ARF专用与ICU通用病情评分法对29例重症急性肾衰竭患者院内死亡判别力的比较

Hospital mortality prediction of five severity scoring models on 29 critical acute renal failure patients

目的比较急性肾功能衰竭(ARF)专用及ICU通用病情评分法对重症ARF患者死亡的判别能力。方法29例ARF患者分为存活组(n=20)和死亡组(n=9),用两种ARF专用病情评分法,即急性肾小管坏死个体病情严重性指数(ATN ISI)、Stuiven berg医院急性肾衰竭评分法(SHARF),以及ICU通用病情评分法,即急性生理学及慢性健康状况评分Ⅱ(APACHEⅡ)、急性生理学及慢性健康状况评分Ⅲ(APACHEⅢ)和死亡概率预测模型Ⅱ(MPMⅡ),对所选病例进行回顾性病情评估,并做受试者死亡判别的工作特征(ROC)曲线。结果存活组各项积分均低于死亡组,其中ATN ISI、APACHEⅡ、MPMⅡ24h、MPMⅡ72h等死亡组与存活组积分差异有显著性(P<0.05)。5种病情评分法ROC曲线下面积由大到小依次为:APACHEⅡ(0.806)、MPMⅡ72h(0.762)、MPMⅡ24h(0.754)、MPMⅡ48h(0.738)、ATN ISI(0.736)、MPMⅡ0h(0.713)、APACHEⅢ(0.706)、SHARF48h(0.660)、SHARF0h(0.586)。结论对ARF患者ICU通用评分法较ARF专用评分法院内死亡判别能力更高,APACHEⅡ能力较优。

Objective To compare the applicability between ARF special severity scoring systems and intensive care unit (ICU) general severity scoring systems on predicting the hospital mortality of critical acute renal failure (ARF) patients. Methods The data of 29 patients with critical ARF were retrospectively analyzed. Two ARF special severity scoring models, acute tubular necrosis individual severity index (ATN-ISI) and Stuivenberg hospital acute renal failure scores (SHARF), and 3 ICU general severity scoring models, Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ), APACHE Ⅲ, Mortality Prediction Model Ⅱ (MPM Ⅱ), were used in the analysis. The receiver operating characteristic (ROC) curves were drawn up to assess the ability of these models in hospital mortality discrimination. Results All severity scores of the patients in survivor group (n=9) were lower than those of the patients in non-survivor group (n=20). There were statistical differentiations in the scores of ATN-ISI, APACHE Ⅱ, MPM Ⅱ-24h, MPM Ⅱ-72h between the survivor group and the non-survivor group (P<0.05). Areas under the ROC curves for these 5 models were (in decreasing order): APACHE Ⅱ(0.806), MPM Ⅱ-72h(0.762), MPM Ⅱ-24h(0.754), MPM Ⅱ-48h(0.738), ATN-ISI(0.736), MPM Ⅱ-0h(0.713), APACHE Ⅲ(0.706), SHARF-48h(0.660), SHARF-0h(0.586). Conclusion ICU general severity scoring models demonstrated better ability to predict hospital mortality on critical ARF patients than ARF special models.