Study design: A retrospective analysis.
Objectives: The objective of this study is to determine whether dipstick protein analysis (DSP) or random urine protein:creatinine ratios (UPC) are accurate in predicting clinical proteinuria in the chronic spinal cord injury (SCI) population.
Methods: A retrospective analysis was performed in 219 veterans with SCI, comparing DSP and 24-h urine protein excretion. Sensitivity, specificity, predictive values (PV) and receiver-operator characteristic (ROC) curves of DSP in predicting clinical proteinuria were calculated with and without correction for specific gravity (SG). A prospective study was also performed in 62 SCI patients, comparing the UPC and 24-h urines. Sensitivity, specificity, PV and ROC curves of UPC in predicting clinical proteinuria were calculated.
Results: Any level of positive DSP had high specificity, but low sensitivity, for detecting the presence of clinical proteinuria. ROC curves of DSP for identifying clinical proteinuria yielded area under the curve of 0.749 (95% confidence interval 0.699-0.794), and adjustment for SG did not significantly improve accuracy. A UPC of <0.3 was sensitive with a high negative PV for ruling out clinical proteinuria, whereas a ratio >0.8 was specific with a high positive PV. A UPC between 0.3-0.8 had an intermediate sensitivity and specificity.
Conclusion: Urine collections of 24-h are still needed in the chronic SCI population for accurate detection of clinically significant proteinuria. DSP may not reliably detect low-grade clinical proteinuria, whereas a UPC below 0.3 may be used to rule out clinical range proteinuria.