Since the 1979 economic reform, rural workers in China have migrated to urban areas and Special Economic Zones for improved employment opportunities and living conditions. Approximately 16% of the current 1.3 billion Chinese population is recognised as the ‘floating or migrant population’ 1 . Shenzhen, the first Special Economic Zone established by the central government, currently hosts 12 million migrant workers, a staggering 86% of its total population 2 . The current ‘Hukou’ registration system in China entitles those who are registered with the local government to benefit from its social welfares, including access to labour employment and healthcare insurance schemes. Migrant workers in Shenzhen who are registered with a non-local Hukou, i.e. from their original place of residence, are not entitled to these benefits. The temporary residence card (TRC) permits migrants to receive selected welfares but many migrants still lack the TRC and are regarded as ‘unregistered’ 3 .

In the past decade, several major reforms in the healthcare financing system have been implemented in an effort to overcome rising health expenditures and disparities in health status, particularly across the urban–rural divide 4 5 6 . This included the introduction of social medical insurance schemes such as the Urban Resident-based Basic Medical Insurance (URBMI), the Urban Employee Basic Medical Insurance (UMBMI) and the New Rural Cooperative Medical Insurance Scheme (NRCMS). Although one of the main reform objectives was to achieve over 90% health insurance coverage by 2010 7 , health insurance coverage for migrants remains a challenge for the lack of access through the Hukou restrictions. Additional/ private insurance provides an alternative to social insurance, but such schemes are often offered at a premium, subject to adverse selection and have restricted coverage for severe illnesses that require hospitalisation 8 .

Population migration is associated with increased risk for communicable, non-communicable, and occupational diseases, mental health disorders, and health disparities 9 10 11 12 , as well as poorer socio-economic status and worse living conditions 13 . As migrants have less access to healthcare insurance and fewer financial resources than local registrants, high costs of healthcare may have catastrophic consequences. The lack of health insurance restricts (for financial reasons) patient’s access to care thus potentially denying them treatment at an appropriate level of care. Conversely, moral hazard deriving from private insurance coverage may lead to the inefficient allocation of resources for tertiary care services 14 .

A direct relationship between the range of services, price, and quality of care exists across the three tier healthcare structure in China (Table 1), with the majority of resources and manpower focused on tier 3 facilities and vice-versa 15 16 . Even though tier 1 facilities are more affordable, these community centres are not designed to cater for more advanced medical procedures that typically require hospitalisation. Migrants, particularly for those uninsured or with self-reported chronic health conditions, may not utilise beyond tier 1 facilities due to affordability 17 .

The objective of this study is to identify the associations between 1) health insurance coverage (both social and private insurance) by Hukou registration status (registrants with the local Shenzhen Hukou vs. migrants who lack local Hukou registration), 2) healthcare utilisation rates in general by Hukou registration and 3) point of access for healthcare services in respect to the three-tier healthcare structure and health insurance status; in Shenzhen, Guangdong, People’s Republic of China.

Among 2863 valid residential telephone numbers, 1543 (response rate 53.9%) subjects completed the survey (Figure 1). The overall weighted sample of respondents had similar socio-demographic characteristics to the Shenzhen census except for educational attainment. Migrants, in contrast to registrants, were younger, single, with lower education and income, and fewer had health insurance coverage (Table 3).

Note: C.I. denotes lower and upper limit of 95% confidence interval.

¹ Categories were constructed to be compatible with published census figures.

² Income categories are constructed by quintiles, where “low” income denotes the first quintile (<20%), “middle” income denotes between second and fourth quintile (20-60%) and “high” income denotes fifth quintile or above (>60%).

³ The proportions may not add up to 100% due to rounding.

⁴ Effect size is a statistical measure that indicates the similarity of the sample compared to the underlying population, where an effect size of 0.1 is considered “small”, 0.3 “medium” and 0.5 “large”.

Statistical significance of p-values are denoted by * (p < 0.05), ** (p <0.01) & *** (p < 0.001).

Migrants were more likely to be current smokers (OR = 1.32; CI 1.04,1.67) and less likely to self-report a chronic health condition (OR = 0.48; CI 0.34,0.67) or a hospital admission in the previous 30 days (OR = 0.51; CI 0.29,0.89). Being a migrant is associated with less likelihood to attend municipal or provincial hospitals (OR = 0.45; CI 0.32,0.61) compared with health utilization at community level (Table 4).

Note: C.I. denotes lower and upper limit of 95% confidence interval.

¹Age-comparative perceived health status was dichotomised (reference = worse).

²Smoking status and alcohol drinking pattern were dichotomised (reference = never smoker/drinker).

³Variables were dichotomised chronic health conditions (reference = absence of disease), health problems in previous 30 days (reference = none), doctor consultation in previous 30 days (reference = none), hospitalisation in previous 12 months (reference = none).

⁴Source of healthcare service utilisation was segregated into three tiers (Table 1) (reference = tier 1).

Statistical significance of p-values are denoted by * (p < 0.05), ** (p <0.01) & *** (p < 0.001).

Table 5 shows strong associations between insurance coverage and Hukou registration status. After adjustment, migrants were more likely to be uninsured (OR = 5.00; CI 3.53,7.07) or have Additional MIS only (OR = 2.99 ,CI 1.66,5.37). Migrants and those unemployed were associated with lack of Basic MIS only coverage (OR_migrant = 0.45; CI 0.35,0.58 vs. OR_unemployed = 0.67; CI 0.50,0.90). Migrants who lack Basic MIS coverage may have contributed to the strong association of purchasing Additional MIS, but the same effect was not observed for the unemployed (OR_migrant = 2.99; CI 1.66, 5.37 vs. OR_unemployed = 1.28; CI 0.72, 2.28). For those who purchased Additional MIS only, in contrast to those uninsured, were far more likely to have a hospital admission in the previous 12 months (OR_{Additional MIS only} = 2.75; 1.07-7.09 vs. OR_uninsured = 0.44; CI 0.21,0.93). The source of healthcare service utilisation is not associated with insurance status except for registrants with Basic + Additional MIS coverage who are more likely to use Municipal/ Provincial hospitals (tier 3) (OR = 1.61; CI 1.04,2.49). Self-reported chronic health conditions did not predict insurance status.

Note: Regression also adjusted for gender, age, marital status and education.

C.I. denotes lower and upper limit of 95% confidence interval.

¹ Registration status was dichotomised (reference = registered).

²Variables were dichotomised chronic health conditions (reference = absence of disease), health problems in previous 30 days (reference = none), doctor consultation in previous 30 days (reference = none), hospitalisation in previous 12 months (reference = none).

³Source of healthcare service utilisation was segregated into three tiers (Table 1) (reference = tier1).

Statistical significance of p-values are denoted by * (p < 0.05), ** (p <0.01) & *** (p < 0.001).

Table 6 shows factors predicting healthcare utilisation stratified by Hukou registration status. After adjustment, self-reported chronic health conditions was a strong predictor for utilisation in both migrants and registrants, (OR_migrants = 2.77; 1.18,6.52 vs. OR_registrants = 3.29; 1.84,5.89). Insurance status did not predict healthcare utilisation by registration status.

Note: C.I. denotes lower and upper limit of 95% confidence interval.

Statistical significance of p-values are denoted by * (p < 0.05), ** (p <0.01) & *** (p < 0.001).

This study has shown that in a city with a predominantly migrant population, inequality in the distribution of social health insurance remains a challenge. Health insurance coverage in a quasi-Bismarck insurance model like China can be used as a proxy of health protection for its population. Achieving universal health coverage for the world’s largest population is no doubt unparalleled in scale and unprecedented; some Chinese cities such as Shanghai 25 , Shantou 26 and Xiamen 27 have already reported near 100% coverage, but others such as Beijing 28 and Zhuhai 26 report only 6% and 2% rural participation respectively. Some local governments have introduced pilot medical insurance schemes- e.g. Medical Insurance System for Migrant Employees (MISM) to redress this inequity, however compliance has been poor with few migrants enrolled 10 . Under the current Hukou system, migrant workers will inevitably find it difficult to access government benefits such as healthcare. This evidence supports policy interventions or indeed the Hukou reform to improve on the transferability and portability of insurance schemes across cities in China, as found in other countries such as Canada 29 with large mobile populations. However, policymakers must also consider the larger social issues from rapid urbanization such as economic gains, societal harmony, the role of both local and central governments in providing housing, education, pension or employment and the financial impact on fiscal reserves from welfare expenses.

The outcomes of this study, lower socioeconomic status, higher rates of self-reported chronic health conditions, increased uptake of Additional MIS insurance and a possible preference for community centres or private clinics (tier 1 facilities) among migrant workers are similar to that found in the literature 30 31 32 . However the choice for community centres or private clinics may be mitigated by affordability, health care policy-guided behaviors or perhaps reflects of the effectiveness of these insurance policies themselves. While self-reported chronic health conditions were not associated with health insurance uptake, it was strongly associated with healthcare utilisation in both registrants and migrants. The Hukou system, for similar reasons explained above, is currently preventing migrant patients acquiring free services from non-registered cities, curbing potential exploits and abuse; but resulting in an unethical consequence of restricting patients to their hometowns through denying them medication elsewhere.

In our study, hospital admissions were associated with migrants who had additional MIS coverage only. Although not significant, these patients appear to be less likely to utilize tier 2 and tier 3 hospitals. This was unexpected as health service utilisation among the whole sample was driven by chronic health conditions. Still, it has been reported elsewhere that illegal private clinics may offer more affordable health care service for migrants 33 , a consequence of which may be poorer health outcomes and supply side moral hazard. From the unpublished data in the same study, more migrants relied on self-medication when reported a problem compared with registrants, and this suggests migrants may increasingly resort to self-medication, perhaps in the form of Traditional Chinese Medicine, instead of seeking health professionals for care.

Despite visions by the Chinese government to achieve universal health insurance coverage by 2012 as part of its health reform target, migrant workers appear to be neglected from its social health insurance net.

Special Economic Zones like Shenzhen, which is constituted almost entirely by migrant workers, should seek to pioneer insurance reforms from the local government level to extend health protection for its migrant population. Efforts should be focused on evaluating the eligibility of insurance schemes for both its registered and non-registered residents, improve the portability and transferability of schemes through collaboration within the Guangdong province or other major origins of migrant workers and broaden insurance reimbursements to extend access to healthcare beyond tier 1 facilities.

Providing health insurance to migrant workers is of paramount importance in maintaining a balanced and healthy workforce for the continual economical development in China.

URBMI: Urban Resident-based Basic Medical Insurance; UMBMI: Urban Employee Basic Medical Insurance; NRCMS: New Rural Cooperative Medical Insurance Scheme; MISM: Medical Insurance System for Migrant Employees.

The authors declare that they have no competing interests.

KL conducted the literature review, performed the statistical analysis and prepared the manuscript. JJ conceived the study and gave careful advice and revised the manuscript. All authors read and approved the final manuscript.