The obesity paradox in stable chronic heart failure does not persist after matching for indicators of disease severity and confounders

To verify whether controlling for indicators of disease severity and confounders represents a solution to the obesity paradox in chronic heart failure (CHF). From a cohort of 1790 patients, we formed 230 nested matched triplets by individually matching patients with body mass index (BMI) > 30 kg/m(2) (Group 3), BMI 20-24.9 k/m(2) (Group 1) and BMI 25-29.9 kg/m(2) (Group 2), according to NT-proBNP, age, sex, and NYHA class (triplet = one matched patient from each group). Although in the pre-matching cohort, BMI group was a significant univariable prognostic indicator, it did not retain significance [heart rate (HR): 0.91, 95% CI: 0.78-1.05, chi(2): 1.67] when controlled for group propensities as covariates. Furthermore, in the matched cohort, 1-year mortality and 3-year mortality did not differ significantly. Here, BMI again failed to reach statistical significance for prognosis, either as a continuous or categorical variable, whether crude or adjusted. This result was confirmed in the patients not selected for matching. NT-proBNP, however, remained statistically significant (log(NT-proBNP): HR: 1.49, 95% CI: 1.13-1.97, chi(2): 7.82) after multivariable adjustment. The obesity paradox does not appear to persist in a matched setting with respect to indicators of disease severity and other confounders. NT-proBNP remains an independent prognostic indicator of adverse outcome irrespective of obesity status.