4项评分系统对重症急性胰腺炎预后评估的受试者工作特征曲线分析

目的：评价入院时Ranson、急性生理学和长期健康评价Ⅱ（acute physiology and chronic health evaluationⅡ，APACHEⅡ）、Bahhazar CT以及胰腺炎结局预测（pancreatitis outcome prediction，POP）评分系统对重症急性胰腺炎（severe acute pancreatitis，SAP）患者的预后评估价值。方法：回顾性研究了2004年1月～2007年8月在四川大学华西医院住院的253例SAP患者，计算患者入院时Ranson、APACHEⅡ、Balthazar CT和POP的评分分值。观察死亡、胰腺局部并发症、中转手术和多器官功能障碍综合征（multiple organ dysfunction syndrome，MODS）发生等4项结局指标。将所有患者的4项评分绘制受试者工作特征（receiver operating characteristic，ROC）曲线，通过计算曲线下面积（areaundercurve，AUC）来比较各评分系统在判断SAP病情严重度和预后上的敏感性、特异性、Youden指数和阳性似然比（likelihood ratio of positivetest，LRPT）。结果：ROC曲线分析显示Ranson评分可用于预测死亡、中转手术及MODS发生（P=0．00），其诊断阈值均为5。APACHEⅡ评分可以预测胰腺局部并发症以外其他结局指标的发生（P=0．00），APACHEⅡ评分对死亡、中转手术、MODS发生的诊断阈值依次为14、10、12。BalthazarCT评分可预测各项结局指标的发生，诊断阈值均为6。POP评分可预测胰腺局部并发症以外的各项结局指标，其对死亡的预测价值尤为突出，当POP评分≥14时，其预测死亡的敏感性为0．90，特异性为0．92。POP评分对死亡、中转手术、MODS发生的诊断阂值依次为14、8、10。结论：4项评分系统预测能力各有侧重。Ranson、APACHE1I和POP评分可以用于疾病的严重程度及全身并发症的预测，对局部并发症的预测效果不佳。POP评分对病死率的预测效果较好；BalthazarCT评分能预测胰腺局部并发症和中转手术的发生及其他结局指标的发生，但敏感性和特异性不高。

Value of four scoring systems for predicting prognosis of severe acute pancreatitis

Objective： To evaluate the value of Ranson, acute physiology and chronic health evaluation Ⅱ （APACHEⅡ）, Balthazar CT and pancreatitis outcome prediction （POP） scoring systems in predicting the prognosis of severe acute pancreatitis （SAP）.
Metheds： A total of 253 consecutive patients with SAP, admitted into West China Hospital of Sichuan University from January 2004 to August 2007 were studied retrospectively. Data associated to Ranson, APACHEⅡ, Balthazar CT and POP scoring systems after admissions were recorded. Four outcome signals including death, local complications, operation transformation, and development of multiple organ dysfunction syndrome （MODS） were analyzed. The receiver operating characteristic （ROC） curves were also drawn and the areas under the ROC curve （AUC） were obtained to compare the sensitivity, specificity, Youden index, and likelihood ratio positive test （LRPT） of the clinical scoring systems in assessing the prognosis of SAP. Results： Through comparison of AUC, Ranson score could be used to predict the outcomes except for local complications （P =0.00）, and all its diagnosis threshold was 5. APACHE Ⅱ score showed that it could assess the outcomes of SAP except for local complications （ P =0.00）. Its thresholds in mortality, operation transformation, and incidence of MODS were 14, 10, and 12 respectively. Balthazar CT could predict all the outcomes, and the threshold was 6. POP score could predict all the outcomes except for local complications, and its predictive value in mortality was particularly prominent. The sensitivity and specificity of POP in predicting death were 0.90 and 0. 92 as the cutoff was no less than 14：. The thresholds of POP in mortality, operation transformation, and incidence of MODS were 14, 8, and 10 respectively. Conclusion： The prognostic ability of the four scoring systems is not the same. Ranson, APACHEⅡ and POP scoring systems can predict the severity of SAP and entire complication, but they can＇t predict the local complications. POP score may predict the mortality well, and Balthazar CT scanning can predict all the outcomes, but the sensitivity and specificity are low.