The Effects of PM2.5 on chronic cardio-cerebrovascular diseases and respiratory diseases in different groups

This paper explored the difference of health damage caused by PM2.5 among the rural mid-aged and elderly in the same pollution level areas. In addition, the reason for the difference is also discussed. The results indicate that one unit concentration increase in PM2.5 is related to a statistically significant 1.82% increase in the probability of suffering chronic cardio-cerebrovascular diseases, respiratory diseases for the low-income individuals. It is 0.0012 higher than those with high income. In addition, the mid-aged and elderly in the areas with high per capita GDP face a higher risk of having the related chronic diseases. Personal health awareness and local medical service infrastructure play positive roles in reducing the health impacts of PM2.5.


Hypotheses
The model of demand for "good health" constructed by Michael Grossman assumed that individuals inherit an initial stock of health that depreciates over time and can be increased by investment. Gross investments in health capital are produced through household production functions. The direct inputs include the time of the consumer and market goods such as medical care, diet, exercise, recreation, and housing, etc [8]. Based on the model presented by Grossman, M.L.Cropper(1981) noted that health capital decays at a rate which depends on air pollution [9]. From the view of Grossman and Cropper, we can infer that as health stock depreciated by the damage of PM2.5, there will be a health gap among different groups because of different economic status, health awareness, and local medical care quality. Based on the views of Grossman and M.L.Cropper, we develop the hypothesis.
Hypothesis: Under certain conditions, the health impacts derived from PM2.5 are different among different groups.

Methods
We use Probit Regression Model to estimate the relationship between PM2.5 and the probability of suffering from the diseases. As the first dependent variable was from the question, "Have you been diagnosed with cardio-cerebrovascular or respiratory diseases in the past two years?" 1 represents "yes" while 0 represents "no". The dependent variable is a binary selection problem, so the regression is established as follows: refers to the PM2.5 average concentration of the cities where the respondents lived in the past two years or present year.

Data
The main database used in this study is from the China Health and Retirement Longitudinal Survey (CHARLS), which was conducted by the National Development Research Institute of Peking University. The dataset is used to analyze the problem of population aging in China, promote interdisciplinary research on the problem of population aging, and provide a more scientific basis for formulating and improving relevant policies in China [10,11].As for the data of PM2.5, we adopt the Global Annual PM2.5 Grids from MODIS, MISR, and SeaWiFS Aerosol Optical Depth (AOD) with GWR. It is a data set released by the NASA Socioeconomic Data and Applications Center. The data of control variables at the city level are from the China Urban Statistical Yearbook.

Variables
Health: The dependent variable was obtained by asking the participants, "Have you been diagnosed with the chronic diseases listed below by a doctor in the past two years?" 1 represents "yes", while 0 represents "no". PM2.5: By using ARCGIS software, the PM2.5 grids were analyzed for the average concentration of PM2.5 at the county level, so that the data could be used for regression analysis. The data is processed with a logarithm to eliminate the influence of the dimension.
Health awareness: The information of the personal health awareness was obtained from CHARLS by asking a question, "Are you now taking any of the following treatments to treat the diseases?" There are four options for the question. Based on the selection of the respondents, we reassigned the values of the respondents' health awareness. The value for the respondent who took treatments to treat his diseases is 1; otherwise, the value is 0. Local medical service infrastructure: It takes a long time to prevent and control the diseases caused by air pollution, especially for the mid-aged and elderly living in the rural areas with poor medical service infrastructure, the cost of time and medical services may higher than that in urban. Therefore, their willingness to receive medical treatment is lower when the diseases are at the early stages, while the sicknesses may turn to chronic diseases which cause greater health losses and economic burden. The number of doctors working in the local city is used to reflect the medical service infrastructure, and the data was collected from the China Urban Statistical Yearbook.
Other control variables: This study controls the variables at individual, household, and city levels. The individual characteristic includes personal health awareness, job, age, the number of cigarettes consumed per day, years of receiving education, marriage, and gender. Meanwhile, the variables reflecting household characteristic includes the per capita household income of last year and the number of family members. Furthermore, the variables at city level include the proportion of local secondary industry output to GDP and the number of doctors working in the local city. Table 1 presents the statistic description of all variables used in the regression. It shows that the percentage of respondents who have been diagnosed with cardio-cerebrovascular or respiratory diseases in the past two years is 11.5%. The average PM2.5 concentration in the past two years was 58.446. 60.3% of the respondents had a good awareness of their health status. A majority of the participants were engaged in agricultural (76.2%), and the mean age was 60.34 years. The mean number of cigarettes consumed per day was 5.45. Most of them were illiterate, so the mean years of education they received were 3.87. 86.6% has a spouse, a little less than half were male (47.8%). The mean per capita household income of last year was about 5,320 Yuan, and each family has an average of 3 members. Among the research cities, the mean proportion of local secondary industry output to GDP was 48.94%. About 11050 doctors were working in each subject cities.

Secondgdp
The proportion of local secondary industry output to GDP (%) 48.941 7.713

Doctor
The number of doctors working in the local city (10,000 doctors) 1.105 0.915

Heterogeneous effect of PM2.5 on the health of rural mid-aged and elderly
Income inequality, personal characteristic, and other socioeconomic factors may change the influence degree of PM2.5 exposure varies by groups and individuals. In consequence, the estimated results of PM2.5 exposure among rural mid-aged and elderly will be provided to explain the influence degree and intrinsic relationship from the socioeconomic perspective. Table 2 presents the estimation results of the effects of PM2.5 on the rural mid-aged and elderly with different income levels. By using the chronic diseases of cardio-cerebrovascular or respiratory diseases as the dependent variables, the first row indicates that the increasing concentration of PM2.5 poses a significant risk to the health of rural mid-aged and elderly. However, the marginal effects of PM2.5 to the probability of related chronic diseases are varied by different income levels. The marginal effect value of the low-income individual is 0.0182 with significance at 1%, which is 0.0012 higher than those with high income. Meanwhile, the significance level of low-incomer is higher than that of mid-incomer(5%). The result suggests that the group with low economic status is more sensitive to the health impact derives from PM2.5. From the results of the other control variables, we can infer that most of the respondents with low income may engage in agricultural, which makes them exposed to the polluted air longer than their counterparts. Meanwhile, due to the limitation of economic conditions, the rural mid-aged and elderly with low income can not afford the better medical service in urban areas. Their spouses often play an essential role in taking care of them when they get sick because of PM2.5 exposure. And local medical service infrastructure has a significance and positive impact on reducing the health risk of the low-incomer.    Table 3 shows the results of the health consequences of PM2.5 on the rural mid-aged and elderly in different areas with different economic development. The results indicate that one unit concentration increase in PM2.5 is related to a statistically significant 2.35% increase in the probability of having the related chronic diseases for the mid-aged and elderly in the areas with high per capita GDP, which is 0.004 higher than those in medium per capita GDP. On the contrary, the PM2.5 does not have a significant influence on the health risk for the participants in rural areas with low per capita GDP. One possible reason is that，in order to release the environmental pressure in urban areas, some urban industrial enterprises were required to move to rural areas. Highemission enterprises contribute to economic development and cause severe air pollution at the same time. The concentration of PM2.5 increases with the development of the local economy, which produces dramatic harm to the health of the local population. While in most cases, low per capita GDP areas locate in the remote mountainous regions without big factories. As a result, the concentration of PM2.5 is too low to have a significant impact on the health of the local residents.