Data collection and quality control
Organization and implementation
The research team carried out a questionnaire survey on rural poor residents in Shaanxi province by recruiting students on the summer holiday social practice program for college students of Xi’an Jiaotong University in 2017. The summer social practice program for college students is organized and implemented by the youth league committee of Xi’an Jiaotong University. The university, together with 12 cities and districts of Shaanxi province, has established a social practice base for college students. Every summer holiday, the university will send outstanding college students to various cities and districts to work as interns in township party, government organizations, and public institutions under its jurisdiction. The research team recruited these college students as investigators to implement the questionnaire survey, and these students from the same county made up the survey team. The questionnaire included basic information of rural poor residents such as health status, income, age, the degree of education, gender and marital status, their living conditions, and their participation in the targeted poverty alleviation policy.
Sampling method
First, a three-stage sampling survey method combining probability and non-probability sampling was adopted. We selected counties (districts) with rural areas from 12 municipals in Shaanxi province, and the typical rural administrative village was selected in the counties (districts). Last, each rural village randomly selected six eligible poor households as the subject of survey. The sampling process was carried out by each survey team in accordance with scientific, typical, and convenient principles.
Data quality control
First, the college students from the social practice that participated in the survey were trained ahead of time, the structure and content of the questionnaire were explained, and the matters needing attention and principles in the questionnaire survey were emphasized. Second, every member of this survey can use the survey data to conduce a research so that they will be more responsible. Third, A preliminary investigation was conducted and the questionnaire was revised according to the problems found in the preliminary survey so that the scientificity of the formal questionnaire was guaranteed. Fourth, we used one-to-one interview to answer the questionnaire, which means that our investigators ask the rural poor residents and fill out questionnaires based on the answer. And problems existing in the survey were solved timely through network communication during the implementation of the survey. Last, we checked the questionnaires the students returned. According to the general requirements of social survey, the questionnaires is valid if there is no logic error and the main questions had been answered. After Eliminating questionnaires that lack key information such as health status, income, age, the degree of education, gender and marital status, a total of 1,233 valid samples were obtained after removing the samples under the age of 16.
Description of the health status of rural poor residents
Descriptive statistical analysis was used to describe the health status of rural poor residents, and chi-square test was used to analyze whether there were statistically significant differences in the health status of poor residents in different regions.
Measurement of health equity
A simple way to measure health equity is to test whether two groups (the poor and the rich) have the same health level. Currently, there are many methods to measure health equity, including the method of concentration curve and concentration index, method of Lorenz curve and gini-coefficient Lorenz curve, atkinson index, and chi-square value Method [19]. Although the method of concentration index and lorenz curve and gini coefficient is similar to the way of expression. The concentration index not only provides an indicator of health inequity but can also be decomposed proportionally into contributions of different inequity of health determinants[20]. Referring to existing studies on health equity of Chinese residents [21,22], this research uses the centralized index method to measure the health equity of rural poor residents. The concentration index is used to investigate the inequity degree of a certain variable associated with social and economic status, which dynamically reflects the effect of the variable influenced by income [23]. The concentration index is calculated using Equation 1 [21].
where C is concentration index, y is health status, u is the mean of health status, r is the fractional rank of income, ranging from 0 to 1. The value of the concentration index is -1 to 1. If the concentration index is 0, this shows that rural poor residents with different economic levels have the same health status. A positive concentration index indicates that people with higher incomes are healthier than those with lower incomes. Conversely, a negative concentration index indicates that people with lower incomes are healthier than those with higher incomes.
The method of decomposition of the concentration index is used to analyze the contributions of various determinants of health to the inequity in health status. Decomposition of the concentration index is proposed by Wagstaff, which is a straightforward way to decompose the measured degree of inequity into the contributions of various explanatory factors [24]. The positive value of contribution means that the variable contributes to pro-rich inequity, that is, richer individuals have a better health status than the poor, and vice versa [20]. We use the OLS linear regression to decompose the health inequity of rural poor residents, because the health status of rural poor residents is a count variable. First, a regression model should be given as Equation 2.
Where yi is the health status; xm is income; xn are need variables; xp are other variables; βm, βn and βp are coefficients; εi is the implied error term, which includes approximation errors. Then the concentration index for y can be written as Equation 3:
concentration indexes of xm, xn, and xp. The terms on the right side of Equation 3 denotes the contributions of income, need variables, other variables, and the implied error to inequity.
Variables
Outcome variable
Health is a complex concept, so there are many methods and indicators to measure the health status of residents. The European five-dimensional health scale (EQ-5D) is widely used by researchers due to its simplicity and high credibility [25-27]. Zhang divided the health of the elderly into three aspects: physical health, cognitive function, and self-evaluated health [28]. Li measured the health status of rural Chinese residents by whether they had been ill in the past two weeks and whether they suffered from chronic diseases [29]. Meanwhile, Lorraine et al. thought that Self-rated health is generally accepted as a valid measure of health status in population studies in their research[30], and Hesketh et al. used the self-rated health explored the health status and access to health care of migrant workers in China[31]. Taking the experience that using self-reported health to measure the health status of people for reference, this study thinks that Self-rated health status can accurately and directly reflect rural poor residents’ understanding of their overall health status, including Both physical health and mental health. Therefore, this study reflects the health status of rural poor residents through self-reported health questionnaires. The self-evaluation questions about health is “how do you feel about your physical health?”, and the answer includes five dimensions of very poor, poor, average, good, and very good, with a value of 1-5.
Independent variables
Since this study uses the method of centralized exponential decomposition to analyze the influencing factors of health equity, the independent variables in this study include three categories: income, need variables, and other variables. Income is measured by self-reported annual household income. Due to the process of targeted poverty alleviation, the financial sources of poor households may be diverse. In the process of the survey, the surveyor will specifically ask whether there are government subsidies and other policy incomes to ensure the accuracy of income. Need variables are closely related to the definition of health equity. In this study, need variables include gender, age, education level, and marital status of rural poor residents. Other variables mainly refer to the targeted poverty alleviation policies that may have an impact on the health and health equity of poor residents. Taking into account the impact of policies and the participation situation of poor residents, the targeted poverty alleviation policies researched in this study include industry development, relocation, employment helping, health poverty alleviation project, and basic living standard guaranteeing. The variable descriptions are shown in Table 2.
Table 2 Socio-demographic characteristics of rural poor residents and variables descriptions
variables
|
Shannxi province
(N=1233)
n(%)
|
Shanbei
(N=240)
n(%)
|
Guanzhong
(N=664)
n(%)
|
Shannan
(N=329)
n(%)
|
P
|
Health status
|
Very poor
|
170(14.98)
|
12(5.00)
|
110(17.13)
|
48(14.77)
|
0.000
|
Poor
|
357(29.58)
|
31(12.92)
|
213(33.18)
|
113(34.77)
|
Medium
|
393(32.56)
|
131(54.58)
|
164(25.55)
|
98(30.15)
|
Well
|
265(21.96)
|
60(25.00)
|
144(22.43)
|
61(18.77)
|
Very well
|
22(1.82)
|
6(2.50)
|
11(1.71)
|
5(1.54)
|
Income
|
13395.11±374.99
|
13770.86±1243.304
|
14001.7±432.80
|
11580±531.54
|
0.029
|
Gender
|
Male
|
911(74.13)
|
161(67.08)
|
528(79.88)
|
222(67.68)
|
0.000
|
Female
|
318(25.87)
|
79(32.92)
|
133(20.12)
|
106(32.32)
|
Age
|
Age<31
|
69(5.60)
|
15(6.25)
|
37(5.57)
|
17(5.17)
|
0.001
|
30<Age<61
|
806(65.37)
|
176(73.33)
|
439(66.11)
|
191(58.05)
|
Age>60
|
358(29.03)
|
49(20.42)
|
188(28.31)
|
121(36.78)
|
Degree of education
|
Unschooled
|
389(31.99)
|
155(64.58)
|
130(20.00)
|
104(31.90)
|
0.000
|
Primary school
|
439(36.10)
|
55(22.92)
|
267(41.08)
|
117(35.89)
|
Middle school and above
|
388(31.91)
|
30(12.50)
|
253(38.92)
|
105(32.21)
|
Marital status
|
Unmarried
|
150(12.44)
|
10(4.17)
|
102(15.91)
|
38(11.69)
|
0.000
|
Married and cohabiting
|
843(69.90)
|
202(84.17)
|
428(66.77)
|
213(65.54)
|
Divorced or widowed
|
213(17.66)
|
28(11.67)
|
111(17.32)
|
74(22.77)
|
Participation in industry development
|
No
|
399(36.84)
|
52(21.67)
|
250(42.52)
|
97(38.04)
|
0.000
|
Yes
|
684(63.16)
|
188(78.33)
|
338(57.48)
|
158(61.96)
|
Participation in relocation
|
No
|
841(82.21)
|
207(86.25)
|
441(79.03)
|
193(85.78)
|
0.014
|
Yes
|
182(17.79)
|
33(13.75)
|
117(20.97)
|
32(14.22)
|
participation in employment helping
|
No
|
823(81.32)
|
191(79.58)
|
459(81.67)
|
173(82.38)
|
0.712
|
Yes
|
189(18.68)
|
49(20.42)
|
103(18.33)
|
37(17.62)
|
Enjoying the health poverty alleviation project
|
No
|
292(26.43)
|
72(30.00)
|
189(32.53)
|
31(10.92)
|
0.000
|
Yes
|
813(73.57)
|
167(70.00)
|
392(67.47)
|
253(89.08)
|
Enjoying basic living standard guaranteeing
|
No
|
463(44.18)
|
108(45.00)
|
310(54.20)
|
45(19.07)
|
0.000
|
Yes
|
585(55.82)
|
132(55.00)
|
262(45.80)
|
191(80.93)
|