感染性眼内炎病原菌菌谱、药敏性及风险预测模型构建

Spectrum and drug susceptibility of pathogens and construction of risk prediction model in infectious endophthalmitis

目的分析感染性眼内炎病原菌菌谱、药敏性,分析影响发生感染性眼内炎的危险因素,建立Logistic回归风险预测模型。方法回顾性分析2017年3月至2022年6月期间在本院眼科住院的20750例患者的临床资料,根据术后是否发生感染性眼内炎将患者分为感染性眼内炎组(n=83)和非感染性眼内炎(n=20667)。统计其术后发生感染性眼内炎标本中致病菌培养和药敏试验结果;记录两组年龄、性别、手术时间等一般资料;采用二分类Logistic回归分析影响眼科手术患者发生感染性眼内炎的因素,并建立风险预测模型,Hosmer-Lemeshow检验评估模型拟合度,ROC检测该模型的预测效能。结果共83例(0.04%)患者术后发生感染性眼内炎。83例送检标本中,检出病原菌株49株,阳性率为59.04%。其中,革兰阳性菌占比65.31%,革兰阴性菌占比26.53%,真菌占比8.16%。药敏试验结果显示,表皮葡萄球菌、肺炎链球菌对万古霉素、加替沙星敏感性最高,铜绿假单胞菌、大肠埃希菌对环丙沙星、头孢吡肟敏感性最高,曲霉菌属和镰刀菌属对伏立康唑和两性霉素B敏感性最高。Logistic回归分析,患者年龄≥50岁(OR=2.34,P=0.039)、合并糖尿病(OR=2.751,P=0.016)、手术时间≥10 min(OR=2.649,P=0.021)及白蛋白<30g/L均为患者发生感染性眼内炎的独立危险因素。建立发生感染性眼内炎的Logistic回归风险预测模型:P=1/[1+e(-2.307+0.089×(年龄)+1.012×(糖尿病)+0.974×(切口大小)],Hosmer-Lemeshowχ^(2)=7.685,P=0.079;ROC分析显示,Logistic回归风险预测模型预测PIFI的AUC为0.782(P<0.05)。结论感染性眼内炎主要致病菌为表皮葡萄球菌、肺炎链球菌和铜绿假单胞菌,药敏结果显示表皮葡萄球菌、肺炎链球菌和铜绿假单胞菌对万古霉素、加替沙星和环丙沙星敏感性高。Logistic回归风险预测模型可以较好地预测感染性眼内炎的发生,临床可重点关注年龄≥50岁、合并糖尿病和手术时间≥10 min的患者。

Objective This study aimed to analyze the spectrum of pathogenic bacteria causing infectious endophthalmitis and their drug sensitivity,identify the risk factors influencing the incidence of this condition,and develop a Logistic regression risk prediction model.Methods Retrospective analysis was conducted on clinical data from 20,750 patients admitted to the ophthalmology department between March 2017 and June 2022.Patients were divided into two groups:infectious endophthalmitis group(n=83)and non-infectious endophthalmitis group(n=20,667),based on whether they developed infectious endophthalmitis after surgery.The study identified the pathogenic bacteria causing postoperative infectious endophthalmitis and analyzed their drug sensitivity.The demographic data of both groups were recorded,and dichotomous Logistic regression was used to evaluate the factors that could influence the incidence of infectious endophthalmitis.Results There were 83 cases(0.04%)with infectious endophthalmitis after surgery.Among these 83 specimens,there were 49 strains of pathogens(positive rate of 59.04%),including Gram-positive bacteria(65.31%)and Gram-negative bacteria(16.33%)and fungi(8.16%).The results of drug sensitivity tests showed that Staphylococcus epidermidis and Streptococcus pneumoniae was the highest sensitive to vancomycin and gatifloxacin.Pseudomonas aeruginosa and Escherichia coli was the highest sensitive to ciprofloxacin and cefepime.Aspergillus and Fusarium was the highest sensitive to voriconazole and amphotericin B.A risk prediction model was established,and its efficacy was assessed using the Hosmer-Lemeshow test and receiver operating characteristic(ROC)analysis.Logistic regression analysis revealed that patients aged≥50 years(OR=2.34,P=0.039),those with combined diabetes mellitus(OR=2.751,P=0.016),operative time≥10 min(OR=2.649,P=0.021),and albumin<30 g/L were independent risk factors for developing infectious endophthalmitis.The Logistic regression risk prediction model of infectious endophthalmitis was as follow:P=1/[1+e(-2.307+0.089×(age)+1.012×(diabetes mellitus)+0.974×(incision size)],Hosmer-Lemeshowχ^(2)=7.685,P=0.079.The developed logistic regression risk prediction model predicted an AUC of 0.782 for PIFI(P<0.05).Conclusion In conclusion,the study found that Staphylococcus epidermidis,Streptococcus pneumoniae,and Pseudomonas aeruginosa were the main causative agents of infectious endophthalmitis,and they exhibited high susceptibility to vancomycin,gatifloxacin,and ciprofloxacin.The Logistic regression risk prediction model effectively predicted the occurrence of infectious endophthalmitis.Clinically,close attentions could be payed on patients with age not younger than 50 years,diabetes mellitus and operation time not shorter than 10min.