Measuring the Albumin Excretion Rate: Agreement Between Methods and Biological Variability

Abstract

A timed urine collection is necessary to determine the excretion rate of albumin (AER) but such specimens are tedious to collect and frequently inaccurate. Albumin excretion can also be quantified by the use of the albumin:creatinine ratio in randomly obtained specimens. In the present study the agreement between AER as measured on a 24‐h urine collection and as estimated from the albumin:creatinine ratio is determined. Previously published studies have examined the correlation rather than the agreement between these methods and not taken into account the biological variability of AER. Thirty patients with diabetes who had normal renal function, but varying degrees of albuminuria, produced two 24‐h specimens and two random daytime specimens of urine. AER was measured on the former and estimated from the latter by multiplying the albumin:creatinine ratio by an estimate of that individual's creatinine excretion rate. Agreement between the methods and the biological variability was determined by using appropriate statistical methodology, the main outcome measure being the limits of agreement between repeat values for both measurements and both estimates of AER, and between the averages of the measurements and the estimates. The limits of agreement between repeated 24‐h measurements were wide, the second specimen being 33 to 490% of the first. The estimates of AER gave values numerically similar to the measurements. The limits of agreement between the two estimates did not differ significantly from those of the measurements, nor did the limits of agreement when the average of the measurements and the average of the estimates were compared (all NS). The regression line fitted to the plot of the average of the measurements against the average of the estimates did not differ significantly from the line of identity. AER shows considerable biological variability when measured on 24‐h collections. The estimate of AER based on the albumin:creatinine ratio gives numerically equivalent values with a similar degree of variability. The imprecision of this method is clinically irrelevant compared with the magnitude of the biological variability. Determining AER in this way obviates the need for timed urine collections.