Association between baseline BMI and in-hospital mortality in critically ill cardiac surgery patients: a retrospective cohort study

BMI–mortality associations in critically ill cardiac surgery lack precise threshold characterization. We investigated the baseline BMI and in hospital mortality in ICU cardiac surgery patients. We included 8032 cardiac surgery patients from eICU (208 US hospitals, 2014–2015). Focus exposure: BMI; key outcome: 30-day mortality. Adjustments encompassed demographics, laboratory values, APACHE-IV scores, comorbidities, and interventions. Nonlinear patterns were examined via penalized spline-fitted generalized additive modeling. Among 8,032 patients (BMI: 29.6 ± 6.8 kg/m2), in-hospital mortality occurred in 289 cases (3.6%). Standard linear modeling detected no BMI–mortality relationship (OR = 0.99, 95% CI 0.97–1.01, P = 0.210). Conversely, spline analysis identified a U-shaped pattern with inflection at 25.0 kg/m2. BMI increases below this point conferred 10% mortality reduction per unit (OR = 0.90, 95% CI 0.83–0.96, P = 0.003), while higher values showed null association (OR = 1.01, 95% CI 0.98–1.03, P = 0.567). Log-likelihood ratio testing favored the nonlinear approach (P = 0.010). Our analysis identified a U-shaped BMI–mortality relationship in critically ill cardiac surgery patients, revealing an optimal risk threshold at 25.0 kg/m2. These findings may provide valuable guidance for perioperative BMI management in this population.