Prevalences of diabetes and cardiovascular disease risk factors in Hindu Indian subcommunities in Tanzania.

OBJECTIVES To seek differences in the prevalence of diabetes mellitus and other coronary heart disease risk factors, and to identify factors associated with these differences within a Hindu Indian community. DESIGN Population based cross sectional survey. SETTING Dar-es-Salaam, Tanzania. SUBJECTS Of 20 Hindu subcommunities categorised by caste in Dar-es-Salaam, seven were randomly selected. 1147 (76.7%) of 1495 subjects aged 15 or over participated. MAIN OUTCOME MEASURES Blood glucose concentrations (fasting and two hours after oral glucose loading), serum total cholesterol and serum triglyceride concentrations, blood pressure, and height and weight. RESULTS The subcommunities differed substantially in socioeconomic characteristics and lifestyle. Overall, 9.8% of subjects (109/1113) had diabetes, 17.0% (189/1113) impaired glucose tolerance, 14.5% (166/1143) hypertension, and 13.3% (151/1138) were obese. The mean fasting blood glucose concentration was 4.9 mmol/l, the blood glucose concentration two hours after oral loading (75 g) 6.0 mmol/l, the total cholesterol concentration 4.9 mmol/l, the serum triglyceride concentration 1.4 mmol/l, and body mass index (weight/height: kg/m2) 24.3. Systolic and diastolic blood pressures were 121 and 77 mm Hg respectively. There were important intercommunity differences even after standardisation for age, sex, and body mass index--for example, in mean fasting blood glucose concentration (range 4.5 (Jains) to 5.9 mmol/l (Patels)), serum total cholesterol concentration (range 4.5 (Jains) to 6.2 mmol/l (Suthars)), systolic blood pressure (range 110 (Limbachias) to 127 mm Hg (Bhatias)), and prevalences of diabetes (range 3.4% (3/87 Limbachias) to 18% (20/111 Navnats)) and hypertension (range 5.7% (5/87 Limbachias) to 19.4% (43/222 Bhatias). Variables which showed significant linear correlation with subcommunity variations were entered into a multiple regression model. Intercommunity variations persisted. The Limbachia and Jain communities had the lowest prevalence of and mean values for coronary heart disease risk factors and the Bhatia and Patel communities had the highest. CONCLUSIONS In this series intercommunity variations in disease and risk factors might have been related to genetic, dietary, socioeconomic, and lifestyle differences but could not be explained by the characteristics studied. Studies of Indian subcommunities are warranted to confirm and extend these descriptive findings and explore the genetic basis of diabetes. Communities of Indian origin should not be perceived as homogeneous.