Physical activity and incidence of diabetes: the Honolulu Heart Program

Few prospective studies have assessed the relation between physical activity and diabetes. The authors examined this relation prospectively among 6,815 Japanese-American men in the Honolulu Heart Program who were aged 45-68 years and initially free of diagnosed diabetes in 1965-1968. A physical activity index was calculated based on time spent per day in different activity levels and a weighting factor correlated with estimated oxygen consumption. Incidence of clinically recognized diabetes was based on self-reported use of diabetic medication at one of two subsequent examinations. The age-adjusted 6-year cumulative incidence of diabetes decreased progressively with increasing quintile of physical activity from 73.8 to 34.3 per 1,000 (p < 0.0001, trend) in all men and from 53.9 to 21.7 per 1,000 (p < 0.0001, trend) among men with a non-fasting glucose level < 225 mg/dl one hour after a 50-gm load, the latter group being less likely to have unrecognized diabetes at baseline. When stratified by tertile of baseline glucose, trends in incidence across physical activity quintiles were statistically significant in the low and middle tertiles but not in the high tertile. Similar inverse trends were observed for men in the lower four quintiles of body mass index, however, these trends were weaker and not significant for men in the upper quintile of body mass index. Age-adjusted odds ratios for diabetes comparing the upper with the lower four quintiles of physical activity were 0.55 (95% confidence interval (CI) 0.41-0.75) for all men and 0.50 (95% CI 0.33-0.74) for men with glucose < 225 mg/dl. After adjustment for age, body mass index, subscapular/triceps skinfold ratio, systolic blood pressure, triglycerides, glucose, hematocrit, and parental history of diabetes, odds ratios were still statistically significant and similar in magnitude. Restriction of analyses to men who remained free of cardiovascular disease during the study period produced similar results, which suggests that inactivity due to subclinical cardiovascular disease is unlikely to be responsible for these findings. Risk factor-adjusted odds ratios for older men (55-68 years) demonstrated that physical activity confers at least the same degree of protection as in younger men (45-54 years). These results indicate that physical activity is associated inversely with incident diabetes and that the beneficial effect does not appear to be mediated through improvements in other risk factors assessed in this study.