Abstract
Smoking is a myocardial infarction (MI) risk factor among people with HIV (PWH). Questions persist regarding the role of smoking behaviors and measurements (e.g., intensity, duration) on MI risk. We used Cox proportional hazards regression to compare the association of smoking parameterization with incidents of type 1 and type 2 MI and whether smoking intensity or duration improves MI risk prediction among PWH. Among 11,637 PWH, 37% reported currently smoking, and there were 346 MIs. Current smoking was associated with type 1 (84% increased risk) but not type 2 MI in adjusted analyses. The type 1 MI model with pack years had the best goodness of fit compared with other smoking parameterizations. Ever or never parameterization and smoking diagnosis data had significantly poorer model fit. These results highlight the importance of differentiating MI types and performing patient-based smoking assessments to improve HIV care and research rather than relying on smoking status from diagnoses.