Diabetes Care, Vol 16, Issue 4 616-620, Copyright © 1993 by American Diabetes Association
Comparison of six commercial techniques in the measurement of microalbuminuria in diabetic patients
SC Tiu, SS Lee and MW Cheng
Queen Elizabeth Hospital, Kowloon, Hong Kong.
OBJECTIVE--To evaluate the advantages and drawbacks of six commercially
available techniques in the measurement of microalbuminuria in diabetic
patients. RESEARCH DESIGN AND METHODS--Timed overnight urine samples from
75 patients in our diabetic clinic were tested with 2 qualitative tests
(Micral-Test and Microbumintest) and assayed with 4 quantitative tests,
which used different methodologies: double-antibody RIA, RID, IT, and NEPH.
All tests were commercially available. The double-antibody RIA was taken as
the golden standard. All urine samples were either negative or trace by
Albustix (Ames, Elkhart, IN). Interobserver variation for the 2 qualitative
tests was assessed by asking 12 experienced nurses to read the color
changes on the dipsticks and the tablets simultaneously, using test
solutions with albumin concentrations of 16, 32, and 64 mg/L. RESULTS--The
75 urine samples contained 0-154.2 mg/L of albumin. Correlation
coefficients of RID, NEPH, and IT with RIA were 0.970, 0.975, and 0.974,
respectively. Intra-assay and interassay CVs ranged from 1.36-11.5%.
Microbumintest had a higher sensitivity (100 vs. 75%), but lower
specificity (82.5 vs. 87.3%) than Micral-Test. Considerable interobserver
variation existed in the matching of colors for both tests. Discrepancies
were especially significant with Microbumintest, with 8 of 12 nurses
misreading the 32 mg/L level. CONCLUSIONS--Correlations of the 3
quantitative methods with RIA were all > 0.970. Choice of test may
depend more on considerations such as time, space, number of specimens to
be handled, and availability of instruments. Both qualitative tests showed
a relatively low specificity. Positive tests must be confirmed with
quantitative assays before microalbuminuria is diagnosed. Microbumintest
had a higher sensitivity, but its unacceptably high interobserver variation
and lower specificity were serious drawbacks.