To explore the association between body mass index (BMI) and mortality in the middle-aged and elderly population of Beijing.
A prospective cohort study was carried out in a natural population (aged 40 and over) in September 1991, totaling 6 209 subjects. After a baseline survey for risk factors, all the subjects were followed-up throughout December 1999. All deaths occurred during the follow-up period were registered and encoded according to ICD-9. Proportional hazard regression was used to examine the relationship between BMI (body weight in kilograms/height in square meters) and overall and cause-specific mortality.
During the past ten years, five leading causes of death in the middle-aged and elderly population of Beijing were cardiovascular disease (CVD), cor pulmonale, digestive malignant tumor, sudden death without definite cause and lung cancer, which accounting for 33.3%, 13.8%, 11.2%, 7.4% and 5.6% of the total deaths, respectively. The minimum overall mortality in all population and in non-smokers were seen in those with BMI of 23.6 - 26.2. Compared with those with BMI of 23.6 - 26.2, the relative risks (RR) for all-cause mortality in all population and non-smokers with BMI less than 21.2 and equal to or greater than 26.2 were 1.7, 1.3, 1.9 and 1.5, respectively after adjusted for age, gender, smoking and alcohol drinking. BMI correlated positively with death risk due to CVD in non-smokers, with an RR of 1.2 for those with BMI > 26.2 to non-smokers with BMI < 21.2 after adjusted for age, gender and alcohol drinking. A U-shaped correlationship between BMI and deaths due to other specific causes, including cor pulmonale, digestive malignant tumor, lung cancer, pneumonia and sudden death without definite causes. With the minimum cause-specific death rates all at BMI of 23.6 - 26.2. In comparison with those with BMI of 23.6 - 26.2, RRs for these five cause-specific mortality were 14.8, 1.1, 5.6, 2.4 and 2.6, respectively for non-smokers with BMI < 21.2 and 3.0, 1.9, 3.7, 2.2 and 1.5, respectively for non-smokers with BMI >/= 26.2 after adjusted for age, gender, and alcohol drinking. The proportion of deaths due to CVD was the highest in each BMI group, with a range of 28.0% - 54.8%.
There exists a U-shaped correlationship between BMI and overall mortality in the middle-aged and elderly population in Beijing. CVD remains the top leading cause of death in natural population, and its death risk increased with BMI.