Effects of rural-to-urban migration on the cognitive aging trajectories of older Chinese adults: results from a prospective cohort analysis

Increasingly, rural older Chinese adults are moving to urban to live with their children who have migrated to urban. However, few studies have examined this pattern of migration and its effects on cognitive function. Our aim was to investigate the effects of domestic rural-to-urban migration on the cognitive function trajectories of older Chinese adults, as well as the explanatory factors for these effects. The data were drawn from three waves of the China Health and Retirement Longitudinal Study (CHARLS). Migrants were defined as participants who had a rural Hukou status (the household registration system of China) but lived in an urban area throughout the study and follow-up period. Cognitive functions were measured using an adapted Chinese version of the Mini-Mental State Examination. We performed multilevel linear regression models to examine the association of internal migration and cognitive function trajectories. Scale; CHARLS: the China Health and Retirement Longitudinal Study; ELSA: the English Longitudinal Study of Aging; HRS: the Health and Retirement Study; MMSE: the Mini-Mental Status Examination; PPS: probability-proportional-to-size

Rural-to-urban migration was positively associated with cognitive function only in women, but did not affect the rate of cognitive decline in either sex. The findings indicate that interventions are warranted to increase the cognitive functions of male migrants and rural, non-migrating older adults.

Background
In China, the estimated number of individuals with cognitive impairment was more than 10.4 million in 2016 [1]. Cognitive impairment is associated with a poor quality of life among older adults, and imposes heavy burdens on families and societies [2]. Moreover, the rapid population aging in China is occurring. However, the health and social care systems are not yet prepared to meet the increasing needs of older patients who have cognitive impairment. On the other hand, the unstable "4-2-1" family structure (four grandparents, two parents, and one child) and the substantial number of working-age adults who migrate from rural to urban areas for employment has undermined traditional practices of family care [3]. Increasingly, older Chinese adults are moving to remain closer to their migrant children, and many provide care to their grandchildren [4]. This unique trend in migration is associated with significant changes in the living environments and social networks of older Chinese adults, which may have important and not fully understood effects on their health and quality of life [5,6].
The extent to which migration may affect cognitive aging is unclear [7]. Various aspects may contribute to the effects of migration on the cognition of older adults, including the socioeconomic, physical and psychological statuses, behaviour and environmental factors [8]. Older adults who migrate to cities alongside their adult children may benefit from increased family interactions, such as the receipt of support from, and provision of support to close relatives [8]. Moreover, urban environments provide more opportunities for various leisure activities, which may also contribute to the maintenance of cognitive function [9]. However, some empirical studies found that rural-to-urban elderly migrants adapted poorly to daily life [6], and that migrants were likely to lose their social networks and interactions associated with their places of origin, which may cause depression [10].
However, these changes may have sex differences [11], as women tend participate in family and social activities more than men [12]. Consequently, women may be more likely to experience cognitive benefits or burdens associated with these engagements [13][14][15].
Most previous studies in this research area have explored the relationship between international migration and cognitive function and have reported conflicting findings, while few have focused on internal migration [7]. In the USA, a retrospective cohort study of 1789 Hispanic-American participants indicated no association between migration and cognitive function [16]. However, a study of 1085 participants in Europe observed poorer cognitive function in non-European migrants than in local citizens [17]. In China, a 12-year longitudinal study indicated that rural-to-urban and rural residents had a more rapid decline in cognitive function than urban residents [18]. However, that study compared rural-to-urban migrants with native urban citizens, rather than their rural nonmigrant counterparts, and did not highlight the potential psychosocial risk factors for cognitive decline in the migrant population. A comparison of rural-to-urban migrants with their rural nonmigrant counterparts may be more appropriate and might better elucidate the changes in the environment and social network (particularly family interactions and social participation) that are consequential to rural-urban migration and its potential effects on cognition. Despite the increasing phenomenon of internal rural-to-urban elderly migration in China, evidence regarding the potential link between this event and cognitive aging remains lacking. Longitudinal studies are also needed to extend the evidence regarding the long-term effects of migration on cognitive aging trajectories.
This study aimed to investigate the sex-specific effects of internal rural-to-urban migration on cognitive aging trajectories, and to explore the potential contributing psychosocial factors in a nationally representative longitudinal sample of community-dwelling older Chinese adults. We hypothesize that rural-to-urban elderly migrants would exhibit better cognitive function and a slower rate of cognitive decline than their non-migrant rural counterparts. Furthermore, it was hypothesized that these associations would be more evident in women than in men and would be partly attributable to family interaction and social engagement.

Study Sample
This study used data from three waves of the China Health and Retirement Longitudinal Study (CHARLS 2011(CHARLS -2015, which was designed based on the Health and Retirement Study (HRS) in the USA. CHARLS comprised a nationally representative sample of adults in China aged ≥ 45 years. The CHARLS sample was obtained using four-stage stratified sampling with the probability-proportional-tosize (PPS) technique [19]. A baseline survey that covered 28

Migrants and Non-migrants
We divided our sample into two groups, migrants and non-migrants, based on the Hukou system, which served as a classifier between rural and urban residents in previous studies [20,21]. The ruralto-urban elderly migrants were defined as older adults (≥ 60 years) with a rural agriculture Hukou status who resided in urban areas during all three study waves (n = 850). Non-migrants were defined as respondents with a rural Hukou status who resided in rural areas during the three study waves (n = 3026).

Cognitive Function
Cognitive function was measured using an adapted Chinese version of the Mini-Mental Status Examination (MMSE), which included similar concepts as those used to measure cognitive function in the U.S. Health and Retirement Study (HRS) [22]. According to previous publications [23][24][25], we divided cognitive function into two dimensions: episodic memory and mental status. We generated an episodic memory score (range: 0-10) as the average of the immediate and delayed recall scores. The mental status score (range: 0-11) was based on the following three items: figure drawing, serial subtraction of 7 from 100 (up to five times) and the ability to identify the date (month, day, year), day of the week and season. The total cognition score, which incorporated both dimensions, ranged from 0 to 21. A higher score indicated better cognitive function.

Psychosocial Factors
The psychosocial factors comprised family engagement and social attachment. Family engagement was measured by assessing the living arrangement, such as whether the household was coupled, whether the respondent lived with his/her children and whether the respondent had provided any informal care to his/her grandchildren during the past year.
The measure of social attachment was adapted from the definition provided in the English Longitudinal Study of Aging (ELSA). Specifically, social attachment was divided into four domains: civic participation, leisure activities, cultural engagement and social networks. However, we excluded cultural engagement to accommodate the Chinese social background of the subjects. We defined civic participation as the performance of activities with a community-related organization or of volunteer or charity activities within one month before the interview, and participants who had one of those activities mentioned above was defined as civic participation. Participation in leisure activities was defined as playing mahjong, cards or chess; visiting a community, athletic, social or other type of club or attending an educational or training course within one month before the interview. Participation in a social network was defined as living in a coupled residence or with children, caring for grandchildren or interacting with friends within one month before the interview.

Covariates
Demographic characteristics, socioeconomic status, health and health behaviours were considered covariates in our study. Educational level was classified according to the prior CHARLS study, as well as the distribution of educational attainment among older Chinese adults, as: illiterate; some primary school (not completed); finished primary school; and higher than primary school [26,27]. The household income was defined as the sum of all annual income at the household level, and was stratified into three levels, low, medium and high, according to the percentiles of respondents in this study. The retirement status was dichotomized as retired or not retired.
The health status was assessed by the number of activities of daily living (ADL) disability, or reported having a chronic disease or depression. ADL was determined by the number of activities during which the participant experienced difficulties (range: 0-6). The chronic disease status was determined from self-reported diagnoses. Depression was measured using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10), with a higher score (range: 0-30) indicating more severe depression. According to previous studies, smoking, alcohol consumption and afternoon napping may affect cognitive function in the elderly [8,22,28]. Therefore, we considered these three items as health behaviours. Respondents were categorized as non-smokers, light/moderate smokers (< 20 cigarettes per day currently or a history of smoking) or heavy smokers (≥ 20 cigarettes per day, currently). The alcohol consumption categories were non-drinkers, less than once per month and more than once per month. Respondents were further categorized as non-nappers, short nappers (< 30 min), moderate nappers (30-90 min) or extended nappers (> 90 min) [28].

Statistical Methods
The characteristics of the sample are described according to sex and migrant status. Continuous variables are reported as means and standard deviations, while categorical variables are reported as percentages. The t-test was used to compare normally distributed continuous variables, while the chisquare and rank-sum tests were used to compare categorical and ranked variables, respectively. We examined differences in cognitive function trajectories using a multilevel linear regression with the follow-up wave as the first level (low level), which were coded as 0, 1 or 2 to represent the longitudinal term, and with individuals as the second level (high level). We assumed that individuals would have different baseline levels of cognitive function and different rates of cognitive decline. Therefore, we estimated random coefficient models. First, we detected the difference in cognitive trajectories between migrants and non-migrants and investigated the presence of a sex-specific difference by establishing a model containing the interacting terms of time, migration status and sex.
As we identified a significant interaction between sex and migration status with respect to the total cognition and mental status scores (see Additional file 1: Table S1), we stratified all analyses by sex.
Given the observed difference in cognitive function between female migrants and non-migrants, we constructed a series of adjustment models to explore the possible underlying factors. Model 1 was adjusted for age group and time of follow-up. Model 2 comprised model 1 plus socioeconomic status, while models 3 and 4 included living arrangement and social attachment, respectively. Finally, health status and health behaviours were added to model 5. We used multiple imputation methods to address missing values. We also conducted a sensitivity analysis by running models in which the missing values had not been imputed, and achieved similar results. In all analyses, statistical significance was based on a two-tailed P value < 0.05. All analyses were performed using R software, version 3.4.5.

Results
Baseline characteristics of the overall sample and according to migration status by sex The total sample comprised 3876 participants (52.2% female, 21.9% migrants), with an average baseline age of 67.5 ± 6.5 years. Of the participants, 43.7% were illiterate, nearly a third had retired, 78.1% lived in a coupled household, nearly half lived with children and 39.9% had cared for grandchildren in the past year. Few reported civic participation (0.9%), less than a fifth had participated in leisure activities during the past one month and 95.2% had a social network. Nearly three quarters of the participants had chronic diseases, and the average depression score was 9.1 ± 6.5. More than half the participants were non-smokers and non-drinkers, and nearly half did not take naps. The average total cognition score was 9.2 ± 3.8.
Although female migrants had better baseline total cognition (8.8 ± 3.8 vs 8.1 ± 3.8, P = 0.001) and mental status scores (5.8 ± 3.0 vs 5.2 ± 3.0, P < 0.001) than female non-migrants, these differences were not observed among male participants. Moreover, no significant episodic memory inter-group differences were observed. The detailed baseline characteristics are presented in Table 1.
Sex-specific differences in cognitive trajectories between migrants and nonmigrants Table 2 presents the results of the multilevel model analyses of the effect of migration on the cognitive function trajectory. After the age groups were adjusted, the time terms in all models were significantly negative, indicating that all cognitive functions declined with time.
Among women, migrants achieved better total cognition (b = 0.77, P < 0.001) and mental status scores (b = 0.68, P < 0.001) than non-migrants except episodic memory (b = 0.09, P > 0.05). Among men, however, there were no significant differences in all kinds of cognitive function levels between migrants and non-migrants. The interacting terms of migrant and time were not significant in all models, indicating that the differences in the rates of cognitive function decline between migrants and non-migrants were not significant. The sex-specific differences in total cognitive function by migrant status are illustrated in Fig. 1.  (Table 3).

Sex-specific contributions of psychosocial risk factors to the association between migration and cognitive function
With regards to the mental status of study participants, among female participants, migrants had better mental status than non-migrants, participation in leisure activities remained significantly associated with a better mental status. Caring for grandchildren was also associated with a better mental status in female participants (see Additional file 1: Table S2). For male participants, the estimate of migrant remained no significant, after controlling for the various factors included in models 1-5 in a stepwise manner. Greater participation in leisure activities was associated with a higher total cognitive level, while living with children was associated with worse cognitive function (detailed results not shown).

Discussion
We observed that rural-to-urban elderly female (but not male) migrants exhibited a better cognition level than their non-migrant counterparts. These differences were evident in the overall level of cognitive function but not the rate of cognitive decline, and were partly explained by variations in psychosocial factors. This work indicated the direction of specific interventions for internal elderly migrants, and revealed the population that require more attention.
In our study, the difference in the rate of cognitive decline between migrants and non-migrants was not significant, and this finding might be explained by the following reasons. First, given the fact the both rural-to-urban migrants and rural non-migrants had rural life experiences, differences in the rates of cognitive decline between rural-to-urban migrants and rural non-migrants were not as significant as those observed between rural and urban residents [18]. Second, the effects of rural-tourban migration on cognitive function may involve a slow and complex process that might not be detectable in the relatively short four-year follow-up period of CHARLS. According to Cattell's categorization of cognitive abilities, cognition can be divided into two dimensions, namely fluid and crystallized abilities [29]. Some researchers claimed that fluid abilities, such as memory, tend to decline linearly from early adulthood and are more difficult to improve or otherwise change in older adults [30,31]. This was consistent with our observation that the effect of migration on episodic memory was not significant.
Our results confirmed that more than half of rural-to-urban migrants lived with their adult children, suggesting that the older adults had moved to follow their adult children. And we observed that living with children was associated with a worse cognitive function performance, indicating that some elderly individuals who were disabled or could not perform self-care may have moved to the cities to receive help from their adult children. However, our results also revealed that some older adults may have migrated to provide help (e.g., voluntary care of grandchildren) to their adult children.
Although both female and male migrants were more likely to provide intergenerational care than nonmigrants, according our findings, this activity only provided cognitive benefits in terms of the mental status among female migrants. Previous studies claimed that providing voluntary care to grandchildren might have a positive effect on the elderly caregiver's cognitive function by enhancing the caregiver's senses of self-esteem and self-worth and providing a new purpose in later life [32][33][34].
However, this effect may be somewhat sex-specific and more benefit female caregiver [13,14]. Our study yielded similar results, this outcome may be partly attributable to traditional social norms in China, where women are typically expected to be responsible for domestic affairs and family life, play a nurturing role and serve as kin-keepers, whereas men are expected to fulfil the role of breadwinner.
Accordingly, grandfathers who migrate to urban areas specifically to provide care for grandchildren would deviate from this traditional social norm. Therefore, female rural-to-urban elderly migrants would more easily benefit from caring for grandchildren.
Our findings also suggested other explanations regarding the sex-specific difference in cognitive function between migrants and non-migrants. We observed that female elderly migrants performed better in terms of social attachment than female non-migrants, particularly in leisure activities. This finding was inconsistent with those of previous studies, which reported problem of poor social adaptation and integration faced by migrants [6,20,35]. This inconsistency may be related to the selection of the control group, which comprised non-migrant rural elderly in our study, however, local citizens in other studies. Compared with rural areas, urban areas feature a wealth of community activities and facilities for leisure activities [36,37], which provide migrants with more opportunities for participation. Consequently, rural-to-urban migrants had more opportunities for leisure activities.
However, increased participation in leisure activities was only observed in female migrants. This may be explained by following reasons, compared to men, women tend to have more larger and varied social networks and to exchange support with a greater number of members in their networks [38]. In contrast, men often depend solely on their spouses and may be less likely to participate in social activities in the community [39].
Our study has some limitations that should be acknowledged. First, although we used the Hukou status to classify participants into rural and urban populations, in accordance with the Chinese special background and previous research [37,40,41], detailed information about migration, such as the migration time, process and reasons, were not available. Further studies should be conducted based on different reasons for migration. Second, it was difficult to match rural-to-urban migrants with rural non-migrants who shared same residential location information in Hukou. Given the significant development gap between different regions of China, it would be more reasonable to compare populations from the same region. Third, although this analysis covered a four-year period, this might not have been sufficient to enable the development of differences in the rates of cognitive decline.
Fourth, rural-to-urban migration was associated with the level of cognitive function performance, but not the rate of cognitive decline. Although possible confounders were adjusted, reverse causation may have existed within those relationships. Further studies are needed to investigate the mechanisms underlying those associations.

Conclusions
In conclusion, our research has advanced the body of knowledge regarding the effects of rural-tourban migration on the cognitive function trajectories of the elderly. Moreover, we have revealed effects of migration may be amplified by sex-related difference, and discussed psychosocial factors behind this. This study indicated that elderly male migrants and rural non-migrants require more attention, and interventions regarding cognitive reservation of elderly internal migrants should be made based on sex-difference. In addition, our study can serve as a basis for further research exploring the mechanism underlying the relationship between migration and cognitive function in older adults.

Additional File
Additional file 1:   The sex-specific differences in total cognitive function trajectories function by migrant status. Fig. 1(A) Relationship between time (horizontal axis) and total cognitive function (range:0-21; vertical axis) of female according to migrant status. the dotted line and solid line represent migrant and stayer respectively, the dots represent the mean total cognitive function of each fellow-up point. Fig. 1(B) Association between time and total cognitive function of male according to migrant status. Over all the total cognitive function of male was higher than that of female. The cognitive performance of migrants was better than stayers, which was more pronounced in female.