|国家科技期刊平台
首页|期刊导航|解放军医学杂志|念珠菌菌血症病原菌特点及短期预后风险预测评分模型的构建与验证

念珠菌菌血症病原菌特点及短期预后风险预测评分模型的构建与验证OA北大核心CSTPCD

Construction and validation of a scoring model for pathogen characteristics and short-term prognosis risk prediction of candidemia

中文摘要英文摘要

目的 分析念珠菌菌血症的病原菌特征,构建并验证其短期死亡风险预测评分模型.方法 回顾性分析2011年1月1日-2020年12月31日联勤保障部队第909医院收治的念珠菌菌血症患者的临床资料,分析病原菌构成、药敏试验结果和住院患者发生率.选取念珠菌菌血症324例随机纳入建模组(190例)、验证组(134例),采用二元logistic回归筛选危险因素,依据比值比(OR)赋分,构建30 d死亡风险预测评分模型.运用预测评分模型对建模组、验证组进行评分,验证预测模型的预测效能.结果 336例念珠菌菌血症患者共检出念珠菌356株,其中白色念珠菌126株(35.39%),热带念珠菌79株(22.19%),近平滑念珠菌74株(20.79%),光滑念珠菌48株(13.48%),季也蒙念珠菌14株(3.93%),克柔念珠菌8株(2.25%),其他念珠菌7株(1.97%);住院患者念珠菌菌血症发生率从2011年的0.20‰上升到2020年的0.48‰.念珠菌对两性霉素B耐药率明显低于氟康唑、伏立康唑、伊曲康唑(P<0.05).纳入模型分析的324例中,确诊后30 d死亡95例,病死率29.32%;建模组男性、发热、肠外营养占比明显高于验证组(P<0.05),慢性肺病、1个月内手术史占比均低于验证组(P<0.05).Logistic回归分析结果显示,慢性肾衰竭、机械通气、重度中性粒细胞降低、72 h内未接受抗真菌治疗、急性生理和慢性健康状况Ⅱ评分(APACHE Ⅱ)≥20分为念珠菌菌血症短期死亡的危险因素(P<0.05),OR值分别为3.179、1.970、2.979、2.080、2.399,风险赋分分别为6、4、6、4、5.风险评分模型对建模组ROC曲线下面积(AUC)为0.792(95%CI 0.721~0.862),Hosmer-Lemeshow(H-L)检验结果为P=0.305;验证组AUC为 0.796(95%CI 0.735~0.898),H-L检验结果为P=0.329.风险评分≤8分为短期死亡低风险组,评分9~15分为中风险组,评分≥16分为高风险组.结论 住院患者念珠菌菌血症发生率呈上升趋势,病死率较高;构建的念珠菌菌血症短期死亡风险预测评分模型预测效能较好.

Objective To analyze the pathogenic characteristics and drug sensitivity of candidaemia,and construct a short-term mortality risk prediction scoring model.Methods The clinical data of patients with candidaemia admitted to the 909 Hospital of Joint Logistics Support Force from January 2011 to December 2020 were retrospectively analyzed,and the composition of pathogen composition,drug sensitivity test results and incidence of hospitalized patients were analyzed.324 cases of candidaemia were randomly divided into modeling group(190 cases)and validation group(134 cases),and the risk factors were screened by binary logistic regression.According to the odds ratio(OR)score,the 30 day mortality risk prediction scoring model was constructed,and the predictive performance of the model was verified both in modeling and validation groups.Results 356 strains of Candida including 126 strains of C.albicans(35.39%),79 strains of C.tropicalis(22.19%),74 strains of C.parapsilosis(20.79%),48 strains of C.glabrata(13.48%),14 strains of C.guilliermondii(3.93%),8 strains of C.krusei(2.25%),and 7 strains of other Candida(1.97%)were detected in 336 patients with candidemia.The incidence of candidaemia among hospitalized patients increased from 0.20 ‰ in 2011 to 0.48 ‰ in 2020.The resistance rate of candida to amphotericin B was significantly lower than that of fluconazole,voriconazole and itraconazole(P<0.05).Among the 324 cases included in the model,95 patients died in 30 days after diagnosis,and the mortality rate was 29.32%.The proportion of males,fever,and parenteral nutrition in modeling group was significantly higher than that in validation group(P<0.05),while the proportion of chronic lung disease and surgical history within one month were lower than those in validation group(P<0.05).Logistic regression analysis showed that chronic renal failure,mechanical ventilation,severe neutropenia,failure to receive anti-fungal treatment within 72 hours,and APACHE Ⅱ≥20 were risk factors for short-term death of candidaemia,the OR values were 3.179,1.970,2.979,2.080,and 2.399,and the risk scores were 6,4,6,4,and 5,respectively.The area under the curve(AUC)of the risk scoring model for modeling group was 0.792(95%CI 0.721-0.862),and the result of Hosmer-Lemeshow(H-L)test was P=0.305;The AUC of validation group was 0.796(95%CI 0.735-0.898),and the H-L test result was P=0.329.A risk score≤8 indicated a low risk group for short-term mortality,a score of 9-15 indicated a medium risk group,and a score≥16 indicated a high risk group.Conclusions The incidence of candidemia in hospitalized patients is increasing and the mortality is high.The risk prediction score model can effectively predict the short-term prognosis and facilitate the early identification of the prognosis.

马建新;林小强;蔡明池;许玉珍;彭俊;梁声强

解放军联勤保障部队第909医院(厦门大学附属东南医院)检验科,福建漳州 363000中南大学湘雅医学院附属株洲医院检验科,湖南株洲 413007

临床医学

念珠菌菌血症药敏试验危险因素预测评分模型

candidaemiadrug sensitivity testrisk factorspredictionscoring model

《解放军医学杂志》 2024 (003)

280-287 / 8

This work was supported by the Military Equipment Scientific Research Project of PLA(JK20191A040324),and the Science and Technology Military Support Project of Zhangzhou(ZZ2020KD04) 军队军事科学装备科研项目(JK20191A040324);漳州市拥军课题(ZZ2020KD04)

10.11855/j.issn.0577-7402.1385.2023.0626

评论