| Objective The inverse probability weighted semi-parameter quantile two-part model was applied to study the medical expenses of the elderly to solve the problems of non-normal,zero inflation and many missing values of the medical expenditure data of the elderly and to explore the response of different medical expenditure levels to its influencing factors,and to provide reference for the realization of healthy aging.Methods Using the China Health and Retirement Tracking Survey(CHARLS)2018database.And a total of 11,032 elderly persons aged 60 years and older were included in the study.R 4.0.4 software was used to establish a inverse probability weighted semi-parameter quantile two-part model to explore the influencing factors of medical decision making and cost in the elderly.Inverse probability weighted parameter quantile regression model and non-inverse probability weighted models were also established to compare the inverse probability weighted effect.Results Older age(β=0.083,P=0.012),female(β=0.0135,P=0.001),urban employee medical insurance(β=0.411,P=0.002),urban employee medical insurance(β=0.320,P=0.006),self-rated health status(β=0.702,P <0.001),poor self-rated health status(β=1.571,P<0.001)will significantly increase the probability of medical expenditure for the elderly.Old age,female,health insurance and poor self-rated health status significantly increased the probability of medical expenditure.The medical expenditure of the urban elderly is higher than that of the rural elderly,and this difference is more obvious in the elderly group with low medical expenditure.The lower self-rated health status,the greater the effect of increasing medical expenditure.There is little difference in the medical expenditure of the elderly between different genders,and females are slightly higher than males in general.Medical insurance will increase the elderly’s medical expenditure,the effect of urban workers’ medical insurance is greater than that of urban and rural residents’ medical insurance.The effect of age on medical expenditure of the elderly is irregular,and its value first increases with the increase of age,and then decreases after reaching the inflection point.And the higher the quantile,the higher the overall effect value.The overall effect of individual annual income on medical expenditure is U-shaped,that is,the effect is higher in low-income and high-income elderly population,while the effect is lower in middle-income elderly population.The result of inverse probability weighted models were more accurate than the unweighted ones.The semi-parametric quantile model fitted better to independent variables with non-linear effects such as age and individual annual incomeConclusion Old age,female,medical insurance and poor self-rated health status will increase the probability of medical expenditure.Age,annual income,gender,urban and rural distribution of residence,self-rated health status and medical insurance are important factors affecting the medical expenses of the elderly.Inverse probability weighted semi-parameter quantile two-part model used in the study on the elderly medical spending had good effect.Because it not only considered the parameters and non-parametric factors at the same time,overcame the medical spending data zero inflation,abnormal distribution,missing value,but also revealed the influencing factors under different levels of medical spending. |
