To examine the long-term effects of drug reimbursement adjustments on drug-switching decisions and to investigate whether patients with complicated or severe conditions are more affected.

A population-based, longitudinal study with a before-and-after design. Analysis of 141,703 patients with type 2 diabetes covered by the universal health insurance program in Taiwan. Observation of five 6-month phases before and after a drug reimbursement adjustment implemented in October 2009. Drug switching was defined as a brand change within the same anatomical therapeutic chemical group between two consecutive physician visits. Generalized estimating equations were employed to control for the random subject effect.

The drug-switching rates in the five phases were 10.85% and 13.71% before implementation and 31.53%, 28.29% and 15.61% after implementation. Results from the regression model revealed a higher likelihood of receiving switched drugs in phases 3, 4 and 5, with odds ratios of 3.16, 2.72 and 1.44 (with 95% confidence interval 3.04–3.29, 2.61–2.84 and 1.38–1.51), respectively, compared with phase 1. Patients with complicated or severe conditions were more likely to have their drugs switched after the reimbursement adjustment.

The drug reimbursement adjustment under the health insurance program resulted in an increase in drug-switching decisions, and patients were not exempt from medication switching regardless of the complications or the severity level of their illness.