Statin Use Over 65 Years of Age and All-Cause Mortality

A 10-Year Follow-Up of 19 518 People

Sophia Eilat-Tsanani, MD; Elad Mor, MD; Yochai Schonmann, MD, MSc


J Am Geriatr Soc. 2019;67(10):2038-2044. 

In This Article



We followed a population-based cohort of 19 518 older adults without preexisting cardiovascular disease for a median of 9.7 years. All-cause mortality rates were 34% lower among those who had adhered to statin treatment compared with those who had not. Adherence to statins was also associated with fewer ASCVD events. The benefit associated with statin adherence did not diminish beyond age 75 and was evident regardless of sex. The findings remained robust through various predefined sensitivity analyses.

Comparisons With the Existing Literature

Prospective evidence supporting the efficacy of statins for primary prevention among older adults is scarce;[3,8,42] recent observational studies attempted to address this gap with conflicting results. In a cohort of 1130 American male physicians who used statins, there were fewer deaths (HR = .82) and cardiovascular events (HR = .86) over a median follow-up of 7.1 years compared with a matched cohort of nonusers.[21] There was no evidence for benefit among those older than 76 years, but this conclusion was based on few participants and did not reach statistical significance. This study was based on self-reports of a specific patient population, implying limited internal and external validity. In another study of routinely collected data from Spain, the authors followed 7500 new users of statins (excluding 39 000 current statin users) and concluded that statins were only beneficial for primary prevention among older patients with diabetes.[23] Adherence to statins decreases substantially with time and age,[17] but the authors of both these studies were not able to account for statin use beyond the baseline that likely biased their results toward the null. Finally, a French population-based study matched 3700 older new users of statins with nonusers, excluding five times as many prevalent statin users. Adherence to statins was modestly associated (HR = .93) with a reduced composite end point of all-cause mortality or an acute coronary syndrome among a subgroup of those with a modifiable cardiovascular risk factor, but the authors could not fully account for risk factors such as smoking, obesity, and hypertension.[22] Both the Spanish and French cohorts included only new statin users, that is, those whose physicians deemed fit for a new treatment at an advanced age. Their results may therefore inform decisions on treatment initiation but are probably less applicable to the real-life setting where the continuation of statins in considered. The overall adherence to statins of cohort participants was low (28% among those <75 y and 25% among those >75 y) but consistent with previous observational studies.[17]

Strengths and Limitations

The strengths of our study include its considerable sample size and the comprehensiveness of the available data. The CHS chronic disease register is continuously updated, independent of any study outcomes,[26] and ongoing organizational measures are in place to assure high-quality standardized care.[43] We were therefore able to account for the relevant comorbidities and risk factors in an unbiased, time-updated manner. We used medication claims, a robust proxy measure for actual use, to quantify the changing patterns of statin adherence. We may have slightly overestimated actual use, but it is unlikely that patients consistently purchased medications without taking them over the entire 10-year period. Our main analysis focused on adherence as a dichotomous variable (ie, >80% PDC), but secondary analyses were also consistent with a dose-response association between level of statin adherence mortality, providing further support for a causal association (Supplementary Tables S4 and S5).

We studied prevalent statin users, reflecting an increasingly common clinical scenario; this approach is necessary and valid when it is not feasible to follow up from treatment initiation.[44] Most statin users fail to adhere to their treatment, and usage patterns vary considerably over time (ie, patients may alternate between periods of statin use and cessation).[17] To account for changing treatment patterns, we modeled statin use as the cumulative, annually updated, proportion of standard doses claimed (DDDs). Our results did not differ substantially through several sensitivity analyses, providing reassuring evidence for their robustness.

People who adhere to preventive treatments generally tend to follow other healthy lifestyles and screening recommendations. This "healthy user/adherer" effect may result in overestimation of the effect of statins on mortality if other health behaviors cannot be accounted for.[18] We were also not able to adjust directly for frailty or physical activity that may have further confounded our results. These limitations, however, were mitigated by the availability of rich prospectively collected data. We adjusted for multiple comorbidities that may serve as a proxy for general frailty[45] and for socioeconomic status that is firmly linked with health behaviors.[46] Furthermore, the results remained robust after adjusting for available lifestyle variables (ie, obesity and smoking). Although residual confounding may remain, these results provide evidence against substantial healthy user bias.

We adjusted for clinical ASCVD components (ie, peripheral artery disease and carotid artery disease) rather than exclude those with a preexisting diagnosis. Our analysis therefore included some individuals not strictly eligible for "primary prevention." Finally, the cause of death is not captured in the CHS register. We therefore relied on a proxy measure to assess the cardiovascular-specific mortality component of ASCVD events (ie, noncancer deaths). However, the CHS registers are independently and prospectively updated, and any misclassification of deaths was therefore not likely to be associated systematically with statin use (ie, nondifferential bias, likely to have biased our results toward the null but not to change the direction of the effect).

Interpretation and Clinical Implications

Increasing numbers of older patients take statins, but the evidence base for informed decisions lags behind. Our study provides reassuring evidence in favor of statin continuation among women and men after the age of 75 years. Prolongation of life and prevention of cardiovascular events are meaningful and intuitive outcomes but may not be the sole consideration for making an informed decision at advanced ages. We could not address important patient-oriented outcomes such as quality of life and adverse events, and our results should therefore be interpreted with caution. Nevertheless, studies like ours provide much needed evidence, assisting clinicians in engaging their patients in a meaningful shared-decision process.

In conclusion, we found that adherence to statins was associated with reduced all-cause mortality and fewer ASCVD events among a large population-based cohort of older adults, supporting the continuation of statin use among older people. These findings should be duplicated in different settings and highlight the need for a prospective interventional trial in an older population.