| Objective1.To investigate the accuracy,sensitivity and specificity of PEWS in identifying critical hand-foot-mouth disease(HFMD)by retrospectively analyzing the clinical data.2.To explore independent risk factors for critical HFMD to provide a scientific basis for the prevention and treatment of critical HFMD and the construction of a HFMD-PEWS model by retrospectively collecting the clinical data.3.To use a scientific statistical method to score independent risk factors for critical HFMD quantitatively and construct the HFMD-PEWS model in combination with PEWS.4.To apply the HFMD-PEWS model constructed in this study clinically,evaluating the clinical application effect of this early warning model.Methods1.Retrospective study:A total of 2488 children with severe and critical HFMD(2166were severe and 322 were critical)who were admitted to Qingdao Women and Children's Hospital between January 2014 and December 2016 were enrolled.(1)By reviewing the electronic medical records(EMRs)archived by the hospital,PEWSs were scored according to the monitoring values of vital signs documented in EMRs;using a receive operating characteristic(ROC)curve,the area under the ROC curve,the cut-off value,the sensitivity,the specificity,and the Youden index were calculated regarding PEWS identifying the exacerbation from severe disease to critical disease,and the accuracy of PEWS in identifying the critical HFMD was assessed.(2)According to the general information,clinical characteristics,laboratory tests,and imaging results,variables meaningful to the critical disease were screened by one-way ANOVA;meaningful variables were introduced into a multivariate logistic regression model to screen out statistically significant independent risk factors,and their P values,odd ratios(ORs)and 95%confidence intervals(95%CIs)were calculated,respectively.(3)Chi-square automatic interaction detection(CHAID)in decision tree technology was used to analyze independent risk factors for critical HFMD,screen out parameters and score them to reflect their disease severity correspondingly;combined with PEWS,HFMD-PEWS model was constructed and further re-validated by using the ROC curve;analysis was conducted on the accuracy,sensitivity and specificity of HFMD-PEWS model in identifying critical HFMD.2.Application study:Children with severe and critical HFMD admitted in 2018 were divided into a control group and an experimental group.In the control group,doctors determined whether to refer the children to the PICU according to related clinical diagnostic markers;in the experimental group,doctors determined whether to refer the children to the PICU according to HFMD-PEWS scores.Using percentage,mean,univariate t-test,and?~2test,both groups of children were compared regarding the accuracy of critical identification,the successful rescue rate,the incidence of neurological sequelae,the mortality,the doctor-patient satisfaction,the average length of stay,and the cost.Results1.For PEWS predicting critical HFMD,the area under the ROC curve was 0.674(95%CI:0.640-0.707);when the Youden index was up to 0.308,the sensitivity was 0.635,the specificity was 0.673,and the cut-off value was 2.5 points.2.A total of five statistically significant independent risk factors were screened out by multivariate logistic regression analysis,including age of<3 years(P=0.001,OR=1.868),body temperature of?39°C(P=0.000,OR=1.525),blood glucose of?8 mmol/L(P=0.000,OR=2.824),limb shaking or tremor(P=0.009,OR=1.576),and herpes in hands(P=0.011,OR=1.753).3.According to the results of the CHAID classification tree,blood glucose and body temperature were included in the final warning model.Specific scoring criteria were set up according to the results of the classification tree:blood glucose,0 for?6.2 mmol/L,1 for6.2-7.0 mmol/L,2 for 7.1-7.9 mmol/L,and 3 for?8.0 mmol/L;body temperature:0 for<37.5°C,1 for 37.5-38°C,2 for 38.1-38.8°C,and 3 for?38.9°C.The HFMD-PEWS model included nine evaluation indexes of five dimensions:nervous system,respiratory system,circulatory system,blood glucose,and body temperature.Each dimension scored 0 to 3,with a total score of 15 points.When the HFMD-PEWS model was re-validated,the area under the ROC curve was 0.911(95%CI:0.896-0.926);when the Youden index was up to0.684,the sensitivity was 0.910,the specificity was 0.774,and the cut-off value was 6.5points.4.Compared with the control group,ventilator-assisted mechanical ventilation cases reduced(P=0.002),doctor-patient satisfaction(P=0.035),with lower incidence of neurological sequelae(P=0.005),shorter average length of stay(t=29.000,P=0.022)and lower costs(t=33.000,P=0.019),and there were significant differences between both groups.There were no differences in successful rescue rate and mortality between both groups(P=0.135).Conclusions1.When PEWS identifies critical HFMD,the accuracy,the sensitivity and the specificity are low.It is necessary to explore independent risk factors for critical HFMD,increase warning scoring markers,and enhance the identification effect of PEWS on critical HFMD.2.Age of<3 years,body temperature of?39°C,blood glucose of?8 mmol/L,limb shaking or tremor,and herpes in hands and feet were independent risk factors for critical HFMD,which can serve as early screening markers for critical HFMD.3.Using the decision tree technology in this study,we have finally integrated the blood glucose and body temperature into the warning model,constructed a HFMD-PEWS model in combination with PEWS,re-validated the accuracy,sensitivity and specificity of its identification of critical HFMD dramatically.It can be used as the basis for the construction of HFMD-PEWS.4.Clinical application of HFMD-PEWS model can early warning the critical HFMD accurately,directing health care providers'clinical intervention and treatment of critical HFMD. |
PDF Full Text Request