Influence of renal biomarker variability on the design and interpretation of occupational or environmental studies

Objectives: To quantify and identify sources of within- and between-subject variability of microalbumin, N-acetyl-beta-D-glucosaminidase (NAG) and ala nine aminopeptidase (AAP), three biomarkers used for early detection of ren al injury, and to assess the consequences of this variability far the desig n and power of epidemiological studies. Methods: Urinary excretion of microalbumin, NAG, AAP and creatinine as well as blood pressure (BP) were measured three times over a 2-year period amon g 142 healthy male workers. To minimise physiopathological and analytical s ources of variation, standardised methods were used for urine collection an d assays, and severe exclusion criteria were applied. At the first and thir d examinations, subjects completed the same questionnaire, providing inform ation about their personal characteristics, tobacco and alcohol consumption , and health. A linear mixed model was used to estimate the within- and bet ween-subject variance components and to analyse the relation between subjec ts' characteristics and the biomarkers. Results: No change, in the mean value of any of the biomarkers was observed over the 2-year period. Intra-class correlation coefficients between repea ted measurements were 0.53, 0.57 and 0.56 for microalbumin, NAG and AAP, re spectively; the between-subject variance was slightly higher than the withi n-subject variance. Subjects' age, BP, body mass index and smoking and drin king habits explained 7.2%, 12.5% and 4.2% of the total variance of microal bumin, NAG and AAP, respectively. Conclusions: In this healthy population of male workers, day-to-day differe nces in biomarker values appeared to be nearly as great as differences betw een subjects. The within-subject variance of these biomarkers is not high e nough to justify systematic repeated measurements in epidemiological survey s. But, in some situations where the number of subjects is limited, measuri ng the subjects twice may improve study power by reducing the total varianc e by about 25% for each biomarker. Taking the above covariates into account would slightly improve study power and the accuracy of parameter estimates for NAG, but would add little to the analysis of microalbumin and AAP. (C) 1999 Elsevier Science Ireland Ltd. All rights reserved.