Physical activity and diabetes risk in postmenopausal women

To evaluate the hypothesis that physical activity independently predicts type 2 diabetes risk in postmenopausal African-American, Hispanic, Asian, and Caucasian women. We prospectively evaluated the relationship between incident type 2 diabetes, walking, and total physical activity at baseline in the Women's Health Initiative Observational Study. Baseline data were collected between September 1994 and December 1998; incident diabetes was identified through August 2002. Hazard ratios for self-reported diabetes adjusted for body mass index (BMI) and other variables were evaluated across categories of physical activity in Caucasian, African-American, Hispanic, and Asian/Pacific Islander women. Incident diabetes was reported by 2.2% of Caucasian, 6.2% of African-American, 4.5% of Hispanic, 3% of Asian, and 5.7% of American Indian women (p <0.0001 across ethnic groups) during 458,018 woman-years of follow-up. Among Caucasian women, walking (multivariate-adjusted hazard ratios 1.00, 0.85, 0.87, 0.75, 0.74; p <0.001 for trend across exercise quintiles) and total physical activity score (hazard ratios 1.00, 0.88, 0.74, 0.80, 0.67; p =0.002) demonstrated a strong inverse relationship with diabetes risk. In BMI-adjusted models, African-American women in higher physical activity categories were less likely to develop diabetes than women in the lowest physical activity category. After adjusting for age and multiple risk factors, however, no significant association between physical activity and diabetes risk was apparent for African-American, Hispanic, or Asian women. These findings suggest a stronger and more independent association of physical inactivity with development of diabetes in Caucasian women than in minority women, but could also be explained by less precise risk estimates in minority women or the role of chance.