Abstract
Introduction: Clinical laboratories have replaced conventional manual urine microscopy with automated urinalysis; however, concerns persist regarding its validity in detecting specific elements of urinary sediment crucial for evaluating kidney diseases. This study aimed to assess the accuracy of urinary sediment analysis performed by a large hospital laboratory compared to a standardized microscopic review, focusing on patients both with and without kidney disease. Methods: Urine samples were randomly selected from routine laboratory specimens at a university hospital. Laboratory analysis was performed using LabUmat 2 and Urised 3 PRO equipment (Abbott Diagnostics). In the automated analysis for sediment examination, technicians have the option to reclassify urinary sediment elements as necessary and, if warranted, conduct manual microscopic evaluations to validate findings. The laboratory’s analysis was compared with a “reference” analysis, which was double-blinded and conducted by two experienced technicians using bright-field and phase-contrast microscopy. Results: 503 samples were selected, with 52.3% originating from nephrology outpatient clinic patients. Overall agreement between the laboratory results and the reference analysis was 42.1%. The sensitivity (SN) of the laboratory examination for detecting pathological casts, lipiduria, and renal tubular epithelial cells was low (<50%), while specificity (SP) was high (>98%). However, for hyaline casts (SN: 50.4%; SP: 80.9%) and dysmorphic red blood cells (SN: 62.3%; SP: 96.2%), accuracy was intermediate. Performance was better for hematuria (SN: 86.1%; SP: 82.3%; intraclass correlation coefficient [ICC]: 0.703; R: 0.828) and leukocyturia (SN: 84.9%; SP: 95.1%; ICC: 0.807; R: 0.861). In patients with kidney disease (N = 248) and in samples manually reviewed by the laboratory (N = 115), accuracy for each urinary element was comparable to the overall sample findings. However, when assessing the ability to identify elements suggestive of nephropathy, only samples manually reviewed by the laboratory showed statistically similar results to those obtained by the reference analysis (p = 0.503, McNemar’s test). Conclusion: Employing automated urinalysis seems to be accurate for detecting hematuria and leukocyturia, as well as for screening patients without kidney diseases. However, clinical laboratories attending complex patients should employ personalized strategies to help decide when to perform manual review, thus avoiding misleading urinalysis results.
