(A-155) Effect of statin discontinuation on mortality among older adults with and without cardiovascular disease

Background: Statin use among older adults (≥75 years) continues to be debated. We previously used the prevalent new-user (PNU) study design to estimate the average treatment effect of discontinuing (ATD) statins on mortality among older adults who discontinued statins. We found that discontinuation increased mortality; however, available randomized trials have not consistently shown harm from discontinuation. Further evaluation is warranted to assess the validity of the PNU design in evaluating statin discontinuation.

Objectives: To estimate the ATD statins on mortality among older adults who discontinued statins without cardiovascular disease (CVD, primary prevention) and with CVD (secondary prevention).

Methods: From a 20% random sample of Medicare beneficiaries, we assembled a cohort of statin initiators (no use in the prior year) who continued statins for ≥5 years (no gaps >90 days) for primary and secondary prevention of CVD (angina, heart failure, ischemic heart disease, myocardial infarction, and stroke), identified by diagnosis codes with all available data. We used the PNU design to estimate the ATD statins on mortality among statin discontinuers overall and by primary and secondary prevention. We computed a monthly-stratified propensity score for discontinuation and used odds weights to adjust for confounding and frailty. We estimated crude and adjusted all-cause mortality rates (MR) per 100 person-years (py) and mortality rate differences (MRD) and mortality rate ratios (MRR) for discontinuers vs. continuers. We derived 95% CIs with the 2.5th and 97.5th percentiles from 500 bootstraps.

Results: We identified 110,102 patients that met our study criteria. At the start of follow-up, 32.0% of patients were receiving statins for primary prevention and 23.2% went on to discontinue statins. The overall crude MRs were 12.4 and 6.2 per 100 py among discontinuers and continuers, respectively. The adjusted MRD was 3.7 (95% CI: 3.5, 4.2) per 100 py and MRR was 1.43 (1.40, 1.50). In the primary prevention cohort, crude MRs were 5.0 and 2.6 per 100 py among discontinuers and continuers, respectively. The adjusted MRD was 1.6 (1.3, 2.4) per 100 py and MRR was 1.49 (1.38, 1.77). In the secondary prevention cohort, crude MRs were 15.6 and 7.8 per 100 py among discontinuers and continuers, respectively. The adjusted MRD was 4.7 (4.3, 5.1) per 100 py and MRR was 1.43 (1.39, 1.48).

Conclusions: We found statin discontinuation was associated with higher short-term all-cause mortality whether patients were receiving statins for primary or secondary prevention. The elevated MRR in primary prevention patients suggests potential residual confounding. Caution is advised in using the PNU design to evaluate the effects of deprescribing on mortality, particularly in primary prevention.