Health status and migration: a propensity score matching with difference-in-difference regression approach.

BACKGROUND: By 2014, about 253 million Chinese migrants had left their rural hometowns and were working in cities. The effect of migration on health is unclear, since it is affected by changes in environments and behaviours over the course of migration. Furthermore, previous studies have not thoroughly considered the "healthy migrant hypothesis" (that is, healthy people are more conducive to migrate). This paper strives to explore the potential mechanisms of migration and health over two successive years to provide a useful reference point for health policy targeted at migrants.

METHODS: Data were collected from 9576 participants aged 15-64 years from the 2012 and 2014 China Labour Force Dynamic Survey (a nationwide cross-sectional survey covering 29 provinces with a multi-stage cluster, stratified, probability sampling strategy). The variables of migration included whether to migrate (the treatment and control groups were migrants and non-migrants, respectively), duration of migration, and migration distance during 2012-14. The variables of health status were changes of self-rated general health, physical health, and emotional health between 2012 and 2014. Controlled variables included medical security, and general and occupational health behaviour during 2012-14. We used propensity score matching (PSM) and difference-in-difference (DID) methods to examine potential mechanisms by controlling for the healthy migrant hypothesis. Ethical approval was obtained from Sun Yat-sen University (Guangdong, China). Informed consent was obtained from all participants.

FINDINGS: After PSM, health status in 2012 remained the same between the matched treatment and the control groups, suggesting that the healthy migrant hypothesis was successfully controlled (that is, the health difference between the two groups can be controlled). Using DID, we found that working overtime and lacking regular health check-ups were factors that worsened the general health of migrant workers after controlling for variables of age, sex, education, working hours, signing the labour contract and buying insurance (p<0·0001); lack of occupational protection measures and lack of regular health check-ups were factors worsening the physical health of migrant workers (p<0·0001). However, we noted no direct migration effect on health.

INTERPRETATION: This study provides convincing evidence that priority should be given to improving occupational safety and medical security for migrants.

FUNDING: None.