Increases in body mass index over a 7-year period and risk of cause-specific mortality in Korean men

Methods We conducted a retrospective cohort study of 473 358 Korean men aged 30-64 years, who had undergone health examinations in both 1992 and 1998 and were followed up during 1998-2004. Cox proportional hazards for cause-specific 7-year mortality in relation to BMI changes after stratification of baseline BMI status were analysed. Results Mortality from cardiovascular disease (CVD) was associated with BMI in both 1992 and 1998. Non-CVD mortality was inversely associated with BMI in both 1992 and 1998. We cross-classified participants into groups based on their baseline BMI levels and percent BMI changes during follow-up; men with the lowest BMI level at baseline (BMI in 1992 < 21 kg/m(2)) and stable BMI during follow-up (percent change in BMI < 5%) were included in the reference category. Compared with the reference group, CVD mortality was higher in initially obese men (BMI in 1992 >= 25 kg/m(2)) with any increase of BMI, and in initially lean men (BMI in 1992 < 21 kg/m(2)) or initially overweight men (BMI in 1992 23-24.9 kg/m(2)) with BMI increases of >= 10%. BMI increases of 5.0-9.9% in men with baseline BMI < 25 kg/m(2) and stable BMI in men with baseline BMI >= 21 kg/m(2) appeared to lower the risk for non-CVD or all-cause mortality, to below the levels seen in the reference group. Conclusions Among middle-aged Korean men, obesity or severe weight gain was detrimental to CVD mortality. An increase in BMI appeared to have a predictive value for CVD mortality, especially when used in combination with baseline BMI level. In contrast, moderate weight gain in non-obese men seemed to protect against non-CVD and all-cause mortality.