Selecting the optimum specimen for assessing slight albuminuria, and a strategy for clinical investigation: novel uses of data on biological variation.

To investigate the optimum specimen for quantifying low, but abnormal, concentrations of albumin in urine, we assessed the analytical and biological components of variation in first morning, random untimed, and 24-h urine specimens from 11 apparently healthy individuals. The results were expressed in terms of albumin concentration, albumin/creatinine ratio, and albumin excretion rate. Analytical methods generally can meet the analytical goal of CV less than or equal to 18%. For reasons detailed herein, we prefer measurement of the albumin concentration in the first morning specimen. Expressing results as an albumin/creatinine ratio has little advantage. Albumin concentrations in first morning urines from 16 diabetic subjects showed larger intra-individual variation than for nondiabetic subjects but clearly fell into two groups: those with consistently normal albumin concentrations in urine and those with abnormal concentrations in some specimens. The intrinsic biological variation of the latter group means that the ideal 100% nosological specificity cannot be achieved with any cutoff point without inclusion of a large proportion of the former. Qualitative testing with a latex-agglutination technique also demonstrates this problem. Use of data on biological variation allows development of an appropriate clinical strategy to investigate diabetic patients.