Variability of urine albumin excretion in normal and diabetic children.

The variability of urine albumin excretion (UAE) was studied in normal and diabetic children and, in addition, the best method of expressing the data was investigated. In 39 timed overnight urine samples from diabetic children, the urine albumin creatinine clearance ratio (CA/CC) was compared with the urine albumin creatinine concentration ratio (UA/UC), the urine albumin excretion rate (UAER) and the urine albumin concentration (UA). UA/UC predicted CA/CC (r = 0.95) better than either UAER (r = 0.83, P less than 0.02) or UA (r = 0.90), 0.1 greater than P greater than 0.05). The within-individual and the between-individual variability in overnight UA/UC in 171 urine samples from 73 normal children was compared with that of 406 urine samples from 119 diabetic children, using a "random effects type 2 nested analysis of variance" model. Geometric mean (range) UA/UC (mg/mmol) in diabetic children, 0.55 (0.04-6.90), was greater than in normal children, 0.33 (0.05-2.10, P less than 0.01), and 18% of diabetics had a value of UA/UC above the normal range. Within-individual variance was the same in normals (0.12) and diabetics (0.12), but between-individual variance in diabetics (0.18) was much greater than in normals (0.03). These data show that within-individual observations for both normals and diabetics are highly but equally variable. Furthermore, from these data, it is possible to infer that a minimum of five estimations are necessary per individual to estimate the true mean value of urine albumin excretion with reasonable confidence.