Effect of reimbursement restriction policy on the use of benzodiazepines in the Netherlands: an interrupted time series analysis

OBJECTIVES: Use of benzodiazepines has health risks. Reimbursement was restricted in the Netherlands from January 2009 onwards with the goal to reduce chronic use and healthcare expenditures. The aim of this study is to assess the initial and long-term effects of this policy on benzodiazepine use. DESIGN: Interrupted time series analysis, segmented regression models, Kaplan-Meier survival analysis and Cox proportional hazards analysis. SETTING: A 10% random sample of benzodiazepine dispensings by outpatient pharmacies between January 2002 and August 2015 were obtained from the PHARMO database. This database covered a catchment area representing about 3.6 million residents in 2015. PARTICIPANTS: 2 500 800 benzodiazepine prescriptions from 128 603 patients were included. INTERVENTION: Reimbursement restriction policy from January 2009 onwards. OUTCOME MEASURES: Changes in: the volume of dispensed prescriptions and doses, the incidence, prevalence of incidental, regular and chronic use and discontinuation rates of benzodiazepines. RESULTS: The volume of dispensed prescriptions and doses decreased by 12.5% (95% CI 9.0% to 15.9%) and 15.1% (95% CI 11.4% to 17.3%) respectively in January 2009 compared with December 2008. A clear initial effect on the overall incidence (-14.7%; 95% CI -19.8% to 9.6%) and the prevalence of incidental (-17.8%; 95% CI -23.9% to 11.7%), regular (-20.0%; 95% CI -26.1% to 13.9%) and chronic (-16.0%; 95% CI -23.1% to 8.9%) use was observed. A statistically significant reduction in the monthly trend per 1000 medication users was observed for the overall incidence (-0.017; 95% CI -0.031 to 0.003) and the prevalence of incidental (-3.624; 95% CI -4.996 to 2.252) but not for regular (-0.304; 95% CI -1.204 to 0.596) and chronic (0.136; 95% CI -0.858 to 1.130) use. Patients who started treatment before policy had a slightly higher probability of discontinuation (HR=1.013; 95% CI 1.004 to 1.022). CONCLUSIONS: The reimbursement policy had a significant initial effect on the volume, incidence and prevalence of benzodiazepine use. In addition, there is a statistically significant reduction in the monthly trend of overall incidence and of the prevalence of incidental use. No statistically significant reduction in the monthly trend of chronic use, the main purpose of the reimbursement restriction, could be demonstrated