Objective: To assess the reliability of estimated urine protein to predict 24 h urine protein excretion in children with glomerular diseases. Methods: Four hundred and forty-three children with glomerular diseases, who were admitted to pediatric department of Peking University First Hospital from January 2001 to December 2021, were enrolled in the cross-sectional study. The 24 h estimated urine creatinine which calculated by 6 previously described equations, 24 h measured urine creatinine, measured urine protein-to-creatinine ratio(UPCR), 24 h urine protein (24 hUP) and urinary sediment analysis with microscopy were collected, estimated urine protein was computed as the product of measured UPCR and estimated or measured 24 h urine creatinine. Spearman correlation analysis, Bland-Altman analysis and linear regression analysis were used to compare the correlation, agreement and accuracy between estimated urine protein and 24 hUP, and the effect of urinary protein level and erythrocyte numbers on their relationship was analyzed. Results: Of 443 children with glomerular diseases (aged (11±4) years, 221 male, 222 female), there were 216 participants with nephrotic syndrome, 78 participants with IgA nephropathy, 47 participants with Alport syndrome, 42 participants with lupus nephritis, 58 participants with purpura nephropathy, and 2 participants with isolated proteinuria. Spearman correlation analysis showed a strong correlation between estimated urine protein and 24 hUP (r=0.90, P<0.05), and the correlation improved after multiplying the measured UPCR by 24 h measured urine creatinine (r=0.94, P<0.05). Improved correlation was also observed using the estimated urine creatinine which calculated by Hellerstein formula, Ghazali-Barratt formula, Ellam formula, Walser formula, Cockcroft-Gault formula, Ix formula (r=0.93, 0.94, 0.90, 0.90, 0.94, 0.93, all P<0.05).Bland-altman analysis showed that the difference between measured UPCR and 24 hUP was (-0.30±2.22) g, consistency limit was -4.65-4.04, and the consistency improved after 24 h measured urine creatinine correction (difference was (0.27±1.31) g, consistency limit -2.30-2.84). The consistency of estimated urine protein was further improved after correction by different formulas, and the Cockcroft-Gault formula showed the best consistency between estimated urine protein and 24 hUP (difference was (0.11±1.18)g, consistency limit was -2.20-2.42). Linear regression analysis showed that measured UPCR had poor accuracy in predicting 24 hUP (R2=0.55, α=0.48, β=0.60, P<0.05), and the accuracy improved after 24 h measured urine creatinine correction, the accuracy of estimated urine protein for predicting 24 hUP was further improved by using different formulas, and Cockcroft-Gault formula was the best (R2=0.81, α=0.18, β=0.96, P<0.05). With the increase of urinary protein level and the decrease of urinary erythrocyte numbers, the correlation, agreement and accuracy between estimated urine protein and measured UPCR and 24 hUP were improved(all P<0.05). Except Ellam and Ix formulas, estimated urine protein using the rest four formulas outperformed measured UPCR(all P<0.05). Conclusion: The 24 h urine creatinine excretion rate (obtained by the Cockcroft-Gault equation)-weighted urine protein-to-creatinine ratio more reliably predicts 24 hUP than measured UPCR alone in children with glomerular diseases.
目的： 评价不同公式估算尿蛋白反映儿童肾小球疾病24 h尿蛋白定量的可靠性。 方法： 横断面研究，收集2001年1月至2021年12月就诊于北京大学第一医院儿科的443例肾小球疾病患儿实测24 h尿肌酐和通过6个相关公式计算得出的估算24 h尿肌酐、实测24 h尿蛋白定量、实测尿蛋白肌酐比（UPCR）、尿沉渣镜检结果，用实测24 h尿肌酐或估算24 h尿肌酐乘实测UPCR计算得到估算尿蛋白。采用Spearman相关分析、Bland-Altman分析法及线性回归分析比较估算尿蛋白和实测UPCR与24 h尿蛋白定量间的相关性、一致性和准确性，并分析尿蛋白水平和尿红细胞数目对于二者关系的影响。 结果： 443例患儿中男221例、女222例，年龄（11±4）岁，其中肾病综合征216例、IgA肾病78例、Alport综合征47例、系统性红斑狼疮42例、紫癜性肾炎58例以及孤立性蛋白尿2例。Spearman相关分析提示实测UPCR与24 h尿蛋白定量具有相关性（r=0.90，P<0.05），应用实测24 h 尿肌酐得到的估算尿蛋白和24 h尿蛋白定量相关性得到改善（r=0.94，P<0.05），应用Hellerstein、Ghazali-Barratt、Ellam、Walser、Cockcroft-Gault、Ix公式估算的尿蛋白与24 h尿蛋白定量的相关性均加强（r值分别为0.93、0.94、0.90、0.90、0.94、0.93，均P<0.05）；Bland-Altman分析法示实测UPCR与24 h尿蛋白定量的差值为-0.30±2.22，一致性界限为-4.65~4.04，应用实测24 h尿肌酐校正后一致性好转［差值为（0.27±1.31）g，一致性界限为-2.30~2.84 g］，不同公式估算尿蛋白一致性进一步好转，其中Cockcroft-Gault公式估算尿蛋白与24 h尿蛋白定量一致性最好［差值为（0.11±1.18）g，一致性界限为-2.20~2.42 g］。线性回归分析示实测UPCR预测24 h尿蛋白定量准确性欠佳（R2=0.55、α=0.48、β=0.60，P<0.05），应用实测24 h尿肌酐校正后准确性好转，应用不同公式估算尿蛋白预测24 h尿蛋白定量准确性进一步加强，其中Cockcroft-Gault公式准确性最佳（R2=0.81、α=0.18、β=0.96，P<0.05）。随尿蛋白定量增加和尿红细胞数目减少，估算尿蛋白和实测UPCR与24 h尿蛋白定量间相关性、一致性和准确性均增加（均P<0.05）；除Ellam和Ix公式外，其余4个公式计算的及实测24 h尿肌酐得到的估算尿蛋白与24 h尿蛋白定量间相关性、一致性和准确性均高于实测UPCR（均P<0.05）。 结论： 基于Cockcroft-Gault公式计算的24 h尿肌酐得到的估算尿蛋白能更准确地评估儿童肾小球疾病24 h尿蛋白水平。.