Validation of HbA1c of 6.5% for diagnosing diabetes mellitus via the use of taxometric analysis.

Purpose: Previous studies in which authors examined the internal and external validity of a glycemic cut-point for diagnosis of diabetes mellitus have provided mixed results. The purpose of the current study was to test the internal validity of the HbA1c 6.5% cut-point with taxometric analysis.

Methods: Data on 14,798 participants were obtained from the National Health and Nutrition Examination Survey, years 1999-2008. Fasting plasma glucose and HbA1c were submitted to the taxometric procedures MAMBAC (mean above minus below a cut) and MAXSLOPE (MAXimum SLOPE). The comparison curve fit index was the outcome measure, with values less than 0.40 and greater than 0.60 indicative of dimensional and categorical solutions, respectively.

Results: In the full sample, MAXSLOPE and MAMBAC procedures yielded CCFIs of 0.778 and 0.872, respectively. Analyses were repeated in subgroups by age, sex, fasting insulin level, ethnicity, and year; the lowest comparison curve fit index measurement from any analysis was 0.706.

Conclusions: These results support a categorical overdimensional model of diabetes, consistent with the new HbA1c cut-point recommendation.