THE COST-EFFECTIVENESS OF STATINS FOR SECONDARY PREVENTION OF CARDIOVASCULAR DISEASE: OBSERVATIONAL STUDY USING DATA FROM ROUTINE ELECTRONIC MEDICAL RECORDS

OBJECTIVES: To evaluate the cost-effectiveness of statins for secondary prevention of cardiovascular disease in routine practice from the perspective of the English NHS. METHODS: Electronic English NHS health records of 6,078 previously untreated patients of age 60+ receiving statins following the occurrence of a myocardial infarction (MI) and a one-to-one propensity score matched control of untreated patients after a MI, resulted in n=12,156 patients. Costs of primary care service use and referrals were calculated from CPRD data and inpatient hospitalisations from HES and HRG codes. QALYs were calculated from ONS mortality records and utilities from the literature applied to MI or ischemic stroke events recorded in HES. Inverse probability of censoring weights (IPCW) were used to estimate counterfactual outcomes in the absence of treatment cross-over in the control arm. RESULTS: The distribution of baseline characteristics was balanced across the two treatment groups, and 43% initially untreated patients crossed-over to statins treatment. For 60-74 year olds, we found an increase in QALY with statins of 0.66 (95% CI: 0.44-0.87) and, in the 75+ group, a gain of 0.81 (95% CI: 0.72-0.88) per patient. Excluding the CV-unrelated costs of inpatient hospitalisations, the total incremental costs were respectively £1,616 (95% CI: 1347-2095) and £2,644 (95% CI: 2034-2963) per patient. Statins had an ICER of £2456 (95% CI: 1814, 2759) in the younger patient group and of £3250 (95% CI: 2843, 3531) in the older group. In contrast, when inpatient hospitalisation costs ICD-10 codes for non-CV events were included, statins resulted in cost savings (95% CI: -1165, 2782), and was consequently dominant, in the younger group, and had incremental costs of £5562 (95% CI: 4356, 6436) for an ICER of £7200 (6221, 8587) in patients aged 75+. CONCLUSIONS: Quasi-experimental evaluation using data from electronic medical records of patients treated in routine practice is feasible. CEA based on decision models may underestimate the cost-effectiveness of statins, due to their omission of hospitalisation costs associated with non-CVD outcomes.