Development of Cardiovascular Risk Equations in People with Overweight or Obesity and Established Cardiovascular Disease Without Diabetes Based on the SELECT Trial

Abstract

Background: Overweight and obesity is a prevalent and growing global health concern associated with a significant healthcare burden. With recent advancements in weight management interventions demonstrating cardioprotective benefits, there is a need for risk equations that can accurately predict cardiovascular event risk in health economic models to support healthcare decision making and resource allocation.

Objective: We aimed to derive risk equations using SELECT trial data that estimate the risk of acute coronary syndrome and stroke in people with established cardiovascular disease and overweight or obesity but without diabetes mellitus for use in health economic modelling.

Methods: Risk equations estimating first in-trial observed acute coronary syndrome and stroke were derived from patient-level data from the SELECT trial, a double-blind randomised placebo-controlled trial comparing semaglutide 2.4 mg with placebo. Risk factors were identified from the literature and by clinical experts. Risk equations were developed using cause-specific Cox proportional hazard models with all-cause mortality as a competing risk. Least absolute shrinkage and selection operator (LASSO) penalisation was applied 100 times in bootstrap samples to refine the risk equations. Final risk equations were validated with clinical experts and using SELECT trial data.

Results: Sex, coronary heart disease, a history of acute coronary syndrome and use of angina medications had the strongest associations with acute coronary syndrome (hazard ratio 1.67-1.82). For stroke, a history of atrial fibrillation, cerebrovascular disorder, stroke and transient ischaemic attack had the strongest associations (hazard ratio 1.40-1.70). In terms of discrimination, the risk equations had an area under the receiver operating characteristic curve (AUC) of 0.66-0.68 for acute coronary syndrome and 0.68-0.71 for stroke, and C-indices of 0.67-0.69 for acute coronary syndrome and 0.66-0.69 for stroke. The risk equations showed good cohort-level predictive capabilities over a 4-year time horizon.

Conclusions: These novel, treatment-specific, trial-derived risk equations are the first to predict the risk of secondary cardiovascular events in people with overweight or obesity and established cardiovascular disease but without diabetes. Incorporating these risk equations into health economic models may improve the accuracy of economic evaluations in this population.