Older adults aged ≥75 years face increasing risk of acute myocardial infarction (AMI) but are underrepresented in trials of statins for primary prevention. Heterogeneity in frailty, multimorbidity, and competing mortality risks complicates extrapolation of trial evidence to very old adults. This study examined the association between baseline statin exposure and incident AMI in community-dwelling adults aged ≥75 years without prior cardiovascular disease using linked real-world health data from the Community of Madrid.
Design and data sources
Investigators performed a retrospective, population-based cohort study using linked primary-care, hospital, laboratory, and pharmacy dispensing records covering the regional health system.
Cohort assembly and exposure definition
An exposure-assessment window spanned 1 January 2018 to 31 December 2019. A landmark date of 1 January 2020 was used to define baseline exposure and begin outcome follow-up. Participants were classified as statin-exposed if they had at least two statin dispensations during the 24-month exposure-assessment period and had no record of lipid-lowering therapy before 2018.
Exclusions
Individuals were excluded if they had prior cardiovascular disease, type 1 diabetes, cancer, dementia, or advanced chronic kidney disease.
Outcome and follow-up
The primary outcome was incident AMI after the landmark date. Follow-up extended for a median of 5 years.
Statistical analysis
Missing covariate data were addressed with multiple imputation. The primary association between baseline statin exposure and incident AMI was estimated with multivariable Cox proportional hazards regression. Sensitivity analyses included propensity-score matching and Fine-Gray competing-risk regression treating all-cause mortality as a competing event.
Cohort characteristics
The final analytic cohort included 174,014 individuals with a mean age of 82.5 years. Of these, 32,698 (18.8%) met criteria for baseline statin exposure.
Incidence of AMI
Over a median follow-up of 5 years, AMI occurred in 533 exposed individuals (1.63%) and 2,722 non-exposed individuals (1.93%); this difference had a p value of 0.0003.
Adjusted associations
In the fully adjusted Cox model, baseline statin exposure was associated with a lower risk of incident AMI (hazard ratio [HR] 0.805; 95% CI, 0.731–0.887).
When accounting for competing mortality using Fine-Gray regression in the full cohort, baseline statin exposure remained associated with a lower cumulative incidence of AMI (subdistribution HR [sHR] 0.823; 95% CI, 0.748–0.905). After propensity-score matching, the competing-risk analysis again showed an association between baseline statin exposure and lower AMI incidence (sHR 0.852; 95% CI, 0.738–0.983).
The observed absolute risk difference for AMI between statin-exposed and non-exposed individuals was 0.30 percentage points (95% CI, 0.14–0.45) over the median 5-year follow-up. This corresponds to an observational number needed to treat (NNT) of 338 over 5 years (95% CI, 222–708), indicating a modest absolute reduction in AMI events at the population level.
Key strengths of the study include a large, population-based sample, linkage across multiple routinely collected health data sources, an explicit exposure-assessment landmark design, multiple imputation for missing data, and use of competing-risk and propensity-score methods as sensitivity analyses.
Limitations noted by the authors include the observational design with potential for residual confounding and selection bias related to survival to the landmark date. Details such as adherence beyond dispensations, statin intensity, and unmeasured confounders cannot be fully addressed in the reported analyses. The authors stress that the modest absolute effect and these limitations mean causality cannot be inferred from these data alone.
In this large real-world cohort of community-dwelling adults aged ≥75 years without prior cardiovascular disease, baseline statin exposure was associated with a lower incidence of AMI across multiple analytical approaches. The absolute benefit observed was modest (0.30 percentage points over ~5 years), and residual confounding or selection effects may influence the observed associations. The authors conclude that randomized evidence is needed to determine whether the association reflects a causal preventive effect of statins in very old adults.
The Regional Research Ethics Committee on Medicines of Madrid (CEIM-R) approved the study (approval reported in the source). Data sharing is currently restricted while infrastructure is developed; selected non-sensitive materials will be deposited in a repository and additional materials and scripts will be made available on project completion. The study declared funding from the Ministry for Digital Transformation and Public Service and the European Union (TSI-100121-2024-79). The authors declared no competing interests.