Non-linear association between body mass index and sudden cardiac death after myocardial infarction: a PROFID pooled cohort sub-analysis

Background The relationship between body mass index (BMI) and sudden cardiac death (SCD) after myocardial infarction (MI) remains incompletely characterised. While BMI demonstrates a J-shaped association with SCD in the general population, data specific to post-MI patients remain limited. Purpose We examined whether BMI is non-linearly associated with long-term SCD risk in a large international pooled cohort of post-MI patients. Methods This analysis included 121,481 post-MI patients from the PROFID pooled dataset with available BMI, survival time, and SCD data. BMI was modelled as a continuous variable using restricted cubic splines to assess potential non-linear associations with SCD risk. A BMI of 25 kg/m² was used as the reference value for hazard ratio estimation. In non-ICD patients, the primary outcome was SCD within 90 months following index MI. In ICD carriers, the outcome was appropriate ICD-therapy as the best available surrogate for SCD. Multivariable cause-specific Cox models were adjusted for demographic and clinical covariates. Model robustness was evaluated using alternative spline specifications, fractional polynomial modelling, proportional hazards diagnostics, competing risk analyses (Fine–Gray), and complete-case comparisons. Results During follow-up, 2,865 SCD events (2.4%) occurred. In multivariable cause-specific Cox models, BMI demonstrated a significant non-linear association with 90-month SCD risk (global spline p<0.001). The lowest estimated hazard occurred at BMI 22.8 kg/m². Relative to BMI 25 kg/m², SCD risk increased progressively with higher BMI, reaching approximately a hazard ratio of 1.45 at BMI 35 kg/m². Lower BMI values were associated with only a modest excess hazard. The extended model demonstrated good discrimination (C-index 0.815) and excellent calibration. The non-linear association remained consistent across alternative spline specifications, fractional polynomial modelling, proportional hazards sensitivity analyses, and competing risk analysis using Fine–Gray models. Conclusions In this large multinational post-MI cohort, BMI demonstrated a non-linear association with long-term SCD risk, with the lowest risk observed at BMI 22.8 kg/m² and progressively increasing risk at higher BMI values. These findings contribute to a better understanding of the association between BMI and SCD in post-MI populations.