Establishment of a model for predicting the efficacy of third-generation cephalosporin in treatment of community-acquired spontaneous bacterial peritonitis
10.3969/j.issn.1001-5256.2022.11.012
- VernacularTitle:第三代头孢菌素治疗社区获得性自发性细菌性腹膜炎的疗效预测模型
- Author:
Longchuan ZHU
1
;
Wei WU
2
;
Bo ZOU
1
;
Dakai GAN
1
;
Xue LIN
1
;
Wei ZHOU
3
;
Molong XIONG
1
Author Information
1. Department of Liver Disease, The Ninth Hospital of Nanchang, Nanchang 330002, China
2. Department of Digestion, Children's Hospital of Jiangxi Province, Nanchang 330000, China
3. Department of Information Technology, The Ninth Hospital of Nanchang, Nanchang 330002, China
- Publication Type:Original Articles_Liver Fibrosis and Liver Cirrhosis
- Keywords:
Liver Cirrhosis;
Peritonitis;
Community-Acquired Infections;
Cephalosporins;
Logistic Models
- From:
Journal of Clinical Hepatology
2022;38(11):2499-2504
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the factors for predicting the efficacy of third-generation cephalosporin (3 rd GC) in the treatment of community-acquired spontaneous bacterial peritonitis (CASBP), and to establish and validate an efficacy predictive model for 3 rd GC in the treatment of CASBP. Methods A retrospective analysis was performed for the clinical data of the patients with liver cirrhosis and CASBP who received 3 rd GC monotherapy for initial treatment in The Ninth Hospital of Nanchang, and according to their treatment outcome, they were divided into non-response group and response group. The patients hospitalized from 2013 to 2018 were included in the modeling cohort (55 patients the non-response group and 110 in the response group), and those hospitalized from 2019 to 2020 were included in the validation cohort (17 patients in the non-response group and 43 in the response group). In the modeling cohort, the two groups were compared in terms of the indices including general information, underlying diseases, past history, clinical manifestation, and laboratory test results. Univariate analyses (the t -test was used for comparison of normally distributed continuous data between groups, and the Mann-Whitney U test was used for comparison of non-normally distributed continuous data between groups; the chi-square test or the Fisher's exact test was used for comparison of categorical data between groups) and a binary Logistic regression analysis were used to identify efficacy predictors, and an efficacy predictive model was established based on the logistic regression equation. The receiver operating characteristic (ROC) curve was plotted to perform internal and external validations of the model in the modeling cohort and the validation cohort, respectively. Results The study population had a mean age of 51.6±12.0 years, and male patients accounted for 80.0%; hepatitis B was the main cause of liver cirrhosis (66.7%), and 3 rd GC had an overall response rate of 68.0%. In the modeling cohort, compared with the response group, the non-response group had significantly lower proportion of patients with the first onset of SBP, polymorphonuclear (PMN) count in ascites, and leukocyte count in ascites (all P < 0.05), as well as significantly higher proportion of patients with exposure to broad-spectrum antibiotic and platelet count (both P < 0.05). The Logistic regression analysis showed that the first onset of SBP (odds ratio [ OR ]=0.158, 95% confidence interval [ CI ]: 0.064-0.392, P < 0.001), ascites PMN count ( OR =0.728, 95% CI : 0.530-0.998, P =0.046), exposure to broad-spectrum antibiotic ( OR =9.152, 95% CI : 1.513-55.351, P =0.016), and platelet count ( OR =1.012, 95% CI : 1.006-1.019, P < 0.001) were independent predictive factors for non-response to 3 rd GC. The efficacy predictive model had an area under the ROC curve (AUC) of 0.840, and based on the maximum Youden index, predictive score ≥ 0.207 was the optimal cut-off value for predicting non-response, with a corresponding Youden index of 0.536, a sensitivity of 89.1%, a specificity of 63.6%, a positive predictive value of 55.1%, and a negative predictive value of 92.1%. This model had an AUC of 0.837 in the validation cohort. Conclusion The first onset of SBP and higher ascites PMN count are the protective factors against non-response to 3 rd GC for the treatment of CASBP, and exposure to broad-spectrum antibiotic and higher platelet count are the risk factors for such non-response. The model established for predicting the efficacy of 3 rd GC in the treatment of CASBP has good predictive performance and thus holds promise for clinical application.
