Study on the diagnosis related groups of cerebral infarction patients based on decision tree model
Objective To explore the case combination scheme of patients with cerebral infarction on diagnosis related groups(DRGs),and provide the reference for relevant departments to formulate the payment standard of cerebral infarction hospitalization expenses.Methods The data were from the network direct report data of the Health Information Statistics Center of Gansu Provincial Health Commission,and 35 039 cerebral infarction patients treated in tertiary public hospitals in Lanzhou,Gansu Province from January 1,2016 to December 31,2020 were selected as the subjects.The first-page medical records data were collected.Mann-Whitney U test and Kruskal-Wallis H test were used to analyze the data.Multiple linear regression and quantile regression were used to analyze the influencing factors of hospitalization expenses;the decision tree model based on the chi-square automatic interaction detection algorithm(CHAID method)was used to construct the case combination of DRGs and formulate the cost standard of each group.The used software was SPSS 26.0 and Stata 17.0.Results The average hospitalization expense of 35 039 patients with cerebral infarction was 14 019.62 yuan,and the median of hospitalization expenses was 9 649.21 yuan.Multiple linear regression analysis showed that ethnic minorities(β=0.046),medical payment method(the basic medical insurance system for urban working people served as reference,new rural cooperative medical insurance system β=0.080,others β=0.068),long hospitalization days(β=0.445)were positively correlated with hospitalization costs(P＜0.01).Female(β=-0.047),higher age(β=-0.038),medical payment method(the basic medical insurance system for urban working people served as reference,basic medical insurance system for urban residents β=-0.050),admission route(the emergency department served as reference,outpatientβ=-0.275,othersβ=-0.040),without complication(β=-0.024)and no operation(β=-0.208)were negatively correlated with hospitalization costs(P＜0.01).Quantile regression analysis showed that ethnic minorities and long hospitalization days were positively correlated with hospitalization costs at the three subsites(Q10,Q50 and Q90),P＜0.01.Female,admission route and no operation were negatively correlated with the hospitalization cost(P＜0.01).Twenty-two case combinations were constructed based on hospitalization days,surgical operation,admission route and medical payment method.Among them,the number of patients with DRGs19(no surgical operation,hospitalization days of 8-14 days,payment method of basic medical insurance system for urban working people)was the largest(9 486 cases).The average hospitalization cost of patients with DRGsl(surgical operation,hospitalization days of 22 to 30 days,payment method of basic medical insurance system for urban residents,new rural cooperative medical system or other)was the highest(97 360.13 yuan).There were 2 030 cases(5.79％)of excessive hospitalization expenses,and a total of 23.18％of medical resources was consumed.Conclusion The subgroups of hospitalization cost constructed by decision tree model is scientific and reasonable,the grouping scheme and cost standard can provide some reference for controlling the hospitalization expenses of patients with cerebral infarction.
Cerebral infarctionDiagnosis related groupsHospitalization expenseDecision treeInfluencing factors
NETL
NSTL
万方数据
