Introduction

Over the past three decades, China has experienced an unprecedented increase in rural‒urban migration1, where individuals move in search of better job opportunities to improve their families’ economic situation. This migration has been accompanied by rapid urbanization, as a significant number of rural laborers flock to cities; China’s rural-to-urban migrant population has been noticed since 2000 and is expected to reach nearly 300 million before 20202. The shift of family members from rural to urban areas has impacted Chinese households, particularly rural ones, by offering paths to higher salaries and increased household income3. Over the past 40 years, this trend has prompted many Chinese residents to relocate from rural areas to urban areas. Migrant labor in China serves as an important means for many farm households to increase their income4,5, a phenomenon also observed in other countries experiencing significant internal and international labor movements6.

Labor mobility offers considerable benefits for households, yet it is critical to acknowledge the ongoing economic development gap between urban and rural areas. This gap leads to a scenario where wealthy farming families move to urban centers, often leaving behind vulnerable groups who continue to engage in agricultural activities7. In their pursuit of economic benefits, workers frequently prioritize their financial interests over health considerations, which detrimentally affects their health8. This situation emphasizes the urgent need to enhance health protection measures for workers, aiming to relieve the health and development challenges they encounter. The 19th Party Congress report highlighted the importance of providing efficient healthcare to migrant populations and ensuring equal access to fundamental public health services; this initiative aims to synchronize income growth with economic development9. As China has progressed socioeconomically, it has become imperative to focus on labor and employment, equitable income distribution, and health protection for rural residents; these strategies are vital for fostering economic growth and maintaining social stability.

Previous research by Chinese scholars has focused mainly on the impact of migrant work on the individual health of left-behind elderly individuals, the impact of migrant work on the health of rural residents, and the impact of parents’ migration on the visual health of left-behind children, with different research themes. Dasong Deng, through structural equation modelling, reported that going out for work indirectly promotes health by relieving relative income deprivation among rural residents10. Lijian Qin, using national longitudinal survey data, reported that going out for work has a certain positive effect on the health of rural residents, with differences observed between genders11. Fangli Zhou used instrumental variable methods to examine the impact of children going out for work on the health of left-behind middle-aged and elderly rural residents and reported that children going out for work have a positive effect on both the physical and mental health of left-behind middle-aged and elderly rural residents12. Kang Du’s analysis of data from the China Education Panel Survey (CEPS) revealed that parental migration reduces both the likelihood of myopia among left-behind children and the probability that myopic left-behind children are corrected13.

Empirical studies on the impact of migration on health are relatively abundant in international research. Sauharda Rai, through in-depth interviews and focus group discussions, analysed the impact of husbands going out for work on their wives’ mental health and identified five key psychosocial factors: communication, children, family support, family migration history, and societal acceptance of migration14. Chakraborty Mona examined the impact of parental migration on the nutritional health and morbidity of children aged 0–14 years in rural India and reported that children who were accompanied by their parents upon their return had a reduced prevalence of underweight15. Zúñiga María, through 26 in-depth interviews with adolescents, parents, and teachers in Yucatán, Mexico, reported that parental migration led to reduced family supervision and deteriorated mental health among adolescents, which in turn promoted alcohol and substance abuse16. Anna Baranowska’s study, which used Swedish registry data, analysed the impact of parental unemployment on children’s birth health and reported that even in high-unemployment areas, the impact of parental unemployment on birth health was relatively limited17. Hayoung Lee’s study, which was based on the eighth wave of the Survey of Health, Aging and Retirement in Europe (SHARE), analysed data from 9133 older adults in 11 Central and Eastern European countries and reported that children’s migration over 500 km was associated with an increased risk of depression among older adults. Frequent parent‒child contact could relieve this effect18.

Self-rated health (SRH) is acknowledged as a broad and sensitive gauge for health assessment, yet its specificity can be ambiguous, sometimes failing to align with objective health indicators19,20,21. SRH appears to merely provide a subjective evaluation of an individual’s health status. However, it also serves as a robust predictor of mortality and functional decline, independent of objective health measures, psychosocial factors, and demographic variables22. At present, studies on the association between going out for work and residents’ health status show complex characteristics, which are significantly influenced by differences in population characteristics, levels of socioeconomic development, and cultural backgrounds across different countries and regions. On the basis of a comprehensive review of the literature, the current research has the following shortcomings. Although existing studies cover multiple countries and some regions in China, there is a clear lack of targeted research on rural residents in the western regions of China, especially in Ningxia, resulting in regional focus bias. The important role of the key mediating variable “utilization of health services” has not received enough attention, and exploration of the mediating pathways through which going out for work affects health is lacking. The studies have not fully considered the endogeneity issue of self-selection bias, which limits the reliability of causal inference.

Compared with those of previous studies, the contributions of this research are reflected in two main aspects. First, this paper contributes to the understanding of the deeper risks associated with going out for work and health. Previous studies have examined the impact of going out for work on health from the perspective of the relationship between parental migration and offspring health but have not fully considered the impact and mechanisms of going out for work on one’s own health and health service utilization. Second, this study expands the research field of the relationship between going out for work and health from the novel perspective of the mediating role of health service utilization. Existing health studies have focused mainly on the impacts of demographic, economic, and social factors. To date, few studies have comprehensively evaluated the role of health service utilization as a mediating variable in the relationship between going out for work and self-rated health.

Theoretically, it is believed that going out for work affects the utilization of health services, which in turn impacts health. However, this hypothesis has not been thoroughly examined. Considering the increased population mobility, prominent health issues, and relatively scarce medical resources in rural Ningxia23, investigating the impact of going out for work on residents’ health and its mediating effect through the utilization of health services is highly important. Therefore, this study aims to systematically evaluate self-rated health associated with going out for work and examine the mediating role of health service utilization. Identifying and assessing relevant variable risk factors is of important practical importance for optimizing health policies and improving the health level of rural residents.

Materials and methods

Participants

The data for this study were sourced from the 2019 and 2022 surveys on the health status and utilization of health services among rural residents in Ningxia. A multistage sampling method was used to conduct surveys in four counties in Ningxia: Haiyuan, Yanchi, Pengyang, and Xiji. In the first stage, all administrative villages under each township in the four counties were classified into three levels—good, medium, and poor—on the basis of their economic development levels. In the second stage, a random number table method was used to randomly select 40% of the sample villages from each stratum. In the third stage, systematic sampling was conducted in each sample village, where one household was selected every five households, resulting in a total of 20–33 households being randomly selected. We also checked the CROSS checklist for survey studies24. The study subjects were family members who had resided in the region for at least 6 months and were aged between 18 and 60 years(questionnaire is provided in Supplementary Material). To maintain data integrity, samples with missing or invalid values for key variables were excluded. Ultimately 13,432 rural residents in 2019 and 12,626 in 2022 were included in the analysis. The specific inclusion and exclusion flowchart and sample county map are shown in the supplementary material Figure S1 and Figure S2.

Survey instrument and variable selection

Explained variable (Y): Self-rated health

In this study, self-rated health (SRH) was used as an indicator to measure the physical health of residents. SRH is a measure of an individual’s overall health perception that is widely used in public health assessments25. The questionnaire item designed based on this variable is “Compared with people of the same age, how do you feel about your current health status?” There are five response categories. In this study, the SRH variable was coded as follows: “very bad” = 1, “bad” = 2, “fair” = 3, “good” = 4, and “very good” = 5. SRH is highly valuable in public health research because it can be directly collected in large-scale surveys and has the ability to predict mortality26.

Explanatory variables (X): going out for work

In this study, the core independent variable “experience of going out for work” was designed on the basis of the following questionnaire item: “Have you worked outside your local area in the past year from the date of the survey?” The response “No” was coded as 0, and “Yes” was coded as 1.

Mediating variable (M): health service utilization

The mediating variable is the utilization of health services. According to previous studies, the utilization of outpatient and inpatient services is often used as an indicator to measure the utilization of health services27,28. In this study, combining the existing variables in the database, the explanatory variable “utilization of health services” was further divided into two aspects: the use of outpatient health services, with the questionnaire item designed on the basis of this variable being “How many times have you visited a doctor in the past 14 days?” To simplify the data analysis, the utilization of outpatient services was dichotomized into a binary variable, with “no outpatient service utilization” coded as 0 and “outpatient service utilization” coded as 1; the use of inpatient health services, with the questionnaire item designed on the basis of this variable being “How many times have you been hospitalized in the past year?” To simplify the data analysis, the utilization of inpatient services was also dichotomized into a binary variable, with “no inpatient service utilization” coded as 0 and “inpatient service utilization” coded as 1.

Control variables

Referring to existing studies29,30, this paper introduces control variables in three aspects, namely, individual characteristics, medical insurance and social assistance, and family characteristics. Individual characteristics include gender, age, marital status, and education; medical insurance and social assistance include whether medical insurance has been purchased and whether the individual is registered as a poor household; and family characteristics include family size, type of drinking water, type of toilet, and type of cooking fuel. The definitions and assignments of all the variables are shown in Tables 1and supplementary material Table S1.

Data analysis

Statistical software, including SPSS 26.0, R (version 4.4.1), and STATA 17.0 were used for data analysis. The relationship between going out for work and self-rated health among rural residents was evaluated via ordered logit regression analysis and propensity score matching methods. To ensure the robustness of the findings, multiple propensity score matching approaches and instrumental variable regression were conducted, with robustness tests conducted through substitution of the explained variable. Furthermore, mediation analysis was performed via the bootstrap method with 5000 resamples and examined at a 95% confidence interval to examine potential mediating pathways between going out for work and self-rated health. All the results were statistically significant at P < 0.05.

Propensity score matching (PSM), proposed by Rosenbaum and Rubin31, is a semiparametric estimation method widely used in observational studies to address covariate imbalance between treatment groups32. On the basis of the counterfactual inference model, this method matches individuals with and without work experience by identifying samples with similar confounding factors, thereby balancing covariates between the two groups as much as possible. PSM matches treated samples with comparable control samples on the basis of their propensity scores and analyses causal relationships between variables by calculating differences between the matched groups. This approach helps relieve selection bias, thereby yielding more accurate causal effect estimates33. This study uses a combined approach of propensity score matching (PSM) and ordered logit regression to address sample selection bias and relieve potential endogeneity issues arising from omitted variables. This methodological integration enables a more robust examination of the relationship between going out for work and self-rated health among rural residents. The analytical procedure involves first constructing matched samples through PSM, followed by ordered logit regression estimation on the balanced sample set.

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Ningxia Medical University (2021–G152). Ethical approval for this study was awarded on March 13, 2021. This approval was secured prior to our most recent follow-up survey (conducted in June 2022). Prior to the survey, the respondents were fully informed about the study’s objectives. Additionally, the questionnaire’s first page contained detailed instructions on informed consent. All the participants gave their informed consent verbally. Furthermore, the study’s methodology was rigorously aligned with the guidelines approved by the committee.

Results

Variable definitions and descriptive statistics

Table 1 Variable definitions and descriptive statistics.
Full size table

Descriptive analyses of variables

This study is based on two waves of longitudinal data from 2019 to 2022, including a total of 26,058 participants. In 2019, 13,432 individuals were included, of whom 2979 individuals had going out for work experience. In 2022, 12,626 individuals were included, of whom 2023 individuals had going out for work experience. The details are shown in Table 2.

Table 2 Distribution of demographic characteristics across different survey years.
Full size table

Ordered logit regression to analyse the relationship between going out for work and self-rated health.

Taking self-rated health as the dependent variable, three ordered logit regression models were conducted, controlling for variables at three levels: individual characteristics, medical insurance and social assistance, and family characteristics. Model 1 included the variable of going out for work and controlled for individual characteristics and year variables; Model 2 added variables related to medical insurance and social assistance on the basis of Model 1; Model 3 further incorporated variables related to family characteristics on the basis of Model 2. The results shown in Table 3 indicate that going out for work is significantly positively associated with the self-rated health of rural residents (β = 0.260, P < 0.01; β = 0.261, P < 0.01; β = 0.278, P < 0.01). Further analysis of the control variables revealed that men have better self-rated health than women do; increasing age is significantly negatively associated with self-rated health, and those who are divorced or widowed have significantly worse self-rated health than those who are unmarried. Higher education is associated with better self-rated health; larger family sizes are associated with better self-rated health; a centralized water supply is associated with better self-rated health than a decentralized water supply; and the use of clean energy for cooking is associated with better self-rated health than the use of nonclean energy. The forest plot of the ordered logit regression is provided in the supplementary material Figure S3.

Table 3 Ordered logit regression results of the association between going out for work and the self-rated health of rural residents.
Full size table

Processing the self-selection problem

To ensure the validity of the matched samples, this study conducted a balance test. The balance test aims to assess whether residents show characteristics of random assignment on the variable of going out for work under the same propensity scores. This study used nearest neighbor matching to match the propensity scores from the logit model. The results of the changes in covariates before and after matching are shown in Table 4. After matching, the standard deviations of the covariates in both groups of samples were reduced, with most covariates showing a significant decrease in standard deviation. Overall, the absolute values of the T values for most covariates decreased, and the P values increased, indicating a reduction in the statistical significance of the covariates and smaller mean differences among the covariates. This indicates that the matching effect was good, and the propensity score matching common-support test and standardized deviation plots for the covariates are presented in the supplementary material Figure S4.

Table 4 Error reduction before and after variable matching.
Full size table

To address potential issues such as selection bias in the research results, propensity score matching regression was used to conduct a robustness test on the model in Table 5. With self-rated health as the dependent variable, three propensity score matching regression models were conducted, controlling for variables at three levels: individual characteristics, medical insurance and social assistance, and family characteristics. Model 1 included the variable of going out for work and controlled for individual characteristics and year variables; Model 2 added variables related to medical insurance and social assistance on the basis of Model 1; Model 3 further incorporated variables related to family characteristics on the basis of Model 2. The results shown in Table 5 indicate that going out for work is significantly positively associated with the self-rated health of rural residents (β = 0.294, P < 0.01; β = 0.296, P < 0.01; β = 0.292, P < 0.01), and the forest plot of the propensity score matching regression is provided in the supplementary material Figure S5.

Table 5 Robustness test with propensity score matching regression methods.
Full size table

Robustness test of different propensity score matching methods

To ensure the robustness of the estimation results, several different methods are used to test whether the previously obtained estimates are robust. The main methods used are radius matching, kernel matching and local linear regression matching as alternatives to nearest neighbor matching. Table 6 shows that under different propensity score matching methods, going out for work has a significant positive effect on the self-rated health of rural residents, which is consistent with the results from the ordered logit regression model and the nearest neighbor matching regression. Therefore, the research results of this paper are relatively robust. The average treatment effects under different propensity score-matched samples are reported in the Supplementary materials Table S5.

Table 6 Robustness test of different propensity score matching regression methods.
Full size table

Robustness test with replacement of the explained variable

Considering that inaccurate recall by respondents in the survey may lead to measurement errors in self-rated health, this study introduced the “two-week prevalence rate” and “chronic disease prevalence” as alternative dependent variables to objectively assess the physical health status of rural residents, as shown in Table 7. This study used logit regression model and propensity score matching (PSM) method to analyse the relationship between going out for work and health status. The logit regression results revealed that going out for work was significantly associated with a lower two-week prevalence rate among rural residents (OR = 0.682, P < 0.01) and a lower prevalence of chronic diseases (OR = 0.654, P < 0.01). The PSM regression analysis also indicated that going out for work was significantly associated with a lower two-week prevalence rate (OR = 0.641, P < 0.01) and a lower prevalence rate of chronic diseases (OR = 0.727, P < 0.01). The robustness test results showed that regardless of the substitution of the dependent variables, the change in estimation methods, or the adjustment of the model settings, the regression coefficients of the core explanatory variables were consistent with the benchmark regression results. This further confirmed that going out for work is significantly associated with improved health among rural residents.

Table 7 Robustness test with replacement of the explained variable.
Full size table

Robustness tests: instrumental variable regression

Although this study has controlled for variables at three levels—individual characteristics, medical insurance and social assistance, and family characteristics—that may affect self-rated health in the empirical model and the propensity score matching (PSM) method can relieve biases arising from sample self-selection, there may still be endogeneity issues caused by omitted variables or reverse causality. Therefore, regression analysis was conducted with an instrumental variable model to perform an endogeneity test. To this end, we used the number of family members going out for work as an instrumental variable regression. First, the number of family members who have already worked outside often influences whether other family members choose to work outside as well. Therefore, the number of family members going out for work within the same household is strongly associated, satisfying the “relevance” requirement. Second, for residents of a household, the act of family members going out for work does not directly affect an individual’s physical health, thus satisfying the exogeneity requirement of the instrumental variable. Table 8 shows that after controlling for other variables, the instrumental variable is not significant in the estimation results, indicating that the instrumental variable does not affect self-rated health and is exogenous. The first-stage estimation results of the impact of going out for work on self-rated health show that the number of family members going out for work has a significant positive effect on going out for work, confirming the validity of the instrumental variable. The second-stage estimation results show that going out for work significantly and positively promotes self-rated health. This finding indicates that going out for work plays a role in improving self-rated health, and this conclusion still holds even after addressing the endogeneity issue.

Table 8 Robustness and endogeneity tests: instrumental variable regressions.
Full size table

Subgroup and sensitivity analyses

To further evaluate the relationship between going out for work and self-rated health among rural residents, subgroup analyses were conducted on the basis of gender, age, marital status, education, medical insurance, registered poor household, family size, type of drinking water, type of toilet, type of cooking fuel, and survey year. Table 9 shows that among rural residents in the subgroups with a marital status of divorced/widowed (β = 0.052, P > 0.05), education level of senior high school or above (β = 0.107, P > 0.05), no medical insurance (β = 0.528, P > 0.05), and flush toilet type (β = 0.088, P > 0.05), there was no statistically significant association between going out for work and self-rated health. In contrast, for all the other subgroups, going out for work was significantly positively associated with self-rated health (P < 0.05), which is consistent with the results observed in the overall population; the forest plot of the subgroup analysis is provided in the Supplement Figure S6.

Table 9 Subgroup analyses of associations between going out for work and self-rated health.
Full size table

Analysis of the mediating effect of the action path

To further examine the potential mediating pathways between going out for work and self-rated health, this study examined the mediating effects of outpatient service utilization and inpatient service utilization while controlling for variables such as gender, age, marital status, education, medical insurance, registration as a poor household, family size, type of drinking water, type of toilet, type of cooking fuel, and survey year. Table 10 shows the results of the bootstrap mediation effect test. When outpatient service utilization was used as the mediating variable, the direct effect β = 0.1226,the 95% confidence interval was (0.0994,0.1468), and P < 0.001,the proportion of the direct effect in the total effect is 96.61%; the mediating effect β = 0.0043, the 95% confidence interval was (0.0005, 0.0081), and P = 0.026,the proportion of the indirect effect in the total effect is 3.39%,the results of direct and indirect effects indicate that going out for work not only directly affects rural residents’ self-rated health but also indirectly influences self-rated health through outpatient service utilization. When the inpatient service utilization was used as the mediating variable, the direct effect β = 0.1141,the 95% confidence interval was (0.0910,0.1387), and P < 0.001,the proportion of the direct effect in the total effect is 89.91%; the mediating effect β = 0.0128, and the 95% confidence intervals were (0.0088, 0.0170), and P < 0.001, the proportion of the indirect effect in the total effect is 10.09%,the results of direct and indirect effects indicate that going out for work not only directly affects rural residents’ self-rated health but also indirectly influences self-rated health through inpatient service utilization. Furthermore, this study also tested the mediating role of total household income; detailed findings are presented in Supplementary Table S6.

Table 10 Mediating role of health service utilization on going out for work—self-rated health.
Full size table

Discussion

During the urbanization process in China, the large-scale migration of rural laborers has become an important social phenomenon34, and the migration of the labor force has reduced the income gap between urban and rural areas35. Whether the out-migration of rural labor for work affects the health status of residents and clarifies the relationship and pathways between the two are highly important for promoting high-quality integration of urban and rural areas and health equity. On the basis of health survey data of rural residents in Ningxia from 2019 to 2022, this paper conducts propensity score matching to balance health-influencing factors and address endogeneity issues. The robustness of the results is verified through various matching methods, and the reliability of the estimation results is enhanced by using the two-week prevalence rate and chronic disease prevalence rate as alternative dependent variables. The research findings indicate that going out for work has a significant positive effect on the health of rural residents in Ningxia, China’s western region. This may be because, prior to going out for work, residents lived in rural areas where the lower economic level resulted in limited access to healthcare services; after going out for work, they moved to cities with higher economic levels and more developed healthcare service provisions, which improved their access to health services. Additionally, going out for work generates income, enhancing residents’ capacity to invest in their health. One primary objective of rural‒urban living for work is to increase household income and improve family economic conditions36. Therefore, going out for work is beneficial to the personal health of rural residents.

Subgroup analysis indicated that the relationship between going out for work and self-rated health among rural residents is heterogeneous. The associations were not significant among individuals who were divorced/widowed, those with a senior high school education or above, those without medical insurance, and those using flush toilets. The insignificant positive health impact of going out for work among divorced or widowed individuals may be related to their lack of family support and the economic and caregiving pressures they bear alone. Some studies have shown that individuals who are divorced, separated, or widowed are more likely to report poor health status than those who are married or never married37,38. In rural Ningxia, where the social structure is tightly knit and traditional family values are strong, these individuals may more easily offset the health benefits of going out for work due to emotional deficits and life stressors. The insignificant positive health impact of going out for work among individuals with a senior high school education or above may be due to their more comprehensive health awareness and higher expectation standards, which lead to stricter self-assessments of health and less noticeable perceptions of health improvements. Research has shown that relatively underdeveloped education in rural areas significantly restricts the mobility of rural labor to high-salary industries with high technical barriers, such as information technology services, thereby widening the urban‒rural income gap39. The insignificant positive health impact of going out for work among those without medical insurance may be due to economic pressures that delay medical visits or lead to forgoing treatment, thereby offsetting the potential health-promoting effects of increased income. Research has shown that medical insurance coverage significantly enhances health service utilization40, and individuals who rate their health status as good are more likely to purchase private medical insurance41,. The insignificant positive health impact of going out for work among those using flush toilets may be due to better sanitary conditions and higher health status in their living environments, making it difficult to highlight the health improvement effects caused by going out for work. Research has shown that access to piped water and flush toilets impacts individuals’ physical and mental health42.

The mediation analysis revealed that the utilization of health services plays a significant mediating role in the relationship between going out for work and self-rated health among rural residents. The mediating effect of outpatient service utilization is significant, likely because working in urban areas prompts residents to shift away from the mindset of “avoidance of medical care for minor illnesses,” enhances their preventive health awareness, and increases their likelihood of seeking timely medical care, thereby influencing self-rated health. Previous studies have demonstrated that the utilization of outpatient services significantly influences residents’ self-rated health43. The mediating effect of inpatient service utilization is also significant, likely because going out for work improves economic capacity and payment ability, enabling residents to access timely inpatient service utilization when their conditions are severe, thus improving self-rated health. Scholars contend that medical insurance is strongly associated with the utilization of healthcare services and that such utilization demonstrably enhances residents’ health outcomes44,45.

Research strengths and limitations

This study has two main strengths. First, the study is based on a population-based dataset, which provides a sufficient sample size and statistical power to examine the relationship between going out for work and the health status of rural residents. Second, the propensity score matching (PSM) model is used to correct for the endogeneity issue caused by selection bias in the group of people going out for work, ensuring the robustness of the empirical results. Third, the study examines the impact mechanisms of health service utilization between going out for work and self-rated health, providing new empirical evidence to further clarify the relationship between going out for work and self-rated health.

This study also has certain limitations. The empirical test in this paper examines the net effect of going out for work on the health of rural residents in the short term. However, changes in health may require a longer process, and the adverse effects of poor working conditions on health may take a longer time. The self-rated health in this survey is self-reported, which may be subject to recall bias. Third, the sample of this study is from a single province; therefore, the main results of this study may not be generalizable to other provinces, as there are differences in economic levels and medical resources among provinces.

Conclusion

This study utilizes two waves of longitudinal data from 2019 to 2022, the “Rural Household Health Inquiry Survey”, in Ningxia to reveal a significant positive association between going out for work and self-rated health among rural residents in Ningxia. By applying various propensity score matching methods and using the two-week prevalence rate and chronic disease prevalence rate as alternative dependent variables, we further validated the robustness of these results. The subgroup analysis results indicate that among individuals who are divorced/widowed, have a senior high school education or above, lack medical insurance, or use flush toilets, the positive association between going out for work and self-rated health is not significant. Additionally, mediation analysis revealed that the utilization of outpatient and inpatient services plays a mediating role in the relationship between going out for work and self-rated health. To improve the health level of rural residents, enhancing medical insurance and accessibility to health services for those going out for work is recommended to promote health equity among rural migrant populations.

Policy implications

The imbalance in regional economic development and the long-standing dual economic structure between urban and rural areas have led to the large-scale migration of young rural laborers to cities. This study revealed that there is a significant positive association between going out for work and self-rated health among rural residents in Ningxia, with health service utilization playing a mediating role, indicating that reasonable policy guidance can further amplify this positive effect. Therefore, action in the following areas is recommended: The cross-regional medical insurance connection mechanism should be improved, medical insurance coverage for workers in their working areas should be expanded, and the accessibility of outpatient and inpatient services should be increased. Health education and health management for working-outside groups should be strengthened, especially by conducting regular health check-ups and health promotion activities at the rural level. Coordinated reforms of household registration, employment, and social security policies should be promoted to enhance the health support network for floating populations in cities. Local urbanization and industrial development should be encouraged to provide more local employment opportunities for rural residents, achieving a two-way health dividend of “going out” and “returning back.” By taking multiple measures, the health level and quality of life of rural floating populations can be effectively improved.