A Risk Model to Predict Statin Non-Adherence Following an Acute Coronary Syndrome

Abstract

Patients at risk of statin non-adherence are often not identified during hospital admission with an acute coronary syndrome (ACS). In 19,942 patients hospitalised for ACS, statin dispensing was determined from the national pharmaceutical dispensing database. A risk score for non-adherence was developed from a multivariable Poisson regression model of associations between risk factors and the statin Medication Possession Ratio (MPR) <0.8 6-18 months after hospital discharge. Statin MPR was <0.8 in 4,736 (24%) patients. MPR <0.8 was more likely in patients with a history of cardiovascular disease (CVD) (RR 3.79, CI 95% 3.42-4.20) and those without known CVD (RR 2.25, 95% CI 2.04-2.48) who were not taking a statin on ACS admission, compared to patients with low density lipoprotein (LDL) cholesterol <2 mmol/L who were on a statin. For patients taking a statin on admission, higher LDL was associated with MPR <0.8 (≥3 versus <2 mmol/L, RR 1.96, 95%CI 1.72-2.24). Other independent risk factors for MPR <0.8 were age <45 years, female, disadvantaged ethnic groups, and no coronary revascularisation during the ACS admission. The risk score, which included nine variables, had a C-statistic of 0.67. MPR was <0.8 in 12% of 5,348 patients with a score ≤5 (lowest quartile) and 45% of 5,858 patients with a score ≥11 (highest quartile). A risk score generated from routinely collected data predicts statin non-adherence in patients hospitalised with ACS. This may be used to target inpatient and outpatient interventions to improve medication adherence.