Background and aim: The American Diabetes Association (ADA) recommends basing diabetes diagnosis on a fasting plasma glucose (FPG) of > or = 7.0 mmol/L and impaired fasting glucose (IFG) on 6.1 < or = FPG < 7.0 mmol/L. The new World Health Organisation (WHO) recommendations also adopt this FPG cut-off, but retain the oral glucose tolerance test (OGTT) where possible and the intermediate group of impaired glucose tolerance (IGT) in addition to IFG. We compare the effect of the new ADA and WHO diagnostic criteria in three ethnic groups.

Methods and results: Three hundred and eighty whites, 340 South Asians and 347 subjects of African descent, aged 40-59 years and not known to have diabetes, were identified through South London general practices. Inevitably, the prevalence of new diabetes was lower under ADA than under WHO criteria (including post-load levels) for all three groups, falling from 5.7% overall to 3.3% (fall 2.4% 95% CI 1.6% to 3.6%). The largest fall was for South Asians from 9.1% to 5.0% (fall 4.1% 95% CI 2.2% to 6.8%). The prevalence of impaired glucose homeostasis under ADA criteria (IFG) was substantially less than under WHO criteria (IFG + IGT). Under WHO criteria, including a glucose tolerance test, there was marked variation by ethnic group in diabetes prevalence (p < 0.001) and IGT (p < 0.0001), both were most prevalent amongst South Asians. Under ADA criteria, (or new WHO criteria without OGTT) diabetes prevalence still differed significantly between groups (p < 0.01), but there was no difference in IFG prevalence (p = 0.43).

Conclusions: Subjects with IGT but normal FPG are at greater risk of coronary heart disease. The new ADA definition fails to identify substantial numbers of such subjects, particularly among South Asians. Our study supports the retention of the OGTT in the new WHO criteria, particularly for South Asians.