Authors’ reply: Poor long-term cardiovascular risk factor management after acute coronary syndrome: An observational cohort study

Abstract

Dear Editor,

We thank Fauchier et al. for their interest in our study and their thoughtful methodological considerations regarding the interpretation of sex differences in post-acute coronary syndrome (ACS) management [1]. The authors raise two main concerns: potential confounding by myocardial infarction (MI) subtype (STEMI vs. NSTEMI) and an insufficient consideration of age as an effect modifier. Both aspects were considered in our analytical framework.

Age was incorporated as a covariate in all regression models, guided by directed acyclic graphs (DAGs). In addition, the observed (sex) differences in statin prescription rates were consistent across all examined age strata (Supporting Information), arguing against residual confounding by age as a likely explanation for our findings.

Regarding MI subtype, current ESC guidelines recommend uniform secondary prevention targets and follow-up for all ACS patients, irrespective of the index event [2]. Our objective was to evaluate whether these targets are achieved in routine care; accordingly, the ACS population as a whole constituted the relevant unit of analysis. Sex was modeled as the exposure in fully adjusted models, and differences in case-mix by MI subtype are therefore unlikely to explain the observed associations.

As discussed by Hernán et al., Simpson's paradox reflects a mismatch between statistical conditioning and underlying causal structure [3]. In our DAG framework, MI subtype lies on the causal pathway between sex and subsequent management as it is partly determined by sex and influences care. Treating MI subtype as confounder would therefore be inappropriate. Conditioning on MI subtype would block part of the causal effect of interest that we aimed to estimate, potentially introducing bias. We therefore estimated marginal associations, adjusted for confounders as identified through our DAGs, which provide an appropriate measure of the overall sex disparity.

This approach aligns with our study aim which was to quantify overall differences in post-ACS care between women and men, while preserving the total effect of sex on outcomes. It also reflects clinical practice, where secondary prevention targets are not stratified by MI subtype [2].

We agree that sex differences in post-ACS care are multifactorial. Nevertheless, the consistent findings observed in our study, including lower LDL-C measurement rates, lower target achievement, and lower statin use in women across age groups and follow-up, support the presence of a clinically relevant disparity in secondary prevention.

Jessica Schubert: Writing – orginal draft, writing – review and editing, conceptualization. Margret Leosdottir: Writing – review and editing. Maria Svensson K.: Writing – review and editing. Elin Täufer: Writing – review and editing. Bertil Lindahl: Writing – review and editing. Håkan Melhus: Writing – review and editing. Thomas Cars: Writing – review and editing. Emil Hagström: Writing – review and editing, conceptualization, supervision. Nina Johnston: Writing – review and editing.

Jessica Schubert: institutional grants and honoraria from Amgen, Arrowhead Pharma, and Pfizer outside the submitted grant. Maria Svensson K.: honoraria from Amgen, AstraZeneca, Boehringer Ingelheim, GSK, and Novo Nordisk outside the submitted work. Margret Leosdottir: institutional grants and honoraria from Amarin, Amgen, AstraZeneca, Bonnier Health Care, Novo Nordisk, and Sanofi outside the submitted work. Bertil Lindahl and Håkan Melhus: no conflicts of interest to disclose. Elin Täufer: institutional grants from Amgen and honoraria from Sanofi, Boehringer Ingelheim, and Novartis outside the submitted work. Thomas Cars is an employee and shareholder in Sence Research AB (an analytical company in epidemiology and biostatistics). Emil Hagström: institutional grants from Amgen, Pfizer, and honoraria from Amgen, Novo Nordisk, Bayer, and AstraZeneca outside the submitted work.