Evaluating pharmaceutical policy impacts using interrupted time series analysis: An Australian case study

Abstract

Determining the impacts of policy on health outcomes is important for policy makers, clinicians and consumers. Interrupted time series analysis is a powerful quasi-experimental method for quantifying change in an outcome after policy implementation. We illustrate the use of interrupted time series analysis for policy evaluation with an Australian case study. Use of prescription medicines in Australia were examined before and after the implementation of pharmaceutical-subsidy changes. Methods: Interrupted time series analysis compares longitudinal data, aggregated into time-units, before and after a change-point. A line of best fit is calculated for the period before and after the change-point and the differences in the level (i.e. height) and trend (i.e. slope) of these lines are quantified. In our case study, dispensings of specified medicines in Australia were compared for 60 months before and 33 months after a substantial increase in prescription costs in January 2005. Results: Interrupted time series analysis quantifies level and trend changes occurring after a change-point and indicates when, and for how long, changes occur. Significant change in the level of a series indicates an immediate policy impact while a significant trend change indicates an on-going impact on an outcome. We found significant decreases in the level or trend of dispensings for 12 medicine classes indicating both immediate and on-going declines in use. Declines were largest for low income patients and for medicines used preventatively to treat asymptomatic conditions.Conclusions: Interrupted time series analysis provides a simple and feasible method of evaluating the impact of already-implemented policies on health outcomes. Findings from the case study, suggested that the January 2005 increase patient co-payments had affected the use of medicines, and that the largest impacts were on low income patients and medicines used to prevent disease progression. These findings had implications for patient care and health service planning. Interrupted time series can be adapted to a range of settings to provide feedback important for future policy formulation.