Effects of non-health-targeted policies on migrant health: a systematic review and meta-analysis

1WHO's Health in All Policies approach to improving population health advocates for consideration of the health implications of public policies across all sectors. This perspective is particularly relevant for the health of migrant populations, who are affected by both general and migrant-specific policies in the destination country. Migrant-specific policies include those pertaining to entry (visa and entry criteria), resettlement (dispersal policies), short-term integration (language classes), long-term integration (anti-discriminatory policies in the labour market, democratic participation, and citizenship policies), and forced and voluntary return migration (deportation procedures). More importantly, the steady increase in international migration, from approximately 155 million migrants in 2000, to 258 million in 2017, has been met with increasingly hostile migration policies worldwide, from expanded EU border control efforts to attempts to rescind legal protection granted to undocumented migrants under the US Deferred Action for Childhood Arrivals programme. Although such policies have not been designed to influence migrant health, their role as social determinants of health is incontestable. Previous systematic reviews have examined the health effects of detention policies, and documentation requirements among specific migrant populations, but have not isolated policy effects with comparator populations or examined policies at multiple stages of the migration process. The aim of this systematic review and meta-analysis was to comprehensively examine the effect of non-health-targeted policies on migrant health.

2For this systematic review and meta-analysis we searched the PubMed, Embase, and Web of Science databases in September, 2017, for peer-reviewed studies published between Jan 1, 2000, and Sept 1, 2017, and written in English, Swedish, Danish, Norwegian, Finnish, French, Spanish, or Portuguese. Search strings were constructed in collaboration with a medical librarian (appendix). Studies were eligible for inclusion if an international migrant population was studied (as per the International Organization for Migration definition); a policy intervention, not primarily introduced to improve health, was implemented at the supranational, national, or local level; a relevant comparison group was used (ie, between intervention and control groups, between countries or regions, or before and after policy implementation within a target population); and a health outcome was assessed. All quantitative and mixed-methods designs were considered. Qualitative studies were excluded as our aim was to quantify policy health effects. We did not search grey literature sources. Forward and backward citation searching (ie, identification of newer publications citing a reference included in our review and previous publications cited within an included study) and consultations with topic experts were used in parallel to identify additional articles for inclusion. Article titles and abstracts were imported into the Covidence systematic-review online management tool. Review team members (SPJ, HH, and ACD) alternated review duties for each paper. Two reviewers independently screened titles and abstracts, retrieved full texts of potentially relevant articles, and assessed article eligibility for inclusion, while discrepancies were resolved by a third reviewer.

3Evidence before this studyThe extent to which health policies are effective in influencing the health of migrants and natives alike has been investigated, yet evidence on the health effects of other public policies among migrants is scarce. To prepare for this systematic review and meta-analysis we used expert recommendations and searched the PubMed and Embase databases from Jan 1, 2000, to July 15, 2017, for English-language review articles that examined the impact of various public policies on the health of migrants. Search terms included “policy”, “migrants”, “refugees”, and “asylum seekers”. We identified three relevant systematic reviews, all published within the last decade, which found increased odds of poor mental health for asylum seekers exposed to detention and undocumented migrants exposed to restrictive (anti-immigrant) policy in the integration phase. Policies that instituted strict documentation requirements were shown to reduce migrant health-care access. However, these reviews considered few aspects of public policy outside of the health-care sector, often did not assess the quality of included studies, and did not assess health among people not exposed to such policies as a comparison group. Previous narrative reviews have highlighted the importance of considering the social determinants of health among migrants.Added value of this studyThis systematic review and meta-analysis provides a comprehensive overview of the impact of public policies on a range of migrant health outcomes. All peer-reviewed evidence came from high-income countries and primarily focused on integration policies, with scarce evidence of the effect of entry or exit policies on migrant health. Our findings highlight a mental health disadvantage in refugees with temporary residence protection and stringent reception upon arrival; decreased health-insurance enrolment and health-service use stemming from restrictive welfare policies; and poor self-rated health and psychological health among non-citizen migrants due to documentation-related and general integration-related fears.Implications of all the available evidenceMigration policies are key social determinants of health and can affect health directly, through access to care, and indirectly, through social and economic pathways. This review highlights the capacity of policy mechanisms to affect migrant health negatively and perpetuate migrant health inequities. These effects have yet to be studied in low-income and middle-income contexts and across other dimensions of migration policy, such as educational and housing opportunities and deportation and exit procedures. Based on our findings, we recommend the use of a Health in All Policies approach, which considers the health effects of all migrant-oriented policies. More inclusive approaches towards integration of migrants into their host societies, rather than the enforcement of stringent reception and border control efforts, is likely to have a positive effect on migrants' health and life opportunities. Such policies are also more in line with the spirit of international agreements made to respect and uphold universal human rights.

4Evidence before this study

5The extent to which health policies are effective in influencing the health of migrants and natives alike has been investigated, yet evidence on the health effects of other public policies among migrants is scarce. To prepare for this systematic review and meta-analysis we used expert recommendations and searched the PubMed and Embase databases from Jan 1, 2000, to July 15, 2017, for English-language review articles that examined the impact of various public policies on the health of migrants. Search terms included “policy”, “migrants”, “refugees”, and “asylum seekers”. We identified three relevant systematic reviews, all published within the last decade, which found increased odds of poor mental health for asylum seekers exposed to detention and undocumented migrants exposed to restrictive (anti-immigrant) policy in the integration phase. Policies that instituted strict documentation requirements were shown to reduce migrant health-care access. However, these reviews considered few aspects of public policy outside of the health-care sector, often did not assess the quality of included studies, and did not assess health among people not exposed to such policies as a comparison group. Previous narrative reviews have highlighted the importance of considering the social determinants of health among migrants.

6Added value of this study

7This systematic review and meta-analysis provides a comprehensive overview of the impact of public policies on a range of migrant health outcomes. All peer-reviewed evidence came from high-income countries and primarily focused on integration policies, with scarce evidence of the effect of entry or exit policies on migrant health. Our findings highlight a mental health disadvantage in refugees with temporary residence protection and stringent reception upon arrival; decreased health-insurance enrolment and health-service use stemming from restrictive welfare policies; and poor self-rated health and psychological health among non-citizen migrants due to documentation-related and general integration-related fears.

8Implications of all the available evidence

9Migration policies are key social determinants of health and can affect health directly, through access to care, and indirectly, through social and economic pathways. This review highlights the capacity of policy mechanisms to affect migrant health negatively and perpetuate migrant health inequities. These effects have yet to be studied in low-income and middle-income contexts and across other dimensions of migration policy, such as educational and housing opportunities and deportation and exit procedures. Based on our findings, we recommend the use of a Health in All Policies approach, which considers the health effects of all migrant-oriented policies. More inclusive approaches towards integration of migrants into their host societies, rather than the enforcement of stringent reception and border control efforts, is likely to have a positive effect on migrants' health and life opportunities. Such policies are also more in line with the spirit of international agreements made to respect and uphold universal human rights.

10Review team members (SPJ, HH, and ACD) extracted the data and assessed risk of bias of included papers. Extracted data included sample characteristics of exposure and control groups (ie, sample size, participants' age, gender, origin, and reason for migration), policy measures, health outcomes, control variables, details of the analytical approaches used in studies, types of outcome measures, and measures of association and statistical significance (data extraction sheet available upon request from authors). Risk of bias was assessed with a modified version of the Effective Public Health Practice Project quality assessment tool (appendix)., Authors were contacted if additional study information was required.

11Our analytical approach entailed narrative synthesis of all included studies and meta-analysis of poolable estimates. Narrative synthesis was done according to the Economic and Social Research Council guidelines. We categorised policies according to the Migration Phases Framework, focusing on the entry, integration, and exit phases. Policies were further classified by generosity—ie, whether migrants' access to health-promoting resources and opportunities was increased (generous) or limited (restrictive) by the policy. Tabulation and modified vote-counting (ie, assessing frequency and direction of effects across studies while accounting for risk of bias) was used as per the Economic and Social Research Council guidance, and was paired with subanalyses to explore the heterogeneity of health effects by migrant population and context, when possible.

12We did the random-effects meta-analysis by policy domain and health outcome when results were amenable to statistical pooling, using the metan command in Stata version 13. Conceptually similar health outcomes assessed with different measures (eg, different indicators of mental health) were meta-analysed together to quantify policy effects on broad categories of health. We grouped outcomes based on whether they were originally reported as dichotomous, which we summarised as odds ratios (ORs; 95% CI), or continuous, which we summarised as standardised mean differences (SMDs; 95% CI). Some measures were dichotomised or transformed from risk ratios to ORs to facilitate inclusion in meta-analysis. Results were displayed in Forest plots with I2 statistics to indicate the proportion of true heterogeneity between studies from the total observed variation. We used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach to summarise and assess the certainty of findings concerning the direction of the effect by outcome., Sensitivity analyses were done to test the robustness of overall findings by excluding studies with high risk of bias, and to explore heterogeneity of health effects by excluding specific migrant populations (eg, older migrants [ie, aged ≥65 years], child migrants, non-citizen migrants, and undocumented migrants).

13Our study was registered with PROSPERO, number CRD42017076104.

14The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

15The database search yielded 43 243 records. After the removal of 11 715 duplicates, 31 232 records were excluded by date, language, or publication type, or on the basis of their title or abstract. Of the remaining 296 potentially relevant full-text articles, 251 did not meet the eligibility criteria. One record was included through citation searching. A final 46 articles were included in the narrative synthesis,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, of which two reported on the same study,, and 19 qualified for meta-analysis,,,,,,,,,,,,,,,,,, (figure 1; table 1; appendix). All included studies were done in high-income contexts, and examined the health-related impact of entry (eight studies),,,,,,,, and integration (37 studies),,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, policies (table 1); no study investigated exit policies. Most studies pertaining to entry had high risk of bias, while integration-related studies generally had low and moderate risks of bias. Findings from the meta-analysis and narrative synthesis are described by policy type (entry or integration), then by health outcome (table 2).

16Entry policies were examined in eight studies primarily in relation to mental health,,,,,,, but also to birth outcomes, mortality, and health-care access., Seven studies,,,,,, were cross-sectional and two, had cohort designs. Study contexts included Australia,,,, the Netherlands,, the USA, comparative studies of Canada and Spain, and refugee sites in multiple countries. Most studies examined refugees or asylum seekers, and two, specifically studied asylum-seeking children.

17Meta-analysis indicated that restrictive entry policies were associated with poor mental health outcomes (ie, psychological distress, depression, anxiety, and post-traumatic stress disorder; SMD 0·44, 95% CI 0·13–0·75, I2=92·1%; figure 2),,,, specifically with regard to temporary (vs permanent) protection visas, restrictive (vs open) border control, and reduced (vs freer) mobility in detainment. Despite high risk of bias and heterogeneous policy contexts and analytical methods across most entry policy studies, narrative synthesis largely supported these findings (appendix).,,,,

18Restrictive entry policies were associated with decreased odds of low birthweight among migrants in one study, perhaps because of positive health selection. However, a high-risk-of-bias study of forced migration to refugee camps, which might be less likely to be affected by health selection than labour-based migration, revealed decreased crude and under-5 mortality with increased (ie, generous) spending on refugee protection and assistance programmes. Other evidence with high risk of bias showed increased or similar levels of difficulties in obtaining access to health care among refugees granted temporary versus permanent protection visas.

19The 37 identified studies pertaining to the integration stage of migration were subdivided into categories of general integration, welfare, and documentation policies (table 1). Studies of general integration policies,,,,, considered the broad approaches adopted by different European destination countries, using the Migrant Integration Policy Index (MIPEX) to categorise policy contexts (inclusive, assimilationist, and exclusionist). Studies on migrant-specific welfare policies in the USA,,,,,,,,,,,,,,,, and in Germany assessed efforts to curtail non-citizen welfare service use (including public health insurance) via waiting times,,,,,,,,,,,,,,, cost barriers,,, and burden of proof for eligibility,,, with controlled before–after and difference-in-difference designs. Finally, studies of documentation policies published in the USA,,,,,,,,,, Italy, and South Korea examined efforts to institute a burden of proof for legal residence, to be used for employment and law-enforcement purposes, while also considering migrants' experiences of fear and discrimination.

20Meta-analysis of integration studies indicated that more restrictive policies across the three subcategories were associated with increased odds of poor self-rated health (OR 1·67, 95% CI 1·35–1·98, I2=82·0%; figure 3a).,, Exclusion of evidence with high risk of bias did not affect this finding (1·68, 1·31–2·05, I2=84·8%). Meta-analysis of MIPEX-based general policy approaches found that the odds of poor self-rated health were increased among migrants in assimilationist (1·31, 1·16–1·45, I2=0·0%) and exclusionist (2·39, 1·99–2·78, I2=0·0%) contexts relative to inclusive contexts. Narrative synthesis of studies with low risk of bias supported this gradient of poor self-rated health by increasingly restrictive general policy contexts., However, examination of both welfare and documentation policies revealed mixed associations with self-rated health (appendix).,,,

21Similarly, relative to inclusive approaches to integration (as per MIPEX), migrant mortality risks were elevated by more restrictive general policy approaches (OR 1·38, 95% CI 1·10–1·65, I2=98·9%; figure 3b), specifically in exclusionist settings (2·14, 1·71–2·57, I2=89·0%), among both men (2·45, 1·74–3·17, I2=89·4%) and women (1·80, 0·78–2·81, I2=94·0%). A decreased risk of all-cause mortality was observed in assimilationist contexts (0·73, 0·58–0·89, I2=96·7%) and among women specifically (0·69, 0·52–0·85, I2=91·0%), possibly attributable to variation in health-care access and quality within these policy classifications.

22Subcategories of integration policies were found to have mixed effects on mental health. With inadequate data for meta-analysis, narrative synthesis of general integration policies revealed that all migrants had worse health than natives,, with the greatest mental health gap in exclusionist contexts, followed by assimilationist, and finally, inclusive contexts (as per the MIPEX score). Protective documentation policy was shown to safeguard undocumented migrants against poor mental health in robust and weak studies alike, while evidence with moderate and high risk of bias for the mental health effects of restrictive documentation policies was mixed.,

23Several studies also assessed the effects of welfare and documentation policy on health insurance and health-care service use. US policies restricting entitlement to welfare support showed no significant reduction in migrants' public health insurance (ie, Medicaid) enrolment (OR 0·89, 95·% CI 0·71–1·07, I2=99·4%; figure 4A) in the meta-analysis.,,,, However, post-hoc analysis that excluded non-working-age migrants showed a reduction in the odds estimate—albeit not significant—for enrolment (0·80, 0·50–1·10, I2=99·8%), suggesting a potential policy effect for some migrants.,, In line with welfare policies' contingency on citizenship, post-hoc analysis similarly found lower odds of enrolment among non-citizens than among their naturalised peers (0·87, 0·74–1·00, I2=96·3%).,, Synthesis of studies with low and moderate risk of bias revealed mixed effects of heterogeneous welfare policies on Medicaid enrolment in diverse migrant populations (appendix).,,,,,,,

24Welfare restrictions likewise did not appear to affect odds of being uninsured (ie, lacking any form of health insurance, public or private) as an unqualified migrant relative to qualified individuals (OR 1·06, 95% CI 0·90–1·21, I2=54·9%; figure 4B), as per our meta-analysis.,, Narrative synthesis of studies with low and moderate risk of bias showed increases in uninsured status for women and children,,, or no change in insurance status among women, children, and older people. This discrepancy could be attributable to age heterogeneity. Post-hoc meta-analysis excluding children and older migrants revealed that increased welfare restrictions did not lead to working-age migrants becoming uninsured (0·97, 0·80–1·13, I2=49·7%). State-level policies in the USA that guaranteed migrants' right to public insurance were mostly unsuccessful in buffering insurance loss relative to states without supplementary insurance,, with only one study with moderate risk of bias showing a state-protective effect against insurance loss after welfare reform.

25Restrictive welfare policies were also associated with reduced migrant use of health-care services (OR 0·92, 95% CI 0·85–0·98, I2=0·0%; figure 4C)., In the narrative synthesis, studies with low and greater,,, risk of bias showed that restrictive welfare and documentation policy discouraged adult,, maternal,, and older migrant service use, but not children's use.,, This was perhaps because migrant parents were more likely to seek care for their children than themselves. A study with high risk of bias on documentation policy also showed decreased service use in paediatric emergency wards, counteracted by increased acute admissions. This was in line with low-risk-of-bias evidence showing increased societal health expenditures in Germany following the postponement of asylum-seeker and refugee welfare eligibility, suggesting that restricting health-care access can delay treatment of existing health issues and ultimately force migrants to seek more costly acute care.

26However, welfare restrictions did not exert large effects on prenatal health-care service use specifically (OR 0·93, 95% CI 0·84–1·02, I2=97·6%; figure 4D)., Subgroup analyses revealed that odds of inadequate prenatal-care use decreased among protected US states with supplementary welfare packages (ie, CA and NY; 0·92, 0·83–1·01, I2=98·2%), but increased for unprotected states (ie, FL; 3·14, 1·45–4·84, I2=0·0%), from before to after the introduction of federal welfare restrictions. Evidence in the narrative synthesis was mixed with regard to the effects of this state-level welfare policy,,, and of documentation policies (appendix), alluding to individual patterns of prenatal-care use dependent on one's nativity and state residency.

27For birth outcomes, one study with low risk of bias showed that welfare restrictions were associated with increased infant mortality among Mexican immigrants in the USA compared with natives. These restrictions did not appear to affect low birthweight (moderate risk of bias),, except by state-level differences in supplementary welfare programmes. Finally, a robust Italian study on documentation policy granting legal rights to previously undocumented women indicated decreased odds of low birthweight in their children.

28Given the observational nature of the included studies, our findings are largely reported with low or very low certainty (table 2). More restrictive entry policies were shown to increase levels of poor mental health, albeit with very low certainty given the predominance of studies with high risk of bias. Restrictive policies in the integration phase were associated with decreased self-rated health, once again with very low certainty, given the studies' risks of bias and imprecision (ie, null effects for documentation policies). However, we have moderate certainty that increasingly restrictive general policies of integration increase migrant mortality—inconsistencies in these effects are probably attributable to differences in the contexts and migrant populations studied within each MIPEX integration category. Mental health was shown to worsen with more restrictive policies in the integration phase, a finding we report with low certainty given some imprecision in the estimates. Restrictive welfare policies were associated with decreased public health insurance and increased proportion of uninsured (low certainty). Inconsistencies in these findings appear to be partly attributable to age differences, with working-age migrants more likely to lose public health insurance but less likely to be uninsured overall than children and older migrants. Finally, we have very low certainty that more restrictive policies in the integration phase decreased health-care service use among migrants. Study findings differed by age group, with children's health-care service use less affected than that of adult and older migrants, and by service type, with decreases in primary care use countered by heightened use of acute care.

29This systematic review and meta-analysis examined the effect of non-health-targeted public policies on migrant health. The included studies focused exclusively on policies in high-income countries related to the entry and integration phases of the migration journey.

30We found evidence for negative effects on mental health in relation to restricted entry, including policies on temporary protection, detention, and restricted asylum reception. Studies also revealed a gradient in health inequalities (poor self-rated health and mortality) between natives and migrants by severity of restrictive policy. Together, these findings suggest that the use of restrictive entry and integration policies should be avoided. If these policies are to be used in exceptional circumstances, they should be implemented alongside strategies to mitigate mental and physical health risks, and to protect the health of refugee migrant groups as per the international responsibilities of nations and states. We also found that increased risks of poor mental health in settings with strict documentation requirements were mirrored by the protective mental health effects of generous documentation policy. This implies that policy makers should not only aim to avoid poor health outcomes by reducing the implementation of harmful policies, but actively work to improve migrant health through the maintenance of generous policy efforts. Finally, there was conflicting evidence that restricting eligibility for welfare support resulted in reduced health insurance coverage. Yet, evidence of policy effects on specific types of health-care use were more consistent, whereby migrants received increased inadequate prenatal care, had reduced perinatal outcomes, and increased the use of acute care services, probably because individuals delayed care for concerns regarding eligibility and affordability. These factors can amplify long-term health-care costs, thus providing an economic incentive for policy revisions. Alternatively, policies ensuring earlier engagement in primary and routine health-care services for migrants would be both health-promoting and cost-mitigating.

31To our knowledge, this is the first systematic review and meta-analysis to examine the effects of non-health-targeted public policies on migrant health. Although previous research has found health inequalities among migrants as a result of societal and institutional factors—many of which are affected by public policies—little has been done to isolate the effects of these policies on migrant health. Furthermore, comprehensive evidence on policies across country contexts and migrant populations has been lacking. Our review aimed to summarise research across these dimensions and revealed a strong regional dispersal of research on specific migrant policy domains. The Australian literature largely focused on entry criteria among asylum seekers,,,, reflecting ongoing international criticism of its detainment procedures. Most European studies centred on general integration efforts, particularly labour-market participation.,,,,, Research from the USA prioritised documentation status and general attitudes towards migrants,,,,,,,,,, as well as policy changes that restricted migrant welfare benefits.,,,,,,,,,,,,,,,,,

32Our review has several strengths, including its systematic and comprehensive approach to synthesising all available evidence from the scientific peer-reviewed literature. We also considered the adequacy of the counterfactuals or comparators used to estimate policy effects., However, our decision to exclude grey literature and restrict our search languages by our linguistic competencies might have excluded some relevant policy evaluations, particularly those outside of the European or North American contexts. The certainty of evidence was largely affected by our reliance on observational studies which often do not employ more advanced analytical strategies found in natural experiment designs to identify policy effects., The heterogeneity of migrant populations and policies, including whether policy introductions or withdrawals were assessed, and the heterogeneity in comparators, health outcomes, country contexts, and study designs, constrained our ability to do meta-analyses. Inadequate reporting of policy implementation and statistical results further limited potential narrative synthesis. Despite these challenges, we were able to synthesise and meta-analyse findings related to entry and integration policies. However, evidence was sparse for exit policies. Additionally, no included studies assessed return or circular migration, preventing assessment of these migration phases. Other frameworks, such as the lifecourse perspective, could also have been used to structure the review or provide alternative insights into our findings.

33The Edinburgh Declaration on Migration, Ethnicity, Race, and Health has advocated for more empirical evidence to understand and tackle inequalities. Our review reveals that non-health-targeted policies contribute to the production of health inequalities among migrants. Given the rapid nature of policy change, future research should aim to update the work summarised here to maintain an up-to-date evidence base on the health effects of public policy in migrants. Substantial study design limitations persist, with shortages of natural experiment studies that use robust analytical approaches to isolate policy effects (eg, difference-in-difference analyses with appropriate counterfactuals); of low-income and middle-income study contexts and international comparisons of such policies; and of health research on resettlement, short-term integration, and voluntary or involuntary return policies. These gaps should also be addressed in future research efforts (panel).Future research directionsResearch should evaluate the effects of non-health-targeted policies on migrant health in low-income and middle-income countries, as the existing evidence base is derived from high-income countries.More studies focusing on the health effects of resettlement (dispersal), short-term integration (language training), and involuntary return (deportation) policies are needed.The differential health impacts of policies targeted at the social determinants of health, on migrants compared with non-migrants, remain poorly investigated; these include major macroeconomic, education, employment, social security, and housing policies which might be expected to affect migrants more than native populations.Consideration of the medium-term and long-term health effects of policies, as well as whether effects differ amongst migrant subgroups, including by gender, age, socioeconomic position, and reason for migration, is needed.Studies should evaluate the health effects of potentially desirable policies oriented towards anti-discrimination, citizenship acquisition, or democratic participation; such studies could inform the implementation of compensatory government policies.Further research on entry policies is needed for non-refugee migrants in relation to visa and selection criteria to understand their long-term health effects and how such policies might impact health selection.Methodological considerationsStudies should attempt to consider how best to create a counterfactual for assessment of causal effects.Adequate reporting of policy implementation and its context is essential to ensure results can be interpreted meaningfully.Further efforts should be made to theorise the potential mechanisms at play between policies and the specific health outcomes under consideration—when possible, empirical analysis should investigate these mechanisms.Data linkage might provide new opportunities for studying the health of migrants in a robust and efficient way, but has been underused thus far.

34Future research directions

35Research should evaluate the effects of non-health-targeted policies on migrant health in low-income and middle-income countries, as the existing evidence base is derived from high-income countries.More studies focusing on the health effects of resettlement (dispersal), short-term integration (language training), and involuntary return (deportation) policies are needed.The differential health impacts of policies targeted at the social determinants of health, on migrants compared with non-migrants, remain poorly investigated; these include major macroeconomic, education, employment, social security, and housing policies which might be expected to affect migrants more than native populations.Consideration of the medium-term and long-term health effects of policies, as well as whether effects differ amongst migrant subgroups, including by gender, age, socioeconomic position, and reason for migration, is needed.Studies should evaluate the health effects of potentially desirable policies oriented towards anti-discrimination, citizenship acquisition, or democratic participation; such studies could inform the implementation of compensatory government policies.Further research on entry policies is needed for non-refugee migrants in relation to visa and selection criteria to understand their long-term health effects and how such policies might impact health selection.

36Research should evaluate the effects of non-health-targeted policies on migrant health in low-income and middle-income countries, as the existing evidence base is derived from high-income countries.

37More studies focusing on the health effects of resettlement (dispersal), short-term integration (language training), and involuntary return (deportation) policies are needed.

38The differential health impacts of policies targeted at the social determinants of health, on migrants compared with non-migrants, remain poorly investigated; these include major macroeconomic, education, employment, social security, and housing policies which might be expected to affect migrants more than native populations.

39Consideration of the medium-term and long-term health effects of policies, as well as whether effects differ amongst migrant subgroups, including by gender, age, socioeconomic position, and reason for migration, is needed.

40Studies should evaluate the health effects of potentially desirable policies oriented towards anti-discrimination, citizenship acquisition, or democratic participation; such studies could inform the implementation of compensatory government policies.

41Further research on entry policies is needed for non-refugee migrants in relation to visa and selection criteria to understand their long-term health effects and how such policies might impact health selection.

42Methodological considerations

43Studies should attempt to consider how best to create a counterfactual for assessment of causal effects.Adequate reporting of policy implementation and its context is essential to ensure results can be interpreted meaningfully.Further efforts should be made to theorise the potential mechanisms at play between policies and the specific health outcomes under consideration—when possible, empirical analysis should investigate these mechanisms.Data linkage might provide new opportunities for studying the health of migrants in a robust and efficient way, but has been underused thus far.

44Studies should attempt to consider how best to create a counterfactual for assessment of causal effects.

45Adequate reporting of policy implementation and its context is essential to ensure results can be interpreted meaningfully.

46Further efforts should be made to theorise the potential mechanisms at play between policies and the specific health outcomes under consideration—when possible, empirical analysis should investigate these mechanisms.

47Data linkage might provide new opportunities for studying the health of migrants in a robust and efficient way, but has been underused thus far.

48The findings of this systematic review and meta-analysis provide empirical evidence that non-health-specific public policies can affect migrant health, supporting the importance not only of adopting a Health in All Policies paradigm, but ultimately embracing a human-rights framework that draws attention to the rights of migrants under international law., The human-rights framework considers health as an interdependent human right, whereby action to support some rights could reinforce others—eg, action to provide secure shelter is an important component of ensuring the right to health. Restrictive migration policies (through detention, reduced access to welfare, and so on), therefore, do not only appear to cause health harms, as our findings show, but fundamentally undermine human rights more broadly. Although international law provides a supportive institutional context for improving the health of migrants, its enforcement to meet human-rights obligations is weak. Achieving healthy migration policies requires an appreciation of the conflicting policy interests, with often hostile domestic views on migration being weighed up against potential improvements to the health and human rights of migrants., While improving the evidence base is an important part of improving migrants' health, political action is also ultimately required.