Quantitative Study On The Diagnosis And Evaluation Criteria Of Gastrointestinal Fever Syndrome In Children

Background:Gastrointestinal heat retention syndrome(GHRS)refers to a condition that is associated with accumulated gastrointestinal heat caused by a metabolic block in energy.The core pathogenesis of GHRS is hyperactivity of heat evil,stagnation of Qi and injury of fluid.GHRS,a syndrome with digestive dysfunction,is associated with exogenous pathogens,diet structure and habits,lack of exercise and insufficiency of natural endowment.Children are more likely to cause digestive disorders,for their digestion function is immature and always vulnerable to suffering GHRS in the childhood stage of growth and development.Children suffer from GHRS commonly,and is closely related to the occurrence and development of a variety of pediatric diseases.Our team have developed the diagnostic scale of GHRS.However,there have still been some problems,such as limited ability to distinguish the diagnosis and evaluation function,high false negative rate,improper classification description of some itemsObjective:To establish the diagnostic model and evaluation scale for children's GHRS and provide a basis for the study on prevention and treatment of diseases associated with GHRS.This study consisted of two parts:one was the establishment and evaluation of the diagnostic model of GHRS;The second was the development and evaluation of the evaluation scale of GHRS.1 The establishment and evaluation of the diagnostic model of GHRSMethods:(1)The information of four examination methods related to GHRS was extracted based on the case reports in the literature and the description in textbooks and guidelines.The initial item pool of the diagnostic model was formed by group discussing.(2)Two-round Delphi method was used to screen the symptoms.65 experts were selected to score each symptom according to the significance in the diagnosis of GHRS using the Likert scale.During questionnaire design,open-ended questions were also designed to fully obtain experts' opinions.The items were selected by boundary value method according to mean expert score,full mark rate and variable coefficient of expert scores.(3)The core items of the diagnostic model were screened by Lasso analysis based on the clinical data of 660 children aged 3-14.(4)The data were divided into training set and test set according to 2:1.Four machine learning algorithms,logistic regression analysis,random forest,support vector machine and eXtreme Gradient Boosting method,were used to construct the GHRS diagnosis model based on the training set,and the optimal model was selected according to the accuracy based on the test set.(5)The diagnostic model was verified internally,and its accuracy,sensitivity and specificity were evaluated based on the test set.Results:41 items were contained in the original item pool,and 38 items were screened by the Delphi method.342 of the 660 cases were in disease states,while the other 318 were in non-disease states.There was no significant difference in gender between the positive GHRS group(n=453)and negative GHRS group(n=207).The core symptoms screened by using Lasso analysis included thick coating,dry stool,abnormal appetite(enormous or poor),reduced frequency of defecation,restlessness at night sleep,halitosis,sweating at night,yellow coating,vexation,red tongue,slippery pulse,feverish feeling in palms and soles,smelly stool,worse after improper diet(crapulence or indulgence for greasy foods),yellow urine,red lips,vomiting,belching,and hard defecation.All items were asked to answered with exist or not.The diagnostic model of GHRS by the eXtreme Gradient Boosting method performed the best.The accuracy of the diagnostic model reached 93.03%.The sensitivity,omission rate,specificity,and mistake diagnosis rate of the model were 95.07%,4.93%,88.14%,and 11.86%,respectively.2 The development and evaluation of the evaluation scale of GHRSMethods:(1)The item pool was developed according to the method of studies as above.(2)Two-round Delphi method was used to screen the symptoms.75 experts were selected to score each symptom according to the significance in the diagnosis of GHRS using the Likert scale.During questionnaire design,open-ended questions were also designed to fully obtain experts' opinions.The items were selected by boundary value method according to mean expert score,full mark rate and variable coefficient of expert scores.(3)Classical Test Theory and Item Response Theory based on the clinical data of 453 children with GHRS were used to select items.Classical Test Theory included discrete tendency analysis,correlation coefficient analysis,Cronbach's a coefficient analysis and factor analysis.The items were selected by at least 4 methods.The evaluation scale was built by equally weighted method.The degree boundary value was set according to the characteristics of data distribution.The reliability,validity,reactivity and differential item functioning of the scale were evaluated by the clinical raw data.Results:41 items were contained in the original item pool,and 38 items were screened by the Delphi method.310 of the 453 cases were in disease states,while the other 143 were in non-disease states.The mean of the GHRS score was 5.80(SD 2.31).The evaluation scale consisted of 26 items,including red complexion,red lips,pharyngeal redness and swelling,red tongue,yellow tongue fur,feverish feeling in palms and soles,rapid pulse,slippery pulse,aversion to heat,halitosis,hot mouth and nasal breath,thirst with preference for cold drinks,abnormal appetite(enormous or poor),abdominal pain,smelly flatus,reduced frequency of defecation,dry stool,difficult defecation,smelly stool,yellow urine,sweating at night,restless sleep at night,vexation and irascibility,nasal crust,susceptibility to respiratory tract infection,and symptoms worsened after improper diet(crapulence or indulgence for greasy foods).There were 11 binary variables and 15 quadruple variables graded by frequency or severity degree.Each item was weighted equally.The degree of GHRS was light if the score was less than 15 points,the degree of GHRS was moderate if the score was 15 points?total score<28 points,the degree of GHRS was severe if the score was more than 28 points.The half-reliability and Cronbach's ? coefficient of the scale were 0.656 and 0.787 respectively.The standard validity was 0.515,and the P value was<0.001.There was difference in scale score between the top 27%and the last 27%of the subjects.Most items had no functional differences between the two groups of 3 years<age?7 years and 7 years<age?14 years.Conclusion:In this study,the diagnostic model and evaluation scale were developed for children aged from 3 to 14 years old.The accuracy,sensitivity and specificity of the diagnostic model were appraised as a high level to identify GHRS.And the reliability,validity and responsiveness of the evaluative scale were good enough to evaluate the degree of GHRS.In addition,most items of scale had no functional differences in age.