Heterogeneous impact of body mass index on in-hospital mortality in acute heart failure syndromes: An analysis from the ATTEND Registry.

BACKGROUND

Although the obesity paradox may vary depending upon clinical background factors such as age, gender, aetiology of heart failure and comorbidities, the reasons underlying the heterogeneous impact of body mass index (BMI) on in-hospital cardiac mortality under various conditions in patients with acute heart failure syndromes (AHFSs) remain unclear.

METHODS

Among 4617 hospitalised patients with AHFSs enrolled in the Acute Decompensated Heart Failure Syndromes (ATTEND) registry, the patient characteristics and in-hospital cardiac mortality rates in those with low BMI (BMI <25 kg/m2, n = 3263) were compared to those with high BMI (BMI ⩾25 kg/m2, n = 1354).

RESULTS

Compared to the high-BMI group, the low-BMI group was significantly older, less likely to be male and to have hypertensive or idiopathic dilated aetiologies and more likely to have valvular aetiologies and a history of prior hospitalisation for AHFS. The low-BMI group also had lower prevalence rates of diabetes, dyslipidaemia, hypertension and atrial fibrillation and higher prevalence rates of anaemia and chronic obstructive pulmonary disease. In addition, cardiac mortality was significantly higher in the low-BMI group than in the high-BMI group (5.5 vs. 1.5%, p < 0.001). Logistic regression analysis demonstrated that low BMI was a predictor of cardiac mortality (odds ratio: 3.89, 95% confidence interval: 2.44-6.21). In subgroup analyses, the impact of BMI on cardiac mortality differed depending on the presence of hypertensive aetiology, hypertension, chronic obstructive pulmonary disease and hyponatremia (all p < 0.05), although there were no interactions between the impacts of BMI and age, gender, other aetiologies, prior hospitalisation, diabetes, anaemia, cardio-renal function and in-hospital management.

CONCLUSIONS

It is necessary to appreciate the obesity paradox in AHFS patients, and a patient's heterogeneous background should also be considered.