Background: The urine protein-creatinine ratio (UPCR) in a spot first-morning urine sample is used to estimate 24-h urine proteinuria (24hUP) in patients who underwent urine protein testing. UPCR cannot be directly compared with 24-h proteinuria. Thus, an equation to estimate 24-h total protein excretion rate, using age, gender, and the UPCR may improve its bias and accuracy in patients who underwent urine protein testing.
Methods: We simultaneously measured 24-h urine protein and the same day's first-morning spot urine from patients with kidney disease. Generalized linear and no-linear models, using age, gender, and UPCR, were constructed to estimate for 24-h urine protein and the best model (NJ equation) was selected to estimated 24 hUP (e24hUP).
Results: A total of 5435 paired samples (including a training cohort of 3803 patients and a validation cohort of 1632 patients) were simultaneously measured for UPCR and 24-h urine protein. In the training cohort, the unadjusted UPCR obviously underestimated 24-h urine protein when UPCR ≤1.2 g/g (median bias - 0.17 g/24 h) and overestimated 24-h urine protein when UPCR > 1.2 g/g (median bias 0.53 g/24 h). In the validation cohort, the NJ equation performed better than the unadjusted UPCR, with lower root mean square error (0.81 vs. 1.02, P < 0.001), less bias (median difference between measured and estimated urine protein, - 0.008 vs. 0.12), improved precision (interquartile range of the differences, 0.34 vs. 0.50), and greater accuracy (percentage of estimated urine protein within 30% of measured urine protein, 53.4% vs. 32.2%). Bland-Altman plot indicated that the agreement of spot and daily estimates was less pronounced with 24 hUP > 2 g than lower values.
Conclusions: The NJ e24hUP equation is more accurate than unadjusted UPCR to estimate 24 hUP in patients with kidney disease and could be used for laboratory application.
Keywords: Prediction model; Proteinuria; Urine protein-creatinine ratio.
© 2022. The Author(s).