The diagnostic accuracy of a urine albumin-creatinine ratio point-of-care test for detection of albuminuria in primary care

Background: Albuminuria is an important sign of chronic kidney disease and is detected routinely by measurement of urinary albumin-creatinine ratio (ACR). A Siemens CLINITEK test designed for use at the point of care is available that can semiquantitatively measure ACR.

Study design: Diagnostic accuracy study evaluating a urinary ACR point-of-care test.

Setting & participants: The semiquantitative ACR test was evaluated at the point of care in a representative primary care population (those with or at increased risk of chronic kidney disease) of 642 patients under standard operational conditions and compared with the reference standard of ACR measurement in the clinical laboratory.

Index test: The point-of-care CLINITEK semiquantitative ACR test. This test uses dye-binding and catalytic assays for albumin and creatinine, respectively, on a Microalbumin 9 strip, which is read by the CLINITEK Status Analyzer, and ACR is calculated automatically.

Reference test: Laboratory measurement of albumin and creatinine on an Abbott Architect analyzer by immunoturbidimetric and enzymatic assays, respectively, and calculation of ACR.

Results: The prevalence of albuminuria (laboratory ACR≥30 mg/g) in the study population was 20.2%. Sensitivity and specificity of the point-of-care test for detecting albuminuria were 83.2% and 80.0%, respectively. Positive and negative predictive values were 51.2% and 95.0%, respectively; positive and negative likelihood ratios were 4.16 and 0.21, respectively. Twenty-three (3.6%) samples measured at the point of care were not analyzed in the central laboratory for a variety of reasons, including laboratory reception data entry errors.

Limitations: Our sensitivity calculation is accurate to an approximately 8% CI.

Conclusions: The instrument-read reagent strip test was a poor rule-in test for albuminuria at the point of care, as evidenced by the low positive predictive value, but was a reasonable rule-out test. Observed sensitivity was lower than reported in earlier laboratory-based studies. This decreased diagnostic accuracy needs to be balanced against the potential advantages of a point-of-care testing approach.