Aim
To determine if an HbA1c diagnostic threshold of less than 6.5% (<48 mmol/mol) could be identified based on a urinary albumin‐creatinine ratio (UACR) of 30 mg/g or higher in subjects not known to have diabetes.
Methods
A UACR was measured for 20 158 participants in the 2011‐2018 nationally representative cross‐sectional National Health and Nutrition Examination Surveys (NHANES; cycles 7‐10 inclusive).
Results
There was a significant trend for an increasing risk with a UACR of 30 mg/g or higher across increasing HbA1c categories (P < .0001). This trend was mainly attributable to the high prevalence of raised UACR in the 7.0% or higher HbA1c subgroup of subjects not previously diagnosed with diabetes. None of the odds ratios in the lower HbA1c subgroups versus the HbA1c subgroup of less than 5.0% reached significance. There were racial/ethnic differences in UACR risk (P < .0001), with White and Black subjects exhibiting little increased risk (vs. HbA1c <5.0%) until they reached an HbA1c of 7.0%, while Asian and Hispanic subjects showed some increased, but non‐significant, risks at lower HbA1c levels. Maximizing the area under receiver operating characteristic curves from logistic regressions predicted an ideal HbA1c threshold of 5.8%, but there was little variation in area from 5.5% to 7.0%.
Conclusion
A clinically useful diagnostic threshold below 6.5% for HbA1c for elevated UACR risk was not identified, with an increased risk only obvious at an HbA1c of 7.0% or higher. Thus, the retinopathy‐derived HbA1c threshold of 6.5% also captures the risk of diabetic nephropathy in NHANES.