Assessment of health inequality between urban-to-urban and rural-to-urban migrant older adults in China: a cross-sectional study

Background Many studies focused on health inequality between migrant older adults and local older adults, while few study concerned the health inequalities between urban-to-urban and rural-to-urban migrant older adults. This study aimed to compare physical health and mental health between these two groups in Hangzhou, Zhejiang Province, China, and to explore the relationship between cognitive social capital, social integration and health among migrant older adults. Methods A two-stage stratified sampling method was employed to recruit participants from May to August 2013 in Hangzhou. Measurement data were compared with student’s t-tests and multivariate analysis of variance (MANOVA). Multiple linear regression was adopted in this study. Results A total of 1000 of participants who met the inclusion criteria were analyzed, consisting of 527 (52.7%) urban-to-urban and 473 (47.3%) rural-to-urban migrant older adults. There were no statistically significant difference in physical health and mental health between urban-to-urban and rural-to-urban groups on the whole. However, urban-to-urban migrant older adults had a higher reciprocity and social integration than did in rural-to-urban group (13.36 vs. 12.50, p < 0.01; 40.07 vs. 38.50, p < 0.01). And both of cognitive social capital and social integration were positively related to physical health (social reciprocity: t = 6.69, p < 0.01; social trust: t = 3.27, p < 0.01; social integration: t = 5.66, p < 0.01) and mental health (social reciprocity: t = 4.49, p < 0.01; social trust: t = 5.15, p < 0.01; social integration: t = 10.02, p < 0.01). Overall, the female, widowed, and the oldest among migrant older adults had a worse health. Conclusions Social capital and social integration were played important roles in health of migrant older adults. The female rural-to-urban migrant older adults, those aged over 70 years, and older adults who were not in marriage should be especially concerned in health policy making.


Background
The proportion of older adults (age 60 years and above) in the whole population was increasing worldwide [1], both in developed and developing countries [2]. In 2010 the Sixth National Population Census showed that people aged 60 years and above accounted for 13.6% of the total population in China [3]. In 2017 China migrant population development report announced that there were 24.5 million migrant [4]. The scale of migrant older adults, aged 60 and above who migrate between regions, increased rapidly. Migrant older adults' health has been a big issue for society as that will become an important ingredient of health for all.
With medical model transformation from a biomedical model to a bio-psycho-social model, recent evidence has been show that factors such as social capital [5], social integration [6] and social-economic factors (gender, education, marital status, immigrant status) may be important for migrant older adults' health. Among them, social capital and social integration were popular in international health research [7] and social science disciplines as well as the field of public health [8,9]. Nan Lin [10] and other researchers [11,12] defined social capital from different perspectives. Social capital refers to those features of social relationships--such as levels of interpersonal trust and norms of reciprocity and mutual aid-that facilitate collective action for mutual benefit [13]. In other words, social capital was defined as the resources available to individuals and groups through social connections and social relations with others [14]. Generally, social capital was divided into cognitive social capital and structural social capital, and horizontal social capital and vertical social capital [15]. Social integration was a broad term that refers to the degree to which an individual is connected to others and embedded in the community [16].
Previous studies suggested that poor social capital or social integration increases risk for poor health, the majority of studies focused on the association between social capital and physical health (or mental health) [17,18] and the health differences between migrant older adults and local residents, but few studies explored health inequality between urban-to-urban and rural-tourban migrant older adults, and the relationships between social integration and health [19], so little is been known about what health inequalities in these two groups. In this study, urban-to-urban migrant older adults refers to those aged 60 years and above who flowed from other cities to inflow areas (Hangzhou) above 6 months as temporary residents with non-agricultural household registration rather than permanent residents; rural-to-urban older adults refers to those aged 60 years and above who flowed from counties to inflow areas (Hangzhou) above 6 months as a temporary resident with agricultural household registration rather than permanent residents. Therefore, this study aimed to assess the differences of physical health and mental health between urban-to-urban and rural-to-urban migrant older adults, and to explore the relationship between social capital and social integration and physical health and mental health among urban-to-urban and rural-to-urban migrant older adults. Actually, social capital of migrant older adults might reduce when they left their birthplace to an unfamiliar city, which could influence their mental health and physical health. Furthermore, health advantage for rural-to-urban migrant older adults might worse than urban-to-urban migrant older adults because it was harder for those from rural area than others from urban area to adapt to an unfamiliar city life. Hypotheses are given as following. First, the rural-to-urban migrant older adults had a worse physical health than urban-tourban group. Second, the rural-to-urban migrant older adults had a worse mental health than urban-to-urban one. Third, social capital and social integration played a crucial role in physical health and mental health among urban-to-urban and rural-to-urban migrant older adults.

Study design
Hangzhou is a well-developed city in China, which had a population of 7 million in 2012, per capital and total gross domestic products of 88,985 CNY (12,940.67 USD) and 780,398 billion CNY (113,489.58 USD), respectively [20], with a migrant worker population of 2.44 million, accounting for 57.5% of the province's total population [21]. A population-based cross-sectional survey was conducted from May to August 2013 in Hangzhou, Zhejiang Province, China. Two-stage stratified sampling method was employed. First, Xihu District and Binjiang District, according to their economic status and the scale of migrant population, were selected as the study site from 13 districts in Hangzhou. Then, one sub-district was randomly selected from these two districts respectively, they were Sandun sub-district in Xihu District and Puyan sub-district in Binjiang District. Finally, two communities, Lilan and Houchengqiao in Sandun and Zhijiang and Lianzhuang in Puyan, were randomly recruited from each sub-district. All older adults met the inclusion criteria were recruited from these four communities.

Participants
Participants were consisting of urban-to-urban and rural-to-urban migrant older adults. Urban-to-urban migrant older adults refers to those aged 60 years and above who flowed from other cities to inflow areas (Hangzhou) above 6 months as temporary residents with non-agricultural household registration rather than permanent residents. Rural-to-urban older adults refers to those aged 60 years and above who flowed from counties to inflow areas (Hangzhou) above 6 months as a temporary resident with agricultural household registration rather than permanent residents.
Recruitment process was shown in Fig. 1. All participants should met the following inclusion criteria: i) being aged 60 years old and above; ii) not being a registered and permanent residence in Hangzhou; iii) having lived in Hangzhou more than 6 months; and iv) being able to read, write, and communicate in Chinese, and not having a cognitive disorder. Exclusion criteria were: i) having not finish a half of a questionnaire; ii) illogical questionnaire (a questionnaire that participants answered inconsistent on particular questions); iii) being lived in Hangzhou more than 20 years. A total of 1521 participants met these inclusion criteria and enrolled, 1316 of them completed a face-to-face interview. A final total of 1000 questionnaires were deemed valid after performing a quality check. Thus, the response rate was 86.5% and rate of valid questionnaires was 76.0%.

Ethical approval
Informed consent was obtained from participants in the form of verbal agreement and ethical approval for the study has been proved by Zhejiang University Ethical Committee (NO. ZGL201608-1).

Socio-demographic characteristics
Socio-demographic characteristic included gender (male, female), age (60 to 69 years old, 70 years old or above), marital status (in marriage, not-in-marriage), educational attainment (primary school or below, junior high school or above), mainly economic source (oneself or spouse, offspring or others), years living in local city (6 months to 1 year, one to three years, three to six years, and six to twenty years) and weight status (low weight: BMI < 18.5; normal: 24 > BMI ≥ 18.5; overweight: 28 > BMI ≥ 24; and obesity: BMI ≥ 28). Weight status was rated by body mass index (BMI, a participant's bodyweight in kilograms divided by body height in squared meters).

Cognitive social capital
"Social capital refers to a sense of community embeddedness, which is in part reflected by group membership, civic participation, and perceptions of trust, cohesion, and engagement" [22]. Cognitive social capital as a proxy of social capital has been widely accepted in the literature [23]. Social trust and reciprocity were used as a proxy for cognitive social capital [16]. The Cronbach's Alpha of social trust was 0.77. Kaiser-Meyer-Olkin (K-M-O) was 0.83 (p < 0.01). The Cronbach's Alpha of reciprocity was 0.8. Kaiser-Meyer-Olkin (K-M-O) was 0.77 (p < 0.01). The scale for measuring social capital and social trust had a good reliability and validity. All items in the Cognitive Social Capital Scale were valid. A 5-point Likert scale was used to measure the degree of agreement to each item of social trust and reciprocity. The responses were collapsed into 5-point scale: 1 = strongly disagree, 2 = disagree, 3 = neutrality, 4 = agree, and 5 = strongly agree. Higher total scores indicated higher social trust and reciprocity.

Social integration
"Social integration refers to the process in which individuals come together as a whole in a community through assimilation" [24], which can be seen as a dynamic and structured process in which all members communicate well with each other to achieve and maintain peaceful social relations. We adopted 10 items to reflect social integration of migrant older adults in this study. A 5-point Likert Scale was used to measure the degree of agreement in each item. Responses were collapsed into dichotomous outcomes: 1 = strongly disagree, 2 = disagree, 3 = neutrality, 4 = agree, and 5 = strongly agree. Higher total scores indicated higher social integration. The Cronbach's Alpha of social integration was 0.62. Kaiser-Meyer-Olkin (K-M-O) was 0.80 (p < 0.01). Unless a item v) my family members always quarrel with me for future living arrangements, the remaining items in the Social Integration Scale were valid.

Physical health and mental health
Chinese version 36-item Short Form Health Survey (SF-36) was used as a comprehensive proxy of health. Calculation of SF-36 can be acquired in published literature [25]. This scale consists of eight components, including Physical Function (PF), Role Physical (RP), Bodily Pain (BP), General Health (GH), these four components constitute Physical Health (PH); the remaining are Vitality (VT), Social Function (SF), Role Emotional (RE), and Mental Health (MH), these four components constitute Mental Health (MH). Higher SF-36 score indicates a better health.

Statistical analysis
Physical health and mental health were dependent variables, category of migrant older adults be designed as the independent variable. Gender, age, marital status, educational attainment, mainly economic source, years living in local city, weight status, cognitive social capital, and social integration were analyzed in this study as the covariates. Frequencies were calculated to describe the participants' socio-demographic characteristics. Exploratory factor analysis was adopted to measure the validity of scales. Multivariate analysis of variance (MANOVA) was used to evaluate differences in physical health and mental health between urban-to-urban and rural-tourban migrant older adults. Student's t-tests and Univariate Analysis of Variance (ANOVA) were employed to evaluate difference in physical health and mental health by gender, age category, mainly economic source, marital status, educational attainment, years living in local city, and weight status. Multiple linear regression was adopted to explore relationships between migrant older adults and health (physical health and mental health) after controlling socio-demographic characteristics. Statistical analyses were conducted using SPSS version 18.0 for Windows.

Socio-demographic characteristics
A total of 1000 migrant older adults finished the survey effectively, consisting of 527 (52.70%) urban-to-urban and 473 (47.30%) rural-to-urban migrant older adults. Overall, the number of female migrant older adults was 549 (54.90%), which was higher than the male 451 (45.10%) (p < 0.01) ( Table 1). The majority of them aged 60 to 69 years both in urban-to-urban (411, 77.99%) and rural-to-urban (413, 87.32%) groups (p < 0.01). Proportion of the married migrant older adults was close in these two groups, 84.06% in urban-to-urban and 81.61% in rural-to-urban group respectively (p = 0.3). In terms of education, 32.14% of rural-to-urban migrant older adults had an education level of junior school or above, far below 76.85% of that in urban-to-urban migrant older adults (p < 0.01). The overwhelming majority (85.58%) of urban-to-urban migrant older adults were self-supporting or dependent on their spouse, compared with only 51.59% of rural-to-urban migrant older adults (p < 0.01). More than 60% urban-to-urban and rural-tourban migrant older adults have lived in Hangzhou more than 3 years (p < 0.01). The weight status distribution between urban-to-urban and rural-to-urban groups was similar (37.76% of overweight in urban-to-urban and 30.49% in rural-to-urban migrant groups).
No difference of social trust score was obtained between urban-to-urban and rural-to-urban groups (21.38 vs. 21.43, p = 0.85), while urban-to-urban migrant older adults had higher score of reciprocity and score of social integration than that of rural-to-urban migrant older adults respectively (13.36 vs. 12.50, p < 0.01; 40.07 vs. 38.5, p < 0.01) ( Table 2).

Comparison of physical health between urban-to-urban and rural-to-urban migrant older adults
Overall, few urban-to-urban and rural-to-urban migrant older adults reported a poor health in survey (3.61% vs. 5.92%). The urban-to-urban group had a higher physical health score than did the rural-to-urban group (309.72 vs. 299.3, p = 0.03). Specifically, the average score of Role Physical (RP) and Bodily Pain (BP) in urban-to-urban group were higher than that in rural-to-urban group respectively (77.47 vs. 72.67, p = 0.05; 82.56 vs. 78.32, p < 0.01) ( Table 3).
The urban-to-urban group had a higher physical score in two age groups than that in rural-to-urban group (315.85 vs. 305.60, p = 0.05; 287.98 vs. 256.12, p = 0.02) ( Table 4). The married urban-to-urban migrant older adults had a higher physical health score than did those in rural-to-urban group (317.46 vs. 303.91, p = 0.01), while there were no difference between these two groups for those who were not in marriage. No differences were obtained between urban-to-urban and rural-to-urban groups for those with different levels of educational attainment (298.06 vs. 292.61, p = 0.50 for those with primary school and below; 313.23 vs. 313.46, p = 0.97 for those with junior high school and above). Self-supporting older adults had a similar physical health in urban-to-urban and rural-tourban group (312.01 vs. 301.65, p = 0.09), so did for those depending on offspring or others (296.13 vs. 296.79, p = 0.95). Only for urban-to-urban migrant older adults lived in Hangzhou three to 6 years had a better physical health than that in rural-to-urban group (319.07 vs. 293.26, p < 0.01), the remaining had no difference in physical health between these two groups. Older adults with normal weight in urban-to-urban group had a higher physical health score than that in rural-to-urban group (312.56 vs. 287.56, p < 0.01).
Comparison of mental health score between urban-tourban and rural-to-urban migrant older adults No difference of mental health was obtained between urban-to-urban and rural-to-urban groups (306.88 vs. 299.09, p = 0.08). Specifically, urban-to-urban group had a higher Social Function (SF) score than that in rural-tourban group (85.82 vs. 82.49, p < 0.01) ( Table 3).
Participants aged 60 to 69 years had a higher mental health score in urban-to-urban group than that in rural-to-urban group (314.73 vs. 303.83, p = 0.02), while no difference was found in those aged 70 years or above (279.49 vs. 267.42, p = 0.32) ( Table 5). The married had a higher mental health score in urbanto-urban group than did in rural-to-urban group (313.99 vs. 301.29, p = 0.01). Self-supporting older adults had a higher mental health in urban-to-urban group than did those in rural-to-urban group (307.23 vs. 291.22, p = 0.01). Participants lived in Hangzhou one to 3 years had a higher mental health score in urban-to-urban group than that in rural-to-urban group (312.19 vs. 288.95, p = 0.02). Older adults with normal weight had a higher mental health score in urban-to-urban group than that in rural-to-urban group (308.37 vs. 288.14, p < 0.01). No difference of mental health score was observed between urban-tourban and rural-to-urban groups for the male (306.82 vs. 302.61, p = 0.52), so did for the female (306.94 vs. 296.67, p = 0.09). And no difference of mental health   was detected between urban-to-urban and rural-tourban groups across educational attainment.

Influencing factors on physical health and mental health based on multiple linear regression
As Table 6 showed, urban-to-urban migrant older adults had a higher physical health score than that in rural-tourban migrant older adults (Model PH1), as well as mental health score was higher in urban-to-urban group than that in rural-to-urban group (Model MH1). However, these differences were disappeared after controlling for sociodemographic factors (Model PH2, Model MH2, Model PH3, Model MH3). For those with a higher cognitive social capital and social integration had a better physical health and mental health than those migrant older adults with a lower cognitive social capital and social integration. With increasing of score of social reciprocity, migrant older adults showed a better physical health (t = 6.69, p < 0.01) and mental health (t = 4.94, p < 0.01). Higher score of social integration indicated higher physical health and mental health (t = 5.66, p < 0.01; t = 10.02, p < 0.01). And higher score of social trust also represented a higher physical health (t = 3.27, p < 0.01) and mental health (t = 5.51, p < 0.01) as well. The young migrant older adults aged from 60 to 69 years had a better physical health and mental health than another group aged above 70 years (t = 2.05, p = 0.04; t = 2.54, p = 0.01). Migrant older adults in marriage had a higher physical health (t = 3.16, p < 0.01) and mental health (t = 3.04, p < 0.01) than those who were not-in-marriage. Compared with female migrant older adults, the male both in urban-to-urban and rural-to-urban groups had a better physical health (t = 2.49, p = 0.01). Years living in Hangzhou did not demonstrated a statistically influencing on physical health and mental health.

Discussion
The urban-to-urban and rural-to-urban migrant older adults indeed had a vital difference across age, educational attainment, and the main source of livelihood. For example, the majority of urban-to-urban migrant older adults (85.58%) were self-supporting or dependent on their spouse, while that was only 51.11% in rural-to-urban migrant older adults. A higher proportion of urban-to-urban migrant older adults had a junior school and above (76.85%), while only 32.14% in rural-to-urban migrant older adults. Briefly, comparing with rural-to-urban migrant older adults, more urban-to-urban migrant older adults were more educated and economically independent.
In general, the score of physical health was higher in urban-to-urban migrant older adults than that in ruralto-urban migrant older adults, specifically for the dimension of Role Physical (RP) and Bodily Pain (BP). While these differences were disappeared after controlling variables such as gender, age, marital status, mainly economic source, cognitive social capital, and social integration. As a matter of fact, this demonstrated that there were no significant difference in physical health and mental health between urban-to-urban and rural-to- urban groups. Seemingly, this phenomenon can be explained by the theory of "health choice" (only those older adults with a better health were more likely to flow, on the contrary, those with poor health had a lower possibility of mobility). Actually, the majority of the urban-tourban and rural-to-urban migrant older adults were drawing to city for taking care of their children or grandchildren, the majority of them were health just as they reported, which was consistent with previous finding that the majority of migrant older adults had a positive self-report health [26].
There was a positive correlation between cognitive social capital (social trust and reciprocity) and physical health, and cognitive social capital and mental health, which was consistent with a previous research in Bogota, Colombia [27,28], Italy [29], and Catalonia [30]. It is widely recognized that social integration had powerful effects on physical health and mental health in community members [31,32]. Similarly, a positive linkage was observed between social integration and physical health, and social integration and mental health in migrant older adults. Neither the differences of physical health nor the differences of mental health between urban-tourban and rural-to-urban migrant older adults were obtained. It is noticeable, however, that urban-to-urban group had a higher reciprocity and social integration than that in rural-to-urban group. That indicated, benefiting from higher cognitive social capital and social integration, that urban-to-urban migrant older adults had a better social adaptability than rural-to-urban migrant older adults. A good social adaptability could be thought as a protective factor for physical health and mental health among migrant older adults.
In addition, demographic factors such as marital status and age were vital factors for physical health and mental health among migrant older adults. Migrant older adults in marriage had better physical health and mental health than those who were not-in-marriage. Migrant older adults aged 60 to 69 years showed better physical health and mental health than others. Compared with female, male migrant older adults had a better physical health than female, while no significant difference was observed in mental health status for these two groups. That indicated, to achieve the goals of health policies for all, making health policies should focus not only on migrant older adults, but also on the most vulnerable groups, such as the female, widowed, and the oldest migrant older adults,.

Conclusions
In this study, social capital and social integration were positively related to the physical health and mental health among migrant older adults, which played important roles in fostering human health. Giving the female, widowed, and the oldest migrant older adults were more vulnerable compared with other groups among migrant older adults, more health policies should be applied for them.

Limitations
There were certain limitations in this study. First, only cognitive social capital was analyzed in this study, which might have reduced the power of the findings. Secondly, this study adopted a cross-sectional survey, which was weak to make a causal relationships between health and migrant status. However, this study compared the physical health and mental health between urban-to-urban and rural-to-urban migrant older adults, which provided a new respective for the research on migrant older adults' health and a new clue for further intervention to improve migrant population's health.