Estimates of Statin Discontinuation Rates are Influenced by Exposure and Outcome Definitions

Abstract

Background Estimates of statin discontinuation rates are generally high but show large variations. Discontinuation rates are possibly influenced by unrecognized supplies from previous prescriptions and the operational definitions of statin discontinuation. Objective: To investigate whether the outcome (discontinuation) was affected by (1) the operational definition used to calculate statin exposure and (2) the operational definition and different cutoff values used to calculate discontinuation. Methods: Data for this study were obtained from the PHARMO medical record linkage system in the Netherlands. Participants were patients with a new statin and no statin prescription in the preceding year. The outcome, discontinuation, was defined based on a variable number of days without medication after exposure (gaps) or based on the availability of supplies 12 months after the inclusion date (at 1 year). Exposure to statins was assessed by 2 methods. The method termed No Overlap accounted for only the supplies of the last prescription to calculate exposure, and the method termed Overlap accounted for all supplies from previous prescriptions. We investigated the effect of 4 exposure outcome combinations on statin discontinuation estimates. Results: The exposure outcome combinations, including overlap, resulted in a 7% unit lower discontinuation rate. At gap lengths of 90 days and longer, no significant differences between No Overlap and Overlap were observed. Shorter minimum gap lengths gave higher discontinuation rates compared to longer minimum gap lengths and ranged from as high as 86% to 21%. Conclusions: If previous supplies are accounted for in the calculation of exposure to statins, lower discontinuation rates are observed, The influence of previous supplies on discontinuation rates is less pronounced than the influence of gap lengths. The calculation of exposure does not influence discontinuation if gaps longer than 90 days are used to assess discontinuation.