' demographic and socioeconomic information, medication conditions, and a proxy health risk estimate using Prospective Risk Score. The study sample was made up of 445,983 patients, of whom 45,850 were identified with benefit changes and 400,133 were not. In the experimental group, 8,049 beneficiaries had claims for diabetes, 36,712 beneficiaries had claims for hypertension, and 20,704 beneficiaries had claims for hyperlipidemia. The final control groups were randomly selected and were three times the number of those in the experimental group for each disease respectively. Mean pre-PDC is significantly higher than post-PDC, and mean pre-GDR is significantly lower than post-GDR for both groups. All models indicated no significant association between adherence and PBD changes. There were, however, significant associations between GDR and PBD changes in two cohorts. Beneficiaries with PBD changes had 0.007(p=0.0002) higher post-GDR in the diabetes cohort; beneficiaries with PBD changes had 0.004 (p<0.001) higher post-GDR in the hypertension cohort. Neither cost-sharing strategy had a significant impact on PDC within studied pharmacy benefit designs. Generic dispensing rate was also not significant associated with most of the cost sharing strategies except the copayment decrease of generics. For beneficiaries with a decrease in the amount of copayment for generic products, the mean post-GDR was 0.017 (p=0.0003) lower than beneficiaries without copayment changes in the diabetes cohort, although not significant in the hypertension and hyperlipidemia cohorts. Within the studied pharmacy benefit designs, design changes did not seem to affect medication adherence, but did positively affect generic drug utilization.