Abstract
Objective: A difference in opinion remains regarding appropriate criteria for diagnosing prediabetes. We examined the risk for incident diabetes in prediabetes using HbA1c and FPG.
Methods: In the period 2007-2010, 4,698 subjects from the Takaoka Health Care Center who had no history of diabetes (HbA1c < 6.5% and FPG < 126mg/dL) were followed up for an average of 6.4 years. They were divided into seven groups according to the results of baseline diagnosis of prediabetes. Diabetes was diagnosed by JDS criteria (FPG ≥ 126 mg/dL and HbA1c ≥ 6.5% or clinician-diagnosed diabetes).
Results: Optimal cutoff points for incident diabetes by receiver operating characteristic curve analysis were FPG 103 mg/dL and HbA1c 5.65%. Multivariate-adjusted hazard ratios for incident diabetes were 2.87 (95% CI 0.86-9.63) in those with FPG 100-125 mg/dL and HbA1c ≤ 5.5%, 8.26 （2.31-29.47）in those with FPG ≤99 mg/dL and HbA1c 5.6-6.4%, 11.26 （3.62-34.97）in those with FPG 100-109 mg/dL and HbA1c 5.6-5.9%, 52.12（15.93-170.52）in those with FPG 100-109 mg/dL and HbA1c 6.0-6.4%, 49.00（15.83-151.65）in those with FPG 100-125 mg/dL and HbA1c 5.6-5.9% and 99.95 （32.75-298.93）in those with FPG 110-125 mg/dL and HbA1c 6.0-6.4%, compared with normoglycemic individuals.
Conclusions: The combination of FPG 100-125 mg/dL and HbA1c 5.6-6.4% was effective in screening for very high risk prediabetes subjects requiring interventions.
