The threshold for definition of impaired fasting glucose in a Japanese population

Aims We examined whether the cut-off value of fasting plasma glucose (FPG) for diagnosing impaired fasting glucose (IFG) should be lowered, using data from a large Japanese population. Methods A retrospective cohort study was conducted from 1998 to 2006. Follow-up (2002-2006) data were merged with baseline (1998-2002) data, yielding 11 129 persons who had participated on both occasions. Among these, 10 475 persons who did not have diabetes (known diabetes or defined as FPG >= 7.0 mmol/l) or suspected diabetes (glycated haemoglobin >= 6.4%) were analysed. Results During follow-up of an average of 5.4 years, 279 (5.2%) out of 5372 men and 98 (1.9%) out of 5103 women developed diabetes. According to the three baseline FPG categories (< 5.6, 5.6-6.1 and 6.2-6.9 mmol/l), 28/3401 (0.8%), 91/1456 (6.3%) and 160/515 (31.1%), respectively, in men and 13/4231 (0.3%), 30/695 (4.3%) and 55/177 (31.1%), respectively, in women developed diabetes. The optimal cut-off FPG value to predict diabetes was 5.7 mmol/l for both men (sensitivity 84.2%, specificity 76.9%) and women (81.6%, 91.0%). However, lowering the cut-off from 6.1 to 5.7 mmol/l increased the prevalence of IFG 2.7-fold in men and 3.0-fold in women. Lowering the value further to 5.6 mmol/l increased the prevalence of IFG 3.8-fold in men and 4.9-fold in women. Conclusions It may be reasonable to retain the conventional lower FPG limit for IFG and treat FPG values of 5.6-6.1 mmol/l as non-diabetic hyperglycaemia, considering the four- to fivefold increase in individuals classified as IFG when the new cut-off is applied.