Uptake of Biosimilar Insulin Aspart Following Implementation of a Regional Prescribing Policy Intervention; An Interrupted Time Series Analysis (dataset)

Abstract

Background: Biosimilar insulin prescribing policy interventions offer the opportunity for healthcare systems to make substantial financial savings through both new initiations and active switching from reference products, without adverse effects on patient outcomes. In September 2024, a large integrated care board in northern England implemented a policy intervention to increase the prescribing of biosimilar insulin aspart. Aims: The aim of this study was to quantitatively evaluate changes in the uptake of biosimilar insulin aspart following the prescribing policy intervention using an interrupted time series analysis and to estimate the associated financial savings. Methods: Prescribing data were extracted from the Bennett Institute for Applied Data Science Open Prescribing Database and imported into Microsoft Excel. Pre-intervention and post-intervention trends in prescribing across nine sublocations within the integrated healthcare system were analysed by device type using segmented regression within a linear mixed-effects model. Serial autocorrelation was accounted for using a first-order autoregressive covariance structure. Financial savings associated with the increased biosimilar usage were then estimated by comparing observed prescribing with a modelled counterfactual scenario. Results: The prescribing policy intervention was associated with a statistically significant increase in the uptake of biosimilar insulin aspart pre-filled pen devices, which reached a market share of 59% after 19 months. No statistically significant post-intervention effects were identified for cartridges or vials. The policy intervention was associated with estimated savings of approximately £350,000 over the study period. Conclusions: Biosimilar insulin prescribing policy interventions can deliver significant changes in prescribing practice as well as meaningful financial savings, but their effectiveness may vary according to device type and factors such as local implementation.