Variability and misclassification of albuminuria in patients with type 2 diabetes mellitus.

Measurement of albuminuria is central in assessment, staging and evaluation of treatment response of diabetic kidney disease. Due to high intra-individual variability of albuminuria, guidelines recommend specific sampling intervals and usually agreement of two out of three samples for correct staging. Nevertheless, different confirmation and staging strategies have been proposed for albuminuria assessment. We report albumin-to-creatine ratio measurements of spot urine samples in a modern European cohort (n = 773) of type 2 diabetes mellitus patients in a primary care setting. Information on albuminuria, its variability and associated baseline characteristics are presented. The amount of "misclassification" of alternative staging strategies compared to the guideline recommended gold standard is reported. Additionally, we applied quantile regression to estimate the conditional quantiles of the second and third measurement, allowing an estimation of likely variability at a given albuminuria level. Overall, the coefficient of variation of albumin-to-creatinine ratio, as a measure of variability, is high with a median of 41%. This variability leads in part to substantial disagreement of alternative staging strategies with the gold standard. In general, higher variability of albuminuria seems to be associated with more severe kidney disease. The high variability of albuminuria necessitates caution and consideration for adequate clinical use. In this study, we provide data regarding the expected variability, correct usage and limitations of albuminuria in a modern primary care diabetes mellitus population.