How effective are screening tests for microalbuminuria in random urine specimens?

The effectiveness of four urine screening tests-microalbumin (MAlb), total protein (TProt), total protein/creatinine ratio (TProt/Cr R), and dipstick (DPalb) test for albumin-were evaluated for the detection of MAlb in random urine specimens. The following criteria were used to assess the effectiveness of each urine screening test: 100% specificity (no false positive results); cost effectiveness; rapidity and ease of performing the screening test; and increased laboratory efficiency. A "gold standard" for presence of MAlb in random urine samples was defined as a microalbumin/creatinine ratio (MAlb/Cr R) of > or = 30 mg/g. The least costly urine screening test was the DPalb, which, if assigned a value of 1.0, allowed a cost ranking order for the screening tests-DPalb (1.0) < urine TProt (1.03) < urine TProt/Cr R (2.1) < urine MAlb (7.0). Two hundred urine samples from diabetic inpatients and outpatients were tested. Only two screening tests--MAlb and DPalb--achieved 100% specificity without increasing laboratory costs (small net savings), whereas the other two screening tests--TProt and TProt/Cr R-only achieved 100% specificity with increased laboratory costs. Theoretical prevalence rate analysis showed that urine MAlb screening would be effective at all prevalence rates for overt nephropathy. TProt and DPalb urine screening testing would be most effective in populations with prevalence rates of > or = 15% for overt nephropathy. The TProt/Cr R ratio would only be effective in populations with prevalence rates of > or = 30%. Of the four urine screening tests, only DPalb would significantly streamline the process of measuring urine MAlb. The dipstick test is inexpensive, easy and rapid to perform, does not delay measuring the ratio, since there is no wait for the screening test result, and can be used by referring laboratories to screen urine specimens before they are submitted to a central laboratory, thereby reducing laboratory workload.