Analysis of influencing factors of hospitalization expenses in elderly patients with injury based on quantile regression and random forest
WANG Lei,YANG Yaya,ZHAO Yini,DOU Feng,HU Xiaobin
DOI: https://doi.org/10.16386/j.cjpccd.issn.1004-6194.2023.04.002
2023-01-01
Abstract:Objective To understand the influencing factors of the hospitalization expenses of injured patients （≥65 years old） in Gansu Province,and provide the reference for effectively carrying out the elderly injury intervention and reducing the direct economic disease burden caused by elderly injury. Methods Multi-stage stratified cluster sampling was used to obtain 12 325items of hospitalization expenses and patient baseline information about injured inpatients （≥65 years old） in Gansu Province in2019. Mann-Whitney U test and Kruskal-Wallis H test were used to analyze the data,the multiple linear regression,quantile regression,and random forest algorithms were used to analyze the influencing factors. The used software was SPSS 22.0 or R 4.1.2.Results The total hospitalization cost of elderly injury patients was 215 712 936.4 yuan,the average hospitalization cost of patients was 17 502.1 yuan. The per capita consumables cost（ 6 911.9 yuan）,per capita drug cost（4 155.1 yuan） and per capita treatment fee（3 052.8 yuan） were the top three expense types,accounting for 39.49%,23.74% and 17.44%,respectively. The multiple linear regression analysis showed that age（β=0.027）,hospitalization duration（β=0.391）,number of comorbidity（β=0.083）,gender（β=0.019）,surgery（β=0.478）,institution type（compared with general hospitals,the traditional Chinese medicine hospital β =-0.105,maternal and child health care hospital β=-0.034） and institution level（compared with provincial level,municipal levelβ=-0.279,district or county level β=-0.218） could significantly influence the average hospitalization cost of patients（P＜0.01）;the results of quantile regression analysis showed that age,hospitalization duration,number of comorbidity,surgery,institution type and institution level had the effects at different quantiles（P＜0.01）;the results of random forest regression analysis showed that the top five influencing factors of importance were surgery,hospitalization duration,institutional level,injury site and number of comorbidity. Conclusion In the study,the burden of hospitalization in elderly injured patients is too heavy,and the cost composition needs to be further optimized,so graded diagnosis and treatment should be encouraged,the hospitalization duration should be reasonably shortened,medical consumables should be introduced into medical insurance,and universal medical insurance coverage should be promoted.