Determining Diabetes Prevalence: a Rational Basis for the Use of Fasting Plasma Glucose Concentrations?

Abstract

The World Health Organization and the National Diabetes Data Group each recommend a diagnostic cut-off point for diabetes of 7.8 mmol l⁻¹ for fasting plasma glucose concentrations as part of the diagnostic criteria for epidemiological studies. However, this cut-off has been shown to be insensitive compared with a screening test based on 2-h plasma glucose levels. In thirteen Pacific populations, from four ethnic groups (Asian Indian, Melanesian, Micronesian, and Polynesian), we have examined whether a different cut-off point for fasting plasma glucose would be more accurate for obtaining an estimate of the prevalence of diabetes when compared with 2-h levels. A fasting plasma glucose diagnostic cut-off of 7.0 mmol l⁻¹ gave an estimate of prevalence not significantly different from that based on the 2-h plasma glucose in 12 of the 13 populations (mean difference 0.27, range −1.51 to +2.44,%). On the other hand, when a cut-off of 7.8 mmol l⁻¹ for fasting plasma glucose was used, the resulting prevalence over-estimated the 2-h glucose prevalence in all populations (mean difference 1.91, range 0.14–5.80,%). Thus for Pacific populations, a fasting plasma glucose cut-off of 7.0 mmol l⁻¹ provides estimates of prevalence that are equivalent to those based on 2-h plasma glucose levels. In epidemiological studies designed to estimate diabetes prevalence, we recommend use of a fasting plasma glucose cut-off of 7.0 mmol l⁻¹ in preference to a detection level of 7.8 mmol l⁻¹, if glucose loading is not possible. The low sensitivity associated with the 7.0 mmol l⁻¹ cut-off however, precludes its use as a definitive screening test for diabetes, particularly in the clinical setting.