Body mass index and mortality in the elderly from general population

OBJECTIVE: To evaluate at a population level whether or not BMI is a risk factor of mortality in the elderly, highlighting possible gender-related differences. METHODS: 3,282 Subjects aged 65-95 years, were recruited from an Italian general population and 12-year events were recorded. Blood tests and anthropometric measurements were performed. BMI as a continuous item was divided into quintiles and, for each quintile, adjusted hazard ratio (HR) with 95% confidence intervals for mortality was derived by classes of age and gender from Cox analysis. RESULTS: BMI inversely predicted overall and cancer mortality in men only. Overall mortality rate was 64.7% (HR = 1.63 [1.23-2.71]) in the 1st quintile of BMI, 54.9% (1.21 [0.92-1.73]) in the 2nd, 54.1% (1.20 [0.85-1.67]) in the 3rd, 53.3% (1.04 [0.82-1.32]) in the 4th and 52.5% in the 5th; cancer mortality rate was 23.1% (HR = 2.35 [1.31-4.23]), 14.2% (HR = 1.19 [0.65-1.80]), 15.8% (HR = 1.49 [0.93-2.39]), 15.8% (HR = 1.36 [0.84-2.16]) and 13.4%, respectively. The relationship between BMI and mortality remained significant only in men aged 76 years or less. No relationship was found between BMI and coronary or cerebrovascular mortality. CONCLUSIONS: BMI <22.7 kg/m2 does not improve survival in the elderly, while it is an independent predictor of cancer mortality in men aged <or=76 years. No prediction based on BMI is possible in women.