Background: Obesity is a risk factor for incident heart failure, but patients with excess weight and heart failure have lower mortality. This "obesity paradox" may be explained either by a favourable effect of the adipose tissue or by confounding. We aimed to assess if body mass index (BMI) is associated with lower mortality after extensive adjustment for prognostic factors in patients with reduced (HFrEF) or preserved (HFpEF) ejection fraction.

Methods: Prospective, observational study including consecutive patients hospitalized for acute heart failure. Two years hazard of mortality was assessed in a multivariable Cox model, in the whole population and separately for HFrEF and HFpEF.

Results: Among 957 included patients (mean age 76 years, 41% women), 500 (47%) had HFrEF and 443 (53%) HFpEF. Four hundred (39%) were in the normoweight, 301(30%) in the overweight, and 256(25%) in the obese category (Class I obesity:144 patients; class II or III: 112). Corresponding mortality was 37%, 26% and 22%. Unadjusted hazard ratio (HR) for mortality was 0.96 (95% CI 0.94-0.98) for each BMI point in the whole population, 0.97 (95% CI 0.94-1.02) in patients with HFrEF, and 0.94 (95% CI 0.92-0.97) in HFpEF. After adjustment for age, sex, atrial fibrillation, diabetes, chronic obstructive pulmonary disease, chronic anaemia, hypertension, glomerular filtration rate, and NT-proBNP, HR was 1.00 (95% CI 0.96-1.02) in the whole population, 1.02 (0.96-1.07) in HFrEF, and 0.98 (95% CI 0.94-1.01) in HFpEF.

Conclusions: Excess weight was associated with an apparent survival benefit in patients with acute heart failure, particularly in patients with HFpEF. This advantage disappeared completely after adjustment for confounding factors including NT-proBNP. The obesity paradox can be completely explained by differences in demographics, co-morbidities, and severity of heart failure.