We have investigated the total prevalence of diabetes and related factors among representative, randomly chosen samples of the five ethnic groups living in Cape Town, and (East) Indians in Durban. Comparisons are hindered by differences in age distribution of the populations, while small, isolated groups were found to be unrepresentative. The variability of a single individual's blood sugar levels led us to require at least three abnormal values on 2 different days for a positive diagnosis. The use of different criteria for the diagnosis of diabetes varying from ‘lax’ to ‘stringent’ alters the discovered prevalence in our groups by the factor of approximately 2.
Mean blood glucose levels rose with middle age but never between childhood and early adulthood. Afternoon screening tests appeared valid, despite the agreed diurnal difference in glucose tolerance figures.
Both high screening blood glucose levels and diabetes itself were most common among Indians and coloured people and least among Whites and Bantu, each of the latter having a total diabetes prevalence of approximately 3·5% over age 15. It is noted that the Cape Coloured have more diabetes than any of the constituent races from which they originated.
The reasons for such racial differences are unclear—obesity cannot be the explanation here, since, to take one example, the fattest group of all, the Bantu women, have the lowest prevalence of diabetes. We found mild diabetes not uncommon among young people under 20 in the Indian, Malay and coloured population but none among White or Bantu. There was little difference between the sexes, and if anything the poorer people had more diabetes than the better-off.