Study On The Hospitalization Expenses And Case Mix For Diabetes Inpatients In Tianjin

Objective:To investigate the structure of hospitalization expenses of diabetes inpatients, the data of diabetes inpatients insured in Tianjin from2003to2007were analyzed, and the medical expenditures were compared between patients with different characteristics. Multinomial linear regression and path model were used to analyze the factors influencing hospitalization expenses. Then, try to establish the case mix model which will be applicable to diabetes inpatients in Tianjin by using decision tree analysis. The proportion of expense exceeding the threshold and patients exceeding the threshold in diabetes inpatients of Tianjin were computed to find the control points and provide the basis for the development of health cost control policies and the improvement of payment of medical expenses.Methods:9347qualified diabetes patients were collected from15%cases which selected randomly from medical insurance database within2003to2007of Tianjin residents. Multinomial linear regression and path model were used to analyze the factors influencing hospitalization expenses with SPSS16.0and AMOS7.0. Establish the case mix model and analyze the proportion of expense exceeding the threshold and patients exceeding the threshold in diabetes inpatients of Tianjin by using SAS9.1.Results:1.The costs of drug, examination and treatment took the largest proportion of the hospitalization expenses, and drug costs accounted for more than half of the total costs. The medical costs and length of hospitalization were compared between patients with different characteristics. The differences of hospitalization expenses of type1diabetes patients were statistically significant in various levels of variables as follows:age, year, occupation, proportion of payment, region, hospital, surgery or not, transfusion or not, occurrence of ketosis or not, occurrence of peripheral artery disease (PAD) or not, occurrence of diabetic foot disease or not. The differences of length of hospitalization were statistically significant in various levels of variables as follows:year, region, hospital, surgery or not, occurrence of heart disease or not, occurrence of PAD or not, occurrence of diabetic foot disease or not and occurrence of hypertension or not. The differences of hospitalization expenses of type2diabetes patients were statistically significant in various levels of variables as follows:sex, age, year, occupation, proportion of payments, region, hospital, surgery or not, transfusion or not, complications, occurrence of hypoglycemia or not, occurrence of eye diseases or not, occurrence of renal diseases or not, occurrence of brain diseases or not, occurrence of neuropathy or not, occurrence of PAD or not, occurrence of diabetic foot disease or not, occurrence of hypertension or not, occurrence of lipid metabolism disorder or not. The differences of length of hospitalization were statistically significant in various levels of variables as follows:sex, year, occupation, proportion of payment, region, hospital, surgery or not, transfusion or not, complications, occurrence of ketosis or not, occurrence of hypoglycemia or not, occurrence of renal diseases or not, occurrence of heart diseases or not, occurrence of neuropathy or not, occurrence of PAD or not, occurrence of diabetic foot disease or not, occurrence of lipid metabolism disorder or not (P<0.05).2. By multinomial linear regression analysis, for type1diabetes patients, the major influential factors included length of hospitalization, surgery or not, transfusion or not, proportion of payments; for type2diabetes patients, the major influential factors were length of hospitalization, hospital’s level, surgery or not, complications, transfusion or not, year, occupation, occurrence of hypertension or not, region, occurrence of lipid metabolism disorder or not, occurrence of diabetic foot disease or not, occurrence of hypoglycemia or not, occurrence of PAD or not.3. By path analysis, for type1diabetes patients, surgery or not, occurrence of diabetic foot disease or not impact on the hospitalization expenses directly and indirectly; age, year, region, complications, occurrence of ketosis or not impact on the hospitalization expenses indirectly. For type2diabetes patients, age, year, occupation, surgery or not, transfusion or not, hospital’s level, complications, occurrence of ketosis or not, occurrence of diabetic foot diseases or not, occurrence of lipid metabolism disorder or not impact on the hospitalization expenses directly and indirectly; proportion of payments, region, occurrence of hypoglycemia or not, occurrence of eye diseases or not, occurrence of brain diseases or not, impact on the hospitalization expenses indirectly.4. The outcomes included40groups, of which7groups classified from type1diabetes mellitus,33groups classified from type2diabetes mellitus. Assess the model by train data with RIV, CV, and nonparametric test. All indicators showed good results. Compute the case mix index(CMI), standard cost and threshold of medical cost according to the case mix model, results showed CMI had positive correlation with medical cost; the proportion of patients exceeding the threshold in type1and type2diabetes inpatients of Tianjin was3.05%and12.46%,respectively; the proportion of expense exceeding the threshold was3.31%and6.22%, respectively;10.97%of the total expenses of type1diabetes was expended by patients exceeding the threshold;23.32%of the total expenses of type2diabetes was expended by patients exceeding the threshold; the results also indicated that the proportion of patients exceeding the threshold had positive correlation with the proportion of expense exceeding the threshold.Conclusion:By using multinomial linear regression and path model analyze the influential factors of diabetes patients’ hospitalization expenses, we not only find out the meaningful factors, but also find out the specific roles of the factors on the hospitalization cost. Additionally, compared with previous studies, the grouping variables of our case mix model included both variables reflected the clinical characteristics and those reflected the social characteristics of patients, achieved the study aim. In subsequent study, we need to enhance the quality of records management, cost measuring, the standardization of clinical pathway and so on.