Research On The Diagnosis Related Groups And Performance Evaluation Of A Hospital In Chengdu

Abjective: To understand the specific process and evaluation indicators of the CHS-DRGs grouping scheme,to explore and establish a systematic DRGs subdivision optimization method,to provide a reference basis for the development of localized DRGs groupers,to clarify the application approach of DRGs grouper in medical units performance evaluation,to provide new ideas for fine management of hospitals.Methods: Based on the CHS-DRGs guidance program,the relevant cases in the MDCC and MDCK primary diagnosis groups of inpatients in our hospital are divided into ADRG basic groups,and the ADRG groups which need to be subdivided are screened using CV and RIV values.Based on multiple linear regression model to screen severe CC categories,calculate patients' CCI.Using data mining technology to analyze influencing factors,taking the important factors in the results as input variables,the hospitalization expenses as output variables,to establish CART and CHAID decision tree groupings model and evaluate the optimization effect of the grouping scheme after subdivision.Using comprehensive index method to evaluate the performance of related departments,DRGs-related indicators(number of DRGs groups,total weight,CMI,cost/time consumption index)are used to evaluate the differences in service capacity and efficiency of each department.Results: 1.The ADRG group classification results show that the MDCC and MDCK both had 4 ADRG groups with a hospitalization cost CV value greater than 0.8,which account for 25.00% and 30.77% of the total number of ADRG groups,respectively.CW1(various types of cataracts)and KF1(skin grafting and/or debridement due to endocrine,nutrition,and metabolic disorders)were the ADRG with the highest cost CV value respectively in MDCC and MDCK,comprehensively considering the optimization effect on the overall grouping plan,finally selected cataract(CB3 + CW1)and diabetes(KS1)patient groups as the DRGs subdivision object.2.The highest proportion of cataract and diabetes patients' hospitalization expenses were material costs(55.02%)and diagnosis costs(46.5%),respectively.There were respectively 12 and 64 kinds of severe CC category for cataract and diabetes patients.Among the three data mining regression models,the prediction results of ANN and LSVM were better.Cataract patients screened out five important factors such as crystal type,number of operations,and CCI and diabetes patients screened out six important influencing factors such as age,CCI,and critical illness/severity.Decision tree models results show that the CART model had a higher correlation coefficient and more accurate cost prediction than CHAID model.Cataract patients and diabetes patients respectively formed 9 and 7 DRG groups.The RIV values respectively were 74.65% and 62.01%,and the CV values of all DRG groups were less than 0.7.3.According to the evaluation of DRGs-related indicators,the medical service capability and efficiency of ophthalmology department 1 were higher than those of ophthalmology department 2.Among the endocrinology departments,the medical service efficiency of the elderly endocrinology department was the lowest,while the medical service capability of the Caotang endocrinology department was the lowest.Conclusion: Combining multiple data mining models can avoid the bias caused by a single analysis method.The CART decision tree model can be used as an effective method for the establishment of DRGs grouper.The degree of cost variability of cataracts and diabetes patients are high,and it is recommended to appropriately increase the subdivision group in the localized DRGs grouping plan.DRGs is an effective tool for performance management which can enhance the comparability between medical service units and urge medical staff to improve the quality of services.