LAUSR

Background Dysmorphic red blood cells (dRBCs) on urine microscopy have been associated with glomerulonephritis (GN). We assessed the prevalence and ability of dRBCs to differentiate GN from other kidney diseases. Methods Adult patients with kidney biopsy performed between 2012 and 2015 at a single center who had a concurrent urinalysis were retrospectively studied. The association of ≥25% dRBCs with the presence of glomerular pathology was assessed. Univariate and multivariate logistic regression were performed on significantly associated variables. Results The mean age of the 482 eligible subjects was 55 years and 47.7% were female. Overall, 173 (35.9%) had <25% and 76 (15.8%) had ≥25% urine dRBCs. Kidney biopsies revealed glomerular disease in 372 (77.2%) (GN 46% and non-GN 54%). At the dRBC threshold of ≥25% used at our center, a sensitivity of 20.4%, specificity of 96.3% and positive predictive value of 94.6% for glomerular disease were observed. In a logistic regression model, urine RBCs [>10 versus ≤10 (P < 0.001)] but not dRBCs ≥25% (P = 0.3) independently predicted the presence of GN. A scoring system (0-3) based on hematuria and proteinuria levels revealed the risk for biopsy-proven GN was 15% when the score was 0 compared with 83% when it was 3. Conclusions The presence of ≥25% urine dRBCs is specific but not sensitive for GN. In this cohort, the combined hematuria (>10 RBCs/high-power field) and proteinuria performed just as well as dRBCs plus proteinuria to predict underlying GN. A model based on the degree of hematuria and proteinuria found on urinalysis was able to predict the presence of GN.