Safety of Ceasing Aspirin Used Without a Clinical Indication After Age 70 Years: A Subgroup Analysis of the ASPREE Randomized Trial

Background: The ASPREE (ASPirin in Reducing Events in the Elderly) trial was a randomized, double-blind, placebocontrolled primary prevention trial of aspirin in 19114 communitydwelling persons aged 70 years and older (≥65 years in U.S. racial minorities). The results of the trial demonstrated that aspirin had no benefit for disability-free survival, prevention of cardiovascular disease events, or prevention of incident cancer, and increased risk for major bleeding and all-cause mortality (1–3). These findings were interpreted by some as being relevant only to aspirin initiation and not aspirin discontinuation (4). The availability of evidence to inform the risks (for example, forgone cardiovascular protection) and benefits (for example, decreased risk for major hemorrhage) from aspirin cessation among older adults is timely, given updated guideline recommendations regarding aspirin use and clinical uncertainty (5). Objective: To investigate the effect of aspirin cessation versus continuation on disability-free survival and other clinical outcomes in a post hoc analysis of ASPREE participants who were regularly taking aspirin before trial enrollment. Methods and Findings: We included participants who reported taking aspirin 2 or more days per week at enrollment in the trial. We compared those randomly assigned to placebo (cessation) with those randomly assigned to aspirin (continuation). The primary outcome was a composite of all-cause mortality, incident dementia, or persistent physical disability (“disabilityfree survival” [1]). Secondary outcomes were all-cause mortality, major adverse cardiovascular events, any cardiovascular event, major hemorrhage, and incident cancer (1–3). All outcomes were adjudicated. Cox proportional hazards regression was used to analyze outcomes. Subgroup analyses were conducted by age (<75 years and ≥75 years), race (White and non-White), and years of pretrial aspirin exposure (<5 years and ≥5 years). The Table (page 762) shows baseline characteristics of the entire ASPREE population and the group for this analysis. Of 19114 recruited participants, 2094 (11%) reported aspirin use before trial entry, of whom 1714 reported taking aspirin 2 or more days per week. Due to higher rates of pretrial aspirin use, U.S. participants were overrepresented in this analysis (40.6%) relative to the entire ASPREE study (12.6%). Over a median follow-up of 4.9 years, evidence of an increased risk for the primary outcome for aspirin cessation versus continuation was weak, and this appeared to be confined to non-White participants (Figure [page 763]). Overall, 116 of 841 participants (13.8%) in the cessation group and 97 of 873 (11.1%) in the continuation group experienced the primary end point (incidence rate, 29.8 vs. 23.4 per 1000 person-years); the hazard ratio for cessation versus continuation was 1.28 (95% CI, 0.98 to 1.68). When the analysis was restricted to those taking prior aspirin 6 or more days per week, the hazard ratio for cessation versus continuation was 1.20 (CI, 0.89 to 1.60). Although it was not associated with any secondary outcome, aspirin cessation appeared to increase cardiovascular disease events among those who reported taking aspirin for 5 years or longer. Age had no evident impact on the effect of aspirin cessation. Discussion: Aspirin cessation through random assignment to placebo among prior regular aspirin users gave rise to increased rates of the primary end point and cardiovascular disease events compared with continuation through random assignment to aspirin, but the CIs were wide and the findings were therefore inconclusive. No substantial increased risks for major hemorrhage and cancer were seen with continued aspirin use. Absence of an effect on hemorrhage may relate to selfselection for tolerance among aspirin users. Noting that the analyses are likely underpowered, our findings cannot conclusively demonstrate clear harm or benefit of either cessation or continuation of aspirin in older adults. We previously performed subgroup analyses from the main trial by prior aspirin use; they can be found elsewhere (1– 3). However, these analyses were based on any previous aspirin use. We restricted our analyses to regular aspirin users to address the question of the effects of aspirin cessation in said users. Our findings were consistent among those taking aspirin for at least 6 days per week. Our study does have limitations. A 4-week placebo run-in phase for the trial occurred before randomization, which meant a gap in continuous exposure for those randomly assigned to aspirin. In addition, the post hoc study design within an ASPREE subgroup reduced statistical power for this exploratory analysis. Considering our study limitations of power and likely selfselection for tolerance, we cannot definitively address the question of the safety of aspirin cessation in older populations without a clinical indication for its use. This would require a well-designed study with a large sample size. However, given the community sampling frame and the findings of the main study, at this time, a pragmatic recommendationmay be to consider aspirin cessation in those with a large medication burden with due caution.