目的 比较8种肾小球滤过率（GFR）估算方程在评估15~18岁慢性肾脏病（CKD）儿童肾小球滤过率中的准确性及适用性。方法 收集2015年1月至2021年3月住院的CKD患儿为研究对象。应用8种肾小球滤过估算（eGFR）公式（改良Schwartz公式、CAPA公式、Counahan-Barratt公式、Filler公式、CKD-EPI-Scr2009、CKD-EPI-CysC2012、LMR18、FAS公式）估算GFR，采用^99mTc-DTPA肾动态显像Gates法为金标准GFR（sGFR）的测定方法。比较各公式的偏倚、精确性、准确性，分析各公式对于肾功能不全的诊断效能。结果 纳入88例患儿，男56例、女32例，中位年龄17.0（16.0~18.0）岁；CKD 1期56例，CKD 2期18例，CKD 3期11例，CKD 4期2例，CKD 5期1例。各公式eGFR与sGFR均呈正相关（P均<0.001）。其中CKD-EPI-Scr2009公式与sGFR相关性最好（r=0.73），Filler公式相关性最差（r=0.39）。CAPA公式、CKD-EPI-Scr2009、FAS公式在整体水平上高估了患儿的GFR，其余公式均存在GFR低估现象。以sGFR为金标准，FAS公式偏倚最小，CAPA公式偏倚最大。在精确性方面，LMR18公式精确性最好，改良Schwartz公式次之，CAPA公式精确性最差。在准确性方面，LMR18公式的准确性最高（P₃₀=73.86%）。ROC曲线分析显示，CKD-EPI-Scr2009公式曲线下面积（AUC）最大（0.91），CAPA、Filler、CKD-EPI-CysC2012公式灵敏度最高（均为95.90%），LMR18、FAS公式的特异度最高（均为85.71%）。结论 8种eGFR公式中，基于CysC公式（CAPA、Filler、CKD-EPI-CysC2012公式）的准确性等方面不及基于Scr的公式（改良Schwartz、Counahan-Barratt、CKD-EPI-Scr2009、LMR18、FAS公式），其中LMR18公式精确性、准确性均最高。

Objective To compare the accuracy and applicability of eight estimation formulas for glomerular filtration rate (GFR) in children aged 15-18 years with chronic kidney disease (CKD). Methods Children with CKD hospitalized from January 2015 to March 2021 wereenrolled. Eight eGFR formulas (update Schwartz formula, CAPA formula, Counahan-Barratt formula, Filler formula, CKD-EPI-Scr2009, CKD-EPI-CysC2012, LMR18, FAS formula) were applied to estimate GFR. The Gates method of 99mTc-DTPA renal dynamic imaging was used as the gold standard GFR (sGFR) determination. The bias, precision and accuracy of each formula were compared, and the diagnostic efficacy of each formula for renal insufficiency was analyzed. Results A total of 88 children were enrolled, including 56 boys and 32 girls, and the median age was 17.0 (16.0-18.0) years. There were 56 cases of CKD stage 1, 18 cases of CKD stage 2, 11 cases of CKD stage 3, 2 cases of CKD stage 4, and 1 case of CKD stage 5. The eGFR in all formulas was positively correlated with sGFR (all P<0.001). The CKD-EPI-Scr2009 formula had the best correlation with sGFR (r=0.73), and the Filler formula had the worst correlation with sGFR (r=0.39). The CAPA, CKD-EPI-Scr2009 and FAS formulas overestimated the GFR on the overall level, while the other formulas underestimated GFR. When sGFR was used as the gold standard, the FAS formula had the smallest bias and the CAPA formula had the largest bias. In terms of precision, the LMR18 formula showed the best precision, followed by the update Schwartz formula, and the CAPA formula showed the worst precision. In terms of accuracy, the LMR18 formula showed the highest accuracy (P₃₀=73.86%). ROC curve analysis showed that the area under the curve (AUC) of the CKD-EPI-Scr2009 formula was the largest (0.907), the sensitivity of CAPA, Filler, and CKD-EPI-CysC2012 formulas were the highest (95.90%), and the specificity of LMR18 and FAS formulas were the highest (85.71%). Conclusions Among the eight eGFR formulas, the accuracy of CysC-based formulas (CAPA, Filler and CKD-EPI-CysC2012 formula) is not as good as Scr-based formulas (update Schwartz, Counahan-Barratt, CKD-EPI-Scr2009, LMR18 and FAS formula), and LMR18 formula has the highest precision and accuracy.