目的 比较慢性肾脏病流行病学合作研究组（CKD-EPI）公式、美国肾脏病膳食改良试验简化（aMDRD）公式、Cockcroft-Gault（G-C）公式、梅奥诊所二次（MCQ）公式及柏林倡议研究（BIS）公式，在评估危重症患者肾脏功能中的表现并探讨临床应用中最适合的方法。 方法 纳入2020年6月至2022年6月在南京医科大学第一附属医院（江苏省人民医院）老年医学科重症监护病房住院的危重症患者，分别采用CKD-EPI公式、aMDRD公式、C-G公式、MCQ公式及BIS1公式与24 h肌酐清除率（CrCl_24h）进行比较，分析各个公式在入院48 h内评估危重症患者肾功能的精确度和准确度。 结果 共534例患者纳入研究。（1）aMDRD公式的偏倚最小（3.91），5种估算肾小球滤过率（estimated glomerular filtration rate， eGFR）公式的准确度均较低，分别为44.4%、42.9%、63.1%、44.9%和54.9%。（2）CKD-EPI公式、aMDRD公式、C-G公式、MCQ公式、BIS1公式对患者肾功能的分级与CrCl_24h的一致性加权κ值分别为0.464、0.555、0.403、0.405、0.159（P < 0.001）。CKD-EPI公式倾向于高估肾功能中至重度下降患者的分级，低估肾功能正常患者的分级。aMDRD公式、MCQ公式、BIS1公式均倾向于高估肾功能减退患者的分级，低估肾功能正常患者的分级。C-G公式倾向于低估肾功能正常及肾功能轻至中度下降患者的分级。（3）对于肾功能明显下降的患者［CrCl_24h ≤ 60 mL/（min·1.73 m²）］，C-G公式的eGFR值最低，BIS1公式的eGFR值最高。对于肾功能正常至中度下降的患者［60 mL/（min·1.73 m²） < CrCl_24h ≤ 130 mL/（min·1.73 m²）］，C-G公式的eGFR值依然最低，而MCQ公式的eGFR值最高。对于肾功能亢进的患者［CrCl_24h > 130 mL/（min·1.73 m²）］，BIS1公式的eGFR值最低，aMDRD公式的eGFR值最高。（4）患者肾功能明显减退［CrCl_24h ≤ 60 mL/（min·1.73 m²）］或肾功能亢进［CrCl_24h > 130 mL/（min·1.73 m²）］时，CKD-EPI及MCQ公式准确度下降。aMDRD公式的准确度随患者肌酐清除率升高而增加。C-G公式呈现出与aMDRD公式相反的变化趋势。BIS1公式则在所有分组中准确度均较低。（5）5种eGFR公式诊断肾功能亢进（ARC）的最佳临界值［mL/（min·m²）］分别为eGFR_CKD-EPI：91.15，eGFR_aMDRD：99.84，eGFR_C-G：76.27，eGFR_MCQ：100.87，eGFR_BIS1：82.36。 结论 5种eGFR公式的精确度及准确度均较低，或不建议用于危重症患者肾功能的评估，收集24 h尿液计算肌酐清除率仍然是评估危重患者肾功能的必要方法。在入ICU早期无CrCl_24h时，eGFR公式的截断值可用于早期识别ARC，aMDRD公式可用于粗略估算危重症患者的肌酐清除率。

Objective To evaluate and compare the performance of the Chronic Kidney Disease Epidemiology Collaboration （CKD-EPI） equation， the abbreviated Modification of Diet in Renal Disease （aMDRD） equation， the Cockroft-Gault （C-G） formula， the Mayo Clinic Quadratic （MCQ） equation， and the Berlin Initiative Study 1 （BIS1） equation in determining renal function among critically ill patients， and to identify the most appropriate method for clinical application. Methods Critically ill patients admitted to the Intensive Care Units of the Department of Geriatric Medicine at the First Affiliated Hospital of Nanjing Medical University （Jiangsu Province Hospital） between June 2020 and June 2022 were included. Their renal function was assessed within 48 hours of admission using the 24-hour creatinine clearance rate （CrCl_24h） as the reference standard， and compared with the CKD-EPI equation， aMDRD equation， C-G formula， MCQ equation， and BIS1 equation. The precision and accuracy of each equation in evaluating renal function in critically ill patients were analyzed. Results Total of 534 patients were included in the study. （1） The aMDRD equation exhibited the least bias （-3.91）， yet the accuracy of the five estimated glomerular filtration rate （eGFR） equations was relatively low， ranging from 42.9% to 63.1%. （2） For renal function grading， the weighted κ agreement values between the CKD-EPI equation， aMDRD equation， C-G formula， MCQ equation， BIS1 equation， and CrCl_24h were 0.464， 0.555， 0.403， 0.405， and 0.159， respectively （all P < 0.001）. （3） Among patients with severe kidney function decline ［CrCl_24h ≤ 60 mL/（min·1.73 m²）］， the eGFR value derived from the C-G formula was the lowest， while that from the BIS1 equation was the highest. In patients with normal or moderately reduced renal function ［60 mL/（min·1.73 m²）< CrCl_24h ≤ 130 mL/（min·1.73 m²）］， the eGFR value of the C-G formula remained the lowest， whereas the MCQ equation yielded the highest eGFR value. For patients with augmented renal function ［CrCl_24h > 130 mL/（min·1.73 m²）］， the eGFR value of the BIS1 equation was the lowest， and the aMDRD equation produced the highest eGFR value. （4） When renal function was severely decreased or augmented ［CrCl_24h > 130 mL/（min·1.73 m²）］， the accuracy of the CKD-EPI and MCQ equations declined. Conversely， as creatinine clearance increased， the accuracy of the aMDRD equation improved gradually. The C-G formula demonstrated an opposite trend compared to the aMDRD equation， and the BIS1 equation exhibited low accuracy across all groups. （5） The optimal critical values ［mL/（min·m²）］ for diagnosing augmented renal clearance （ARC） using the five eGFR equations were as follows： eGFR_CKD-EPI： 91.1， eGFR_aMDRD： 99.84， eGFR_C-G： 76.27， eGFR_MCQ： 100.87， eGFR_BIS1： 82.36. Conclusions The precision and accuracy of the five eGFR equations are relatively low， making them unsuitable for assessing renal function in critically ill patients. Collecting 24-hour urine to calculate creatinine clearance remains an essential method for evaluating renal function in this population. In the absence of early CrCl_24h data upon ICU admission， the cut-off values of the eGFR equations may serve as a tool for the early identification of ARC. Additionally， the aMDRD equation can provide a rough estimate of creatinine clearance in critically ill patients.