A risk model for decline in health status after acute myocardial infarction among older adults

Abstract Background Health status is increasingly recognized as an important patient‐centered outcome after acute myocardial infarction (AMI). Yet drivers of decline in health status after AMI remain largely unknown in older adults. We sought to develop and validate a predictive risk model for health status decline among older adult survivors of AMI. Methods Using data from a prospective cohort study conducted from 2013 to 2017 of 3041 patients age ≥75 years hospitalized with acute myocardial infarction at 94 U.S. hospitals, we examined a broad array of demographic, clinical, functional, and psychosocial variables for their association with health status decline, defined as a decrease of ≥5 points in the Short Form‐12 (SF‐12) physical component score from hospitalization to 6 months post‐discharge. Model selection was performed in logistic regression models of 20 imputed datasets to yield a parsimonious risk prediction model. Model discrimination and calibration were evaluated using c‐statistics and calibration plots, respectively. Results Of the 2571 participants included in the main analyses, 30% of patients experienced health status decline from hospitalization to 6 months post‐discharge. The risk model contained 14 factors, 10 associated with higher risk of health status decline (age, pre‐existing AMI, pre‐existing cancer, pre‐existing COPD, pre‐existing diabetes, history of falls, presenting Killip class, acute kidney injury, baseline health status, and mobility impairment) and four associated with lower risk of health status decline (male sex, higher hemoglobin, receipt of revascularization, and arrhythmia during hospitalization). The model displayed good discrimination (c‐statistic = 0.74 in validation cohort) and calibration ( p > 0.05) in both development and validation cohorts. Conclusions We used split sampling to develop and validate a risk model for health status decline in older adults after hospitalization for AMI and identified several risk factors that may be modifiable to mitigate the threat of this important patient‐centered outcome. External validation of this risk model is warranted.