Investigators reported a substantial association between initiating GLP-1 RAs on a background of statins and lower rates of major clinical events in patients with established ASCVD and obesity. The matched analysis (119,911 pairs) produced a lower all-cause mortality proportion in GLP-1 RA users compared with statin-only controls (4.0% vs. 7.5%), with a reported hazard ratio of 0.479 (95% CI 0.463–0.496; p<0.001). The cohort showed reduced risk for myocardial infarction (HR 0.762; 95% CI 0.737–0.788; p<0.001) and for ischemic stroke (HR 0.671; 95% CI 0.588–0.766; p<0.001). These outcomes were the primary clinical signals described by the authors.
The dataset and reported effect sizes are drawn from real-world, aggregated electronic health record data. The study team observed consistent directionality across the three major endpoints; the magnitudes reported in the preprint suggest clinically meaningful associations but, as the authors emphasized, the work is a preprint and has not been peer reviewed.
The study used the TriNetX Research Network, which aggregates de-identified records from 108 U.S. healthcare organizations. Initial identification yielded 166,278 patients who initiated GLP-1 RAs while on statins, and 920,176 statin-only controls. Propensity score matching across 34 covariates produced 119,911 matched pairs for the primary analysis, followed for up to three years. The source did not report granular dosing regimens, exact timing of GLP-1 RA initiation relative to ASCVD events, or adherence metrics; those details were not provided in the preprint.
The authors specified a preplanned subgroup with BMI ≥35 kg/m2 (62,385 matched pairs) and conducted multiple sensitivity checks to probe robustness. Institutional review board approval was not required because the analysis used de-identified aggregate data, and the authors noted that data can be requested from the corresponding author via TriNetX access information.
The investigators reported that a greater proportion of GLP-1 RA users achieved LDL <55 mg/dL by a risk analysis metric (17.1% vs. 15.3%; relative risk 1.118; p<0.001). However, the time-to-event Kaplan–Meier analysis for attaining that LDL threshold was non-significant (HR 1.007; p=0.538). The preprint did not specify whether LDL measurements were fasting, the timing of laboratory draws relative to therapy initiation, or the statin intensity distribution across groups; those methodological specifics were not provided in the source text.
This discrepancy between a cross-sectional risk ratio and a time-to-event analysis was reported by the authors without extended exploration of potential measurement timing or residual confounding that might explain the divergence.
The study team carried out four washout sensitivity analyses using windows from 30 to 120 days; outcomes were described as consistent across these tests. The BMI ≥35 kg/m2 subgroup analysis (62,385 matched pairs) similarly showed consistent benefit signals for mortality, MI, and ischemic stroke per the preprint. The source did not provide subgroup event counts or hazard estimates for every sensitivity test in the abstract, so readers should consult the full preprint PDF for detailed tables and figures.
To test whether inherited severe dyslipidemia altered results, investigators performed an exclusion analysis removing patients with familial hypercholesterolemia, producing 161,312 matched pairs; the mortality benefit reportedly persisted in that cohort. The authors noted the robustness of primary associations across this and other sensitivity checks. The preprint did not list how familial hypercholesterolemia was ascertained (ICD codes, genetic confirmation, or chart review); that ascertainment method was not reported in the abstract.
This analysis used large, de-identified electronic health record data from TriNetX and is subject to the usual limitations of observational research: potential residual confounding despite matching, incomplete or variable capture of medication exposure and adherence, and possible misclassification of diagnoses or outcomes. The preprint did not report dosing schedules, individual GLP-1 RA agents, or duration of therapy beyond the three-year censoring window; those specifics were not provided in the source abstract.
The study team declared no competing interests, and they stated that IRB approval was not required because of de-identified data. The paper remains a medRxiv preprint and has not been peer reviewed; the authors and readers will need peer review and further replication or randomized data to determine whether the associations reflect causal cardioprotection, effects mediated by weight and metabolic change, or residual confounding. Specific unanswered questions documented in the source include the lack of detailed drug-level exposure and the methods for identifying familial hypercholesterolemia; these should be clarified in peer review and subsequent reporting.