外科手术部位感染病例的早期预测研究

目的：建立数学模型，对每个手术病例进行预测，确定出外科手术部位感染（SSI）的高危病例，及早采取针对性的预防措施，减少和预防SSI的发生。方法：采用尤度法，所用初选自变量为14个，最后入选自变量为7个：X4、X5、X7、X8、X9、X12、X14，计算出SSI非高危病例（Y1）和SSI高危病 列（Y2）的条件概率。结果：符合率67．95％，灵敏度78．67％，特异度67．27％，假阳性率32．73％，假阴性率为21．33％；应用于1999年的资料进行回代性判别，总符合率68．95％，灵敏度80．37％，特异度68．27％，假阳性率31．73％，假阴性率19．63％。结论：应用判别分析方法预测出SSI的高危病例，在术前的手术方案制订以后即可进行；预测方法比较简单，可在各类医院推广应用。

Early Discrimination Study of High Risk Surgical Site Infection

OBJECTIVE To design a mathematic model and to calculate and forecast highly risk surgical site infection(SSI) cases in order to prevent the surgical site infection and reduce the rate of SSI. METHODS By applying maximum likelihood discriminating analysis method, 7 independent variables were selected to creat the discriminant model: X 4(age), X 5(nutritive condition), X 7(skin preparation), X 8(type of surgical site), X 9(emergency case), X 12 (the length of operating time), X 14 (drainage), to separately calculate and compare the condition probabilities of SSI and non SSI. RESULTS The results of return discrimination from this model indicated that the fitness was good: crude agreement was 67.95%, the sensitivity was 78.67%, the specificity was 67.27%. The result of retrospective discrimination showed that crude agreement was 68.95%, the sensitivity was 80.37% and the specificity was 68.27%.CONCLUSIONS This discriminant model was very simple for calculation and discrimination, it was an useful method for primary screening the highly risk surgical site infection cases at the different level hospitals.