Differences in relative and absolute effectiveness of oral P2Y ₁₂ inhibition in men and women: a meta-analysis and modelling study

Figure 1. Overview of data sources used in modelling. Panel A shows the data sources used to estimate the relative treatment efficacy of newer P2Y12 inhibitors in women and men and panel B shows the data sources used to estimate the absolute treatment effect. PLATO, Platelet Inhibition and Patient Outcomes Trial; TRILOGY ACS, Targeted Platelet Inhibition to Clarify the Optimal Strategy to Medically Manage Acute Coronary Syndromes; TRITON-TIMI 38, Trial to Assess Improvement in Therapeutic Outcomes by Optimising Platelet Inhibition with Prasugrel-Thrombolysis in Myocardial Infarction 38.

Figure 2. Relative treatment efficacy for cardiovascular events stratified by sex. Rate ratios and 95% credible intervals estimates of sex-specific treatment effects for each trial. Estimates for each trial were estimated independently in generalised linear models. Points represent the point estimates, and bars represent the 95% credible intervals. ACS, acute coronary syndrome.

Figure 3. Relative treatment efficacy for cardiovascular events. Meta-estimates and 95% credible intervals estimates of sex–treatment interactions for each trial as well as the overall sex–treatment interaction for P2Y12 inhibitors. All estimates were obtained from generalised linear models. The overall sex–treatment interaction was estimated in a hierarchical model. Points represent the point estimates, and bars represent the 95% credible intervals. The point-size for each trial is proportional to the inverse of the SE squared for each interaction estimate. The vertices of the diamond indicate the 95% credible intervals for the meta-estimate, which was obtained from the ‘shared’ model. ACS, acute coronary syndrome.

Figure 4. Baseline risk of death and estimates of the absolute treatment effects of oral P2Y12 inhibition in women and men with myocardial infarction in Scotland. Panel A shows estimated risks of deaths obtained from a regression model run on the national Scottish data. A generalised linear model with a log-link and multinomial distribution was fitted to all three outcomes (cardiovascular death, bleeding death and non-cardiovascular, non-bleeding death), which included parameters for age and sex as well as age–sex interactions. Panel B shows the modelled absolute risk reduction in this population, obtained from applying to this model the relative treatment efficacy for cardiovascular events and bleeding events in men and women. Most patients in the clinical trial data were aged less than 75 years, and we were unable to examine whether treatment effects showed heterogeneity by age. As such, estimates for older patients should be treated with caution.