Metabolic Health Modifies the Obesity Paradox in Critically ill Patients: A Phenotype-Based Risk Stratification Tool from the MIMIC-IV Database

BackgroundThe prognostic significance of obesity in critically ill patients remains controversial, partly due to the simplistic use of body mass index (BMI) without considering metabolic heterogeneity. We aim to evaluate whether combining BMI with metabolic health status provides superior risk stratification for all-cause mortality in ICU patients using the MIMIC-IV database.MethodsThis retrospective cohort study analyzed 31 171 adult ICU patients from the MIMIC-IV database. Patients were classified into six phenotypes based on BMI categories (normal, overweight, or obese) and metabolic health status (healthy vs unhealthy, defined by hypertension, diabetes or hyperlipidemia). The primary outcomes included 7-day, 30-day, and 90-day all-cause mortality. We applied survival curve analysis and Cox proportional hazards regression, with subgroup analyses stratified by sex, age, and illness severity.ResultsAmong metabolically healthy patients, BMI category did not significantly affect mortality. However, in metabolically unhealthy patients, a clear graded protective association emerged: the metabolically unhealthy obese (MUO) phenotype conferred the strongest survival benefit (adjusted HRs: 0.63, 0.65 and 0.64 for 7-day, 30-day and 90-day mortality, respectively), followed by metabolically unhealthy overweight (MUOW, adjusted HRs: 0.73, 0.70 and 0.69 for 7-day, 30-day and 90-day mortality), while the metabolically unhealthy normal weight (MUNW) patients showed the weakest protection. Such a survival benefit was significantly amplified in males and in patients with higher illness severity.ConclusionsAmong ICU patients, metabolic health status, rather than BMI alone, is the key determinant of prognosis. For metabolically unhealthy patients, higher BMI was associated with a graded survival advantage, particularly in those with greater illness severity. The proposed six-phenotype classification provides a simple, admission-available bedside tool for early risk stratification and may inform individualized nutritional and metabolic management in critical care.