| Objective Logistic regression model combined with Roc curve was used to construct and verify the clinical differential model of spotted fever(SF)and severe fever with thrombocytopenia syndrome(SFTS).To provide a reference for early clinical differentiation of SF and SFTS.Methods A total of 177 laboratory-confirmed cases of SF and SFTS seen in secondary and higher medical institutions in Lu’an City between May 2017 and May 2021 were selected as study subjects,and 124 cases(70%)were randomly selected from them according to 7:3 as the modeling group for constructing the scoring model,and the remaining 53 cases(30%)were selected as the internal validation group,and a total of63 clinically diagnosed cases of SF and SFTS seen in secondary and higher medical institutions in Lu’an City between June 2021 and May 2022 were selected as the external validation group for validating the efficacy of the scoring model.Demographic,epidemiological,clinical and laboratory data of the cases were collected,and each characteristic index related to the diagnosis of the two diseases was extracted as an independent variable,and a regression equation was established using a dichotomous multi-factor logistic regression analysis.The regression model was then returned to the corresponding case data to calculate the corresponding scores for each case,and then the receiver operator characteristic(ROC)curve was plotted to determine the optimal cut-off value and the area under curve(AUC)to evaluate the effectiveness of the regression model.The discriminatory effect of the regression scoring model was evaluated.The Hosmer-Lemeshow test was used to evaluate the model fitting effect.The internal validation group and external validation group populations were substituted into the established model and optimal cutoff values,while laboratory tests were performed on the external validation group population samples to assess the accuracy of the differential model based on the calculated sensitivity,specificity and Youden index.Results A total of 240 cases were included in this study.The results of the multifactorial logistic regression analysis of the modeling group showed that the presence of rash(OR=174.988,95% CI:8.924-3431.117)and elevated C-reactive protein(OR=39.356,95% CI:2.035-761.285)were positively associated with SF disease,and the presence of thrombocytopenia(OR=0.061,95% CI:0.005-0.830)and leukopenia(OR=0.054,95%CI:0.003-0.822)were negatively correlated with SF disease;while the above four indicators in SFTS patients were correlated with SF in the opposite direction.The rash(β=5.165),C-reactive protein(β=3.673),thrombocytopenia(β=-2.795),and leukopenia(β=-2.927)were assigned scores of 5,4,-3,and-3,respectively,according to the standard regression coefficient,with a total score of-6 to 9.The backgeneration results of the SF and SFTS discrimination score model constructed with these four indicators showed that the area under the ROC curve of the modeled group was 0.990,and when the cutoff value was taken as 2,the sensitivity of SF and SFTS discrimination was93.02%,the specificity was 98.77%,and the Youden index was 0.918.The Hosmer-Lemeshow test P>0.05 and the Kappa value was 0.928.The internal validation group had an area under the curve of 0.995,sensitivity of 94.74%,specificity of 97.06%,and the Youden index of 0.918.The Hosmer-Lemeshow test P>0.05 and the Kappa value of 0.918.The external validation group had an area under the curve of 0.875,sensitivity of 72.73%,specificity of 90.24%,and the Youden index of 0.630.The Hosmer-Lemeshow test P>0.05 and the Kappa value of 0.643.Conclusion The high accuracy and reliability of the differential scoring model for SF and SFTS based on four factors: rash,C-reactive protein,thrombocytopenia,and leukopenia suggest that the differential model has good discriminatory ability for SF and SFTS,and can be used for early differential of SF and SFTS,which can provide a reference basis for the differentiation of SF and SFTS. |
PDF Full Text Request