BISAP评分系统对急性胰腺炎严重程度评估的荟萃分析

Meta analysis of BISAP score system in predicting the severity of acute pancreatitis

目的评估BISAP评分系统在预测急性胰腺炎（AP）严重程度的临床应用价值。方法计算机检索Medline、EMBASE、ScienceDirect、Springerlink、CBM、中国知网、万方以及维普数据库2000年1月至2013年3月的文献，按照严格的纳入标准收集BISAP评分系统预测AP严重程度的文献，采用QUADAS量表进行文献质量评价，利用Meta—Disc1．4统计软件进行异质性分析和定量合成，计算汇总的敏感度、特异度、阳性似然比、阴性似然比和受试者特征性工作（ROC）曲线下面积（AUC），结果均采用95％可信区间（95％CI）表示。结果共纳入文献11篇，包括7篇中文论著和4篇英文论著。按QUADAS量表进行分级，其中A级4篇，B级5篇，C级2篇。6篇文献以BISAP2分为cutoff值、9篇文献以BISAP3分为cutoff值（4篇文献采用两个cutoff值）预测SAP。前者汇总的诊断比值比为8．03（95％C15．66～11．38），后者为7．49（95％C15．35～10．49），两组文献均存在中等程度的异质性（I^2=63．3％，P=0．018；I^2=56．1％，P=0．019）。以BISAP2分为cutoff值预测AP严重程度的汇总的敏感度、特异度、阳性似然比、阴性似然比和AUC分别为59％（95％CI56％-63％）、82％（95％CI80％-83％）、3．50（95％CI 2．96～4．14）、0．45（95％CI 0．36～0．56）和0．82；以BISAP3分为cutoff值时分别为44％（95％CI41％～47％）、90％（95％CI89％-91％）、4．59（95％CI3．31-6．37）、0．64（95％C10．61-0．68）和0．64。前者有较高的敏感度，较低的特异度，AUC较大；后者敏感度低，特异度高，AUC较小。结论BISAP预测SAP的最佳的cutoff值为2分。其漏诊率较低，且误诊率在可接受范围内，适合在临床应用及推广。

Objective To investigate the value of BISAP scoring system in predicting the severity of acute pancreatitis. Methods Medline, EMBASE, Science Direct, Springer link, CBM, Cnki, Wan fang and VIP database were retrieved by computer between January 2000 and March 2013, and articles of BISAP score system in predicting the acute panereatitis were collected according to strict inclusion criteria. Quality assessment was made by QUADAS scale. Meta-Disel. 4 software was used to analyze the heterogeneity of included articles and perform quantitative synthesis, as well as calculate the pooled sensitivity and specificity, positive likelihood ratio, negative likelihood ratio, and draw the ROC curve, and the results were presented with 95% CI. Results A total of 11 articles were included, and there were 7 Chinese articles and 4 English articles. According to QUADAS scale, there were 4 articles of A grade, 5 articles of B grade, 2 articles of C grade. There were 6 studies using 2 as BISAP cut-off value, and 9 articles using 3 as cut-off value, while 4 studies using 2 cut-off values. Pooled analysis showed diagnostic odds ratio of 8.03 （95% CI 5.66- 11.38）when cut-off value was 2, and diagnostic odds ratio of 7.49 （95% CI 5.35-10.49 ） when cut-off value was 3. There were moderate heterogeneity in both groups （ I^2 = 63.3 % , P = 0.018 ;I^2 = 56.1% , P = 0. 019 ）. When BISAP cut-off value was 2, the pooled analysis of sensitivity, specificity, positive likelihood ratio, negative likelihood ratio and area under curve （AUC） were 59% （95% C156% 453% ）, 82% （95% CI 80% -83% ）, 3.50 （ 95 % CI 2.96-4.14 ） , 0.45 （ 95 % CI 0. 36-0.56 ） and 0.82 ; and when BISAP cut-off value was 3, the corresponding values were 44% （95% CI 41%-47% ） , 90% （95% CI 89%-91% ） , 4.59 （95% CI 3.31- 6.37 ）, 0.64 （ 95% C10. 61-0.68 ） and 0.64. The former had a high sensitivity, low specificity and large AUC, while the latter had a low sensitivity, high specificity and small AUC. Conclusions The best cut-off value of BISAP in predicting SAP is 2. When cut-off value is 2, the misdiagnosis rate is low, and the false positive rate is in the acceptable range, which is suitable for clinical application.