SUMMARY
Background Pandemic response and preparedness plans aim at mitigating the spread of infectious diseases and protecting public health, but migrants are often side-lined. Evidence amounted early that migrants are disproportionately affected by the COVID-19 pandemic and its consequences. However, synthesised evidence is lacking that quantifies the inequalities in infection risk and disease outcomes, or contextualises the consequences of pandemic measures and their underlying mechanisms.
Methods Systematic review searching 25 databases and grey literature (12/2019 to 11/2021). We considered empirical articles covering migrants, refugees, asylum-seekers, and internally displaced persons reporting SARS-CoV-2 cases, hospitalisation, ICU admission, mortality, COVID-19 vaccination rates or health consequences of pandemic measures. Random-effects meta-analysis of observational studies and qualitative analysis were performed for evidence synthesis. A Protocol was registered with PROSPERO (CRD42021296952).
Findings Out of 6956 studies, we included 241 in the review. For the quantitative studies (n=46), meta-analysis with over 40 million study participants showed that compared to non-migrants, migrants have an elevated risk of infection (RR = 2·33; 95%-CI: 1·88-2·89) but similar risk for hospitalisation (RR = 1·05; 0·80-1·37), while the likelihood of ICU admission was higher (RR = 1·36; 1·04-1·78). Among those hospitalised, migrants had a lower risk of mortality (RR = 0·47; 0·30-0·73), while their population-based excess mortality tended to be higher (RR = 1·31; 0·95-1·80). The qualitative synthesis (n=44) highlighted the complex interplay of social and COVID-19-related factors at different levels. This involved increased exposure, risk, and impact of pandemic measures that compromised the health of migrants.
Interpretation Even in the advanced stages of the pandemic, migrants faced higher infection risks and disproportionately suffered from the consequences of COVID-19 disease, including deaths. Population-level interventions in future health emergencies must better consider socio-economic, structural and community-level exposures to mitigate risks among migrants and enhance health information systems, to close coverage gaps in migrant groups.
Funding None.
INTRODUCTION
Global health emergencies like the COVID-19 pandemic occur unpredictably, and effective responses require well elaborated and actionable preparedness strategies in line with the International Health Regulations.1 National pandemic preparedness and response plans are part of such strategies and are supported by the WHO, e.g., through a strategic plan published on February 4, 2020, with subsequent ongoing elaboration.2, 3 However, migrants have been side-lined in such plans, prompting a call for urgent global action to consider migrants in pandemic responses from the Lancet Migration early in the pandemic (April 10,2020).4
Research in context
Evidence before this study
Reviews on inequalities in COVID-19-related health risks published in the early phase of the pandemic found that migrants were disproportionally affected by the pandemic. High infection risk and all-cause mortality, as well as severe mental burdens among migrants, were identified. However, the magnitude of inequalities in the risk of infection, hospitalisation, admission to intensive care units (ICU), mortality, or vaccination coverage between migrants and non-migrants has not been quantified. Studies conducted during the early phase of the pandemic found that poor working conditions, crowded housing, language barriers, and legal aspects are among the social determinants that intersect with migration, resulting in increased COVID-19-related risks. Evidence summaries published since then (covering studies until October 2021) focused on individual health effects of the pandemic, e.g., mental health, risk of infection, or severity of SARS-CoV-2 infections among migrants, and were limited in their geographic scope (e.g., focus on single countries, or high-income countries), the outcomes considered, or covered only single or specific migrant groups.
Added value of this study
We mapped the global empirical literature and synthesised the available qualitative and quantitative literature (published in English, German and Spanish) on multiple COVID-19-related outcomes among diverse categories of migrants, as well as the impacts of the COVID-19 pandemic on migrant populations worldwide by the end of 2021. We quantified absolute and relative inequalities among over 40 million study participants (including migrant and resident populations) using random-effects meta-analyses for several outcomes such as risk of SARS-CoV-2 infection, hospitalisation, ICU admission and mortality. Our results showed higher infection risk among migrants compared to non-migrants, and hinted at different patterns by geographical or contextual exposures. As opposed to previous (narrative) systematic reviews, we found no evidence for increased hospitalisation risk, but higher risk for ICU admission and death as far as population-level estimates are considered.
The comparability of the data was hampered by the heterogeneity of studies and poor reporting quality, while the lack of disaggregated reporting for some outcomes (e.g., vaccination coverage) made it impossible to synthesise evidence on important dimensions of inequality between migrant and resident populations. Only seven studies (2·9% of all studies) addressed vaccination coverage. Three of those studies were conducted in the European context and reported lower vaccination rates among migrants (i.e., foreign-born individuals) compared with the non-foreign-born population. The opposite was found in a study from China. Two modelling studies recommended including migrants in vaccination strategies to prevent cases and increase cost-effectiveness, while one study focused on the effectiveness of a COVID-19 vaccine (rather than vaccination coverage). Findings from the qualitative synthesis uncover not only the ways in which interrelated social risks and inequalities (lack of social protection, working conditions, housing, legal uncertainties) engender severe and unique impacts on migrants; but they also pinpoint potential sources of resilience at individual, community and societal level (e.g., mindset, financial resources, mutual support, access to health information, trust in authorities, state assistance).
Implications of all available evidence
Not only in the pandemic’s early phases, but even by the end of 2021, migrants were at higher risk of SARS-CoV- 2 infection. Once infected, migrants seemed to have more severe courses of disease requiring admission to ICU. Deaths among clinical populations were lower (likely due to different age-structures), but population-based excess mortality was higher in migrants. The venue for reducing such inequalities appears to be through population-level rather than clinical interventions. Socio-economic structures, risks in communities and specific contexts (worksites, accommodation centres) appeared as major drivers for higher infection risk and are likely to be part of the underlying causes of higher mortality. To address the challenges posed by the impact of COVID-19 on migrant health, and to better prepare for future health emergencies, it is urgently required to improve health information systems and ensure the inclusion of migrant populations in national pandemic response plans. Social and health equity policies and measures are needed to avoid future pandemic measures from unfolding (unintended) negative consequences for migrants.
Evidence from the early phase of the pandemic suggests that migrants were disproportionally affected by the COVID-19 pandemic. Reviews of the early literature found an increased risk of SARS-CoV-2 infection in migrant populations and elevated all-cause mortality compared to non-migrants.5, 6 Migrants living in crowded housing conditions, but also undocumented migrants and migrant healthcare workers, were found to be at higher infection risk.5, 6 A study examining outbreaks in German accommodation centres for asylum seekers and refugees found a significantly higher attack rate if indiscriminate mass-quarantine was applied to all camp inhabitants compared with targeted contact tracing.7 Several studies identified risk factors such as precarious working conditions, crowded housing, language barriers, or legal restrictions among migrants and negative effects of pandemic control measures on mental health.5, 6, 8–10
Despite these early efforts to compile evidence on differential risks and exposures between migrant and resident populations, there is a lack of synthesised evidence quantifying the magnitude of inequalities in infection risk, consequences of disease, or vaccination rates. Furthermore, there is still a dearth of consolidated knowledge on the impact of pandemic response strategies on migrant health beyond studies published at the onset of the pandemic.
We conducted a systematic review covering the literature from 12/2019 to 11/2021 to map the global landscape of the empirical literature and synthesise the evidence in this field. We investigated the risk of SARS-CoV-2 infections and the consequences of disease (measured by hospitalisation, intensive care unit (ICU) admission, and mortality rates) among asylum seekers, refugees, migrants and internally displaced persons (IDP) compared to non-migrants; vaccination coverage among migrants and non-migrants; and the impact of lockdown and pandemic control measures on migrant health.
METHODOLOGY
Search strategy and selection criteria
We conducted a systematic literature review in line with the Preferred Reporting for Systematic Reviews and Meta-Analyses (PRISMA) guideline. Studies were eligible for inclusion if migrants, IDP, refugees or asylum seekers were studied (following definitions of the International Organisation for Migration (IOM)11 and the United Nations High Commissioner (UNHCR)12); health effects of SARS-CoV-2 (cases, hospitalisation, ICU admission, mortality, vaccination) or corresponding policy measures among refugees and migrants were assessed; and if written in English, German or Spanish. We performed a search of the Cochrane Library and the WHO COVID-19 Research Database (representing a comprehensive source of 24 bibliographic databases), thereby updating and complementing the search of a previously published review covering all studies from 12/2019 to 11/2021.6 All study designs were considered except case series, theoretical research papers, or policy analyses without empirical data. We included peer-reviewed articles, but also preprints and official reports from the IOM and European Public Health Association websites, as well as studies from previous reviews known to the authors. For systematic reviews, only the primary studies were considered. The review protocol was registered with PROSPERO (CRD42021296952).
Search terms were developed, refined and validated among the review team, and databases searched by experienced reviewers (SR, MH) of the Rapid Response Review Unit (RRRUN).13 Website searches were performed by three reviewers (JL, SA, JO). Identified records were uploaded to a management tool for systematic reviews (Covidence), where two reviewers each screened title and abstracts independently. The same applied for full-texts. Discrepancies were resolved by a third reviewer (KB) and decisions were consented in constant discussion among the review team.
The volume of included articles required adjustments to the protocol regarding data extraction and quality assessment. Data extraction was done individually and cross-checked by MH. The following items were extracted: Generic bibliographic information (author, title, year published, journal); study objectives, hypothesis and research questions; study characteristics (research method, sample size, geography); population and context characteristics; findings (e.g., main outcomes in quantitative studies, major themes/minor themes in qualitative studies) and conclusions as reported.
The Joanna Briggs Institute’s critical appraisal tools were used for the quality assessment, which was performed by pairs of two reviewers independently.14 A quality score (range 0-100%, from lowest to highest quality) was constructed based on the answers of applied checklists (questions indicated as “not applicable” were not counted into the overall quality score in order to avoid artificial downrating of studies). For each study, the scores obtained from two independent ratings were averaged, and studies were grouped based on their scores and classified as high (100-75% of possible score), moderate (74-50%), and low (<50%) quality studies (Supplementary File, Table S5a-f). Rating discrepancies in case of considerably different scores were resolved by discussion among the team. The quality of modelling studies was assessed using an adapted instrument derived from existing tools as used in previous reviews.6, 15–18
Evidence Synthesis
We descriptively synthesised and mapped all studies (Panel 3), but applied different strategies and criteria for the analytical synthesis of qualitative and quantitative data. For the quantitative synthesis, we included only moderate- or high-quality studies that reported SARS-CoV-2 cases, hospitalisations, ICU admissions, and mortality among migrant and non-migrant populations, in order to draw reliable conclusions from the literature. We checked 169 quantitative studies with medium- or high-quality for their eligibility; studies reporting outcomes among migrants without comparison groups were excluded from the quantitative synthesis.
We performed meta-analyses for the binary outcomes (SARS-CoV-2 cases, hospitalisation, ICU admission and mortality) using the effect sizes relative risk (RR) and risk difference (RD), which allow estimation and interpretation of pooled proportional and absolute risk, respectively, associated with migratory status compared to non-migrants. Given that the denominators for mortality differed (i.e., deaths based on hospitalised cases or all deaths) we performed separate analyses. As we assume that studies are likely to have functional differences and that the true effect size θ varies between studies, we fitted random-effects models using inverse variance weighting.19 We carefully assessed the heterogeneity between studies, used sensitivity analyses to check for robustness of models, and performed explorative subgroup analyses (Panel 2).
Measure of heterogeneity, sensitivity and subgroup analyses
The measure of between-study heterogeneity τ2 estimates the variations in the true effect size θ. It was calculated using the restricted maximum likelihood (REML) method and the Q-Profile method to compute corresponding 95% confidence intervals (95%-CI).19, 20 The measure I2 was not suitable for assessing heterogeneity, as it tends toward 100% when studies have large sample sizes, which is the case for many of the included studies. The measure τ2, however, is insensitive to the precision of the included studies. Based on τ2, we calculated a prediction interval (PI) that allows to quantify the range into which a future study might fall based on the evidence considered. The PI provides a meaningful interpretation of τ2.18 We used the Kenward-Roger method to calculate the 95%-CI for both the pooled estimate θ̂ and PI.21 In addition, we performed subgroup analyses for study regions and migration indicators where appropriate, to identify possible explanations for heterogeneity. We used sensitivity analysis to carefully check the robustness of the models for each outcome and effect size and, as a result, excluded some studies from the meta-analysis. Therefore, we calculated influence diagnostics to detect extremely influential studies that have a high impact on θ̂ and τ2. We excluded studies based on extreme or implausible values of standardised residuals, Cook’s distance, hat value, and leave-one-out cross-validation (LOO-CV) to measure direct impact on τ2and θ̂.22 See supplementary file pp. 102-117 for more information on the sensitivity analysis. Analysis and visualisations were performed using R 4.2.0 statistical programming language. Meta-analyses and sensitivity analyses were performed using the meta and dmetar packages.22, 23
Overview of identified global empirical literature on SARS-CoV-2 health outcomes and effects of pandemic control measures (e.g., lockdown) among migrants (12/2019 - 11/2021), N = 241 studies
Study designs
Out of 241 included studies, 186 had a quantitative design (e.g., cross-sectional, longitudinal, cohort), including 18 modelling studies, 42 were qualitative research, and 13 studies had a mixed method design.
Geographic scope
Most studies were conducted in high-income countries (n=167), 35 studies took place in (upper) middle-income countries, 35 in lower middle-income countries, and only five studies were conducted in low-income countries. Some studies didn’t have a specific country context (e.g. modelling studies) or looked at multiple countries.
Migrant groups
The migrant groups studied varied widely. Study populations consisted mainly of a) international migrants (n=128) as defined by (a combination of) various indicators such as region of origin, country of birth, parents’ country of birth, ethnicity, nationality, language-proficiency, or migrants/immigrants without further specification; b) international migrants with a specific legal status that we know of, including international students (n=10), refugees (n=36), asylum seekers (n=14), labour migrants (n=40), undocumented migrants (n=9), ICE detainees (n=3); c) IDP (n=5); d) internal migrants (mainly in India and China) (n=14); e) returnees (n=8); f) ecological / net migration flows (n=19).
Health outcomes
50 studies investigated infection risk, 28 transmission risk, 84 health outcomes of disease (i.e., SARS-CoV-2 cases, hospitalisation, ICU admission, mortality, or pneumonia), 16 health services access, 49 mental health, 69 looked at the effects of pandemic control measures (e.g., lockdown, mask wearing, etc.), and only seven studies addressed vaccination.
Study quality
Less than half of the studies (n=112, 46·5%) were classified as high quality, 92 (38·2%) as medium quality, and 37 (15·3%) as low quality.
See supplementary file Table S6, p.35ff. for study characteristics of all included studies.
For qualitative synthesis, a coding scheme following the model of Diderichsen et al. was developed and the extracted data (i.e., main findings as reported by the included qualitative studies) was coded accordingly, using ATLAS.ti software.20 The development of the coding scheme involved both inductive and deductive coding strategies. During the first round, codes were developed inductively from the data. They were then compared with and integrated in the existing model, which describes COVID-19-related inequalities for migrants in three categories: exposure, risk, and impact. Through iterative meetings (first within the qualitative analysis-team, then the entire review-team), we elaborated the original model as follows: 1) through inductive coding, we added further codes to the categories; 2) the original model was linear (from exposure to risk to impact), while our data showed many interrelations and feedback loops within and among the categories; 3) within each category, we distinguished between factors on the micro-, meso- and macro-level; 4) we added the category “sources of resilience”, which interacts with the other three categories.
Role of the funding source
There was no funding source for this study.
RESULTS
The search yielded 7045 records from databases and 75 from websites. After removal of 164 duplicates, 6956 records were screened based on their title and abstract; 508 reports were sought for retrieval and 480 assessed for their eligibility whereof 255 did not meet the criteria (Fig. 1). Another 16 records were included based on previous reviews (n=8) or snowball sampling (i.e., primary studies from reviews) (n=8). A total of 241 studies were included to map the global empirical landscape (Panel 3). Of these, 44 qualified for qualitative synthesis (Table 1A) and 46 for meta-analysis (Table 1B).
Based on the sensitivity analysis, few studies were not eligible for the quantitative synthesis and therefore excluded from the meta-analysis (Supplementary File, Chapter 6.1; pp.103-106). Among the 26 studies reporting on SARS- CoV-2 cases,21–46 20 were eligible for quantitative synthesis.21, 22, 24, 25, 27, 29–34, 36, 37, 39–42, 44–46
Sample sizes comprised 40,019,567 individuals for cases, 1,310,219 for hospitalisation, 96,691 for ICU admission, and over five million for mortality. The risk of infection among migrants was 2·33 (95%-CI: 1·88-2·89) times the risk of non-migrants, with a RD of 7% (95%-CI: 0·02-0·13) (Fig. 2a). Inequalities in infection risk between migrants and non-migrants living in North America (RR = 2·65; 95%-CI: 1·34-5·26) or northern Europe (RR= 2·81; 95%-CI: 1·99-3·95) seemed to be higher than inequalities observed in southern Europe (RR = 1·70; 95%- CI: 1·23-2·35), whereby the actual RD compared to non-migrants was 9% (95%-CI: 0·01-0·17) in North America and 3-4% (95%-CI: 0·00-0·05; 0·01-0·08) in European countries (Fig. 2b).
Legend: Events: SARS-CoV-2 cases; Population: denominators; CI: Confidence Interval; I2 and Tau2: Measures of heterogeneity. Labberton et al. grouped migrants into two subgroups (foreign-born/foreign-born parents), hence the study appears twice in the analysis.34 The same applies to Warszawski et al. who reported cases for May and November 2020.46
Legend: Events: SARS-CoV-2 cases; Population: denominators; CI: Confidence Interval; I2 and Tau2: Measures of heterogeneity. Labberton et al. grouped migrants into two subgroups (foreign-born/foreign-born parents), hence the study appears twice in the analysis.34 The same applies to Warszawski et al. who reported cases for May and November 2020.46
International migrants turned out to be the main migrant category studied (i.e., 76% of studies), and were mostly defined by indicators such as region of origin, country of birth, language, or nationality. We grouped studies based on the underlying indicators of migratory status and provide an explorative subgroup analysis in the supplement (Figure S16, p.118). Among these groups, inequalities in infection risk compared to non-migrants were highest in migrants living in any kind of shared accommodation (refugee camp, dormitory, detention facility) (RR = 3·91; 95%-CI: 1·71-8·97), albeit with wide PI indicating high between study variations.21, 33, 40 Infection risk in migrants whose migratory status was defined by residence status, nationality of parents, country of birth, or region of origin was more than twice the risk of non-migrants (with narrow PIs), but inequalities were less pronounced among foreign workers and when migration status was defined via language (Supplementary File, Figure S16, p.118).
13 studies reported on hospitalisation of migrant and non-migrant SARS-CoV-2 infected individuals,27, 28, 32, 34, 39, 47–54 whereof nine studies were eligible for quantitative synthesis.27, 28, 32, 34, 39, 47, 48, 51, 54 Overall, the risk of hospitalisation appeared to be similar among both population groups (RR = 1·05; 95%-CI: 0·80-1·37) (Fig. 3). However, studies from southern Europe showed a lower risk in migrants compared to non-migrants,51, 54 whereas studies conducted in northern Europe reported slightly elevated risk among migrants (Fig. 3).27, 28, 32, 34, 39, 48
Legend: Events: hospitalised cases; Population: denominators (SARS-CoV-2 cases); CI: Confidence Interval; I2 and Tau2: Measures of heterogeneity.
The meta-analysis of the outcome ICU admission included eight out of ten studies and showed 1·36 (95%-CI: 1·04-1·78) times the risk in migrants compared to non-migrants (Fig. 4). 27, 40, 47, 51, 54–59 The risk difference among the groups was relatively low (RD = 0·04; 95%-CI: -0·00-0·07) (Fig. 4).
Legend: Events: ICU admissions; Population: denominators (hospitalised cases); CI: Confidence Interval; I2 and Tau2: Measures of heterogeneity.
As for mortality, 16 out of 18 studies were eligible for meta-analysis (Supplementary File, pp.112- 117).27, 28, 39, 40, 47, 51, 53, 56–66 Studies with mortality as outcome used different denominators: hospitalised cases vs. all- deaths within the respective time-period and geographic region. This resulted in different trends and patterns of inequality. With hospitalised cases as denominator for incident deaths, the mortality risk in migrants was almost half that of non-migrants (RR = 0·47; 95%-CI: 0·30-0·73) with a risk difference of -7% (95%-CI: -0·12-(-0·02)) (Fig. 5) based on a sample size of 22,561 study participants. If studies reported excess mortality (i.e. using all-deaths in the population), the risk for fatal outcomes due to or associated with SARS-CoV-2 was 1·31 (95%-CI: 0·95-1·80) the risk in migrants compared to non-migrants, however the difference between the groups was comparably low (RD = 0·03; 95%-CI: -0·02-0·09) (Fig. 5). For this analysis 5,062,317 study participants were included.
Legend: Events: SARS-CoV-2-related deaths; Population: denominators (hospitalised cases A) and B); all deaths C) and D)); CI: Confidence Interval; I2 and Tau2: Measures of heterogeneity.
The qualitative synthesis was based on 44 high- to moderate-quality articles and provided insights into the syndemic nature of the COVID-19 pandemic by showing the complex interactions between social- and COVID- 19-related factors that have resulted in relative disadvantages for migrants.43, 67–109 Figure 6 illustrates exposures, risks, and impact of COVID-19 measures as well as sources of resilience for migrant populations, each at micro-, meso-, macro-levels (i.e., at the level of the individual, the family/community, and state/society, including policies and institutions such as the healthcare system or the labour market). Factors in the different categories and on the different levels interacted and compromised the physical and psychosocial health of migrants in severe and sometimes unique ways.
Our synthesis shows that migrants faced exposures at meso- and macro-level; among these crowded housing conditions, and lack of work security, and/or governmental support emerged as particularly critical.43, 67, 77, 103, 108 Some studies further pointed to the role of social norms in creating exposures, describing e.g. that some (more communitarian) communities may tend to put pressure on their members to participate in gatherings and events. Risks (e.g., higher risk of infection) existed at all three levels. At the individual level, the included studies identified unaffordability of personal protective equipment, low health literacy and language barriers.67, 74, 79 These risk factors can be cross-referenced to the macro-level, as other studies reported late, insufficient or no provision of health information in ways that accommodate linguistic diversity formal barriers to health services and distrust in institutions, including public health services.67, 79 Risks at the meso-level include the (threat of) penalisation of preventive and response measures by employers and camp managers, which may deter migrants, e.g., from demanding protective equipment or testing.
Among the impacts of the COVID-19 pandemic and pandemic measures, three interrelated individual-level factors figured prominently across the literature: economic consequences of lockdown measures (such as job loss), impacts on migration prospects, and trajectories (mainly related to disruption of migration related services such as visas and work permit renewal or the processing of asylum applications), which put migrants in states of legal and social insecurity, and, as a corollary, uncertainties and fears of failing one’s migration project.68, 71, 73, 75, 77, 80, 83, 88, 90, 95, 99 These factors are intertwined with meso-level factors: e.g., the inability to fulfil social roles (such as acting as the family’s breadwinner, sending remittances), impacted social and family dynamics, including gender and intergenerational roles).68, 75, 80, 85, 89, 90, 102 Further, impacts on the macro-level include increased discrimination and, in some settings, the effects of differential pandemic measures. For instance, migrants in camp-like settings (e.g. asylum-seekers, refugees, labour migrants in congregate housing) as well as in-house domestic workers were often subjected to severe mobility restrictions; also in many camp-like settings, collective response measures such as mass quarantine or labour quarantine were imposed on the residents. Our analysis thus highlights the severe, to some extent unique, and in part unintended consequences of pandemic control measures for migrants.69, 71, 72, 80, 81, 90, 94, 104 More than the pandemic itself, these unintended consequences of pandemic measures have contributed significantly to the severe psychosocial impact on migrants.69, 84
Our analysis pinpoints sources of individual, community and systemic resilience that can counteract some of the above described exposures, risks and impacts of pandemic measures. Among the main individual-level sources of resilience are optimistic framings of the crisis (e.g., “We’ve been through worse.”) and the exchange with family and peers via internet and social media. Mutual (material and/ or psychosocial) support as well as decent work conditions that allow for a sense of control and participation are important community-level sources of resilience. Macro-level sources of resilience include access to governmental assistance, the tailored provision of health information, trust in the authorities, and legal status, which is key to being able to assert one’s rights. Effective intersectoral exchange and collaboration enhance health system resilience. However, our analysis also shows that many of these sources of resilience have been inaccessible or unavailable for migrants during the COVID-19 pandemic, especially those on meso- and macro-levels.43, 77
Eventually, our analysis highlights the multifaceted interrelations among the various factors in the different categories and on the different levels, with some key factors triggering cascading effects and feedback loops (Panel 4).
Cascading effects and feedback loops for different migrant groups deriving from pandemic control measures
Vignette a) “Migrants with precarious legal status”
The pandemic-related closure of offices and services obstructed the renewal of residence and work permits for migrants with precarious legal status. Informal labour markets were severely impacted by pandemic measures, leading to widespread loss of jobs and livelihoods, with no social safety nets in place. The resulting legal and economic insecurities amplified existing power differentials, put migrants at risk of exploitation, and generated major psychosocial distress due to the uncertain future prospects in the host country. The pressure to send money to families and communities in their home countries exacerbated such distress. The coping strategies of migrants and their families sometimes involved extremely hazardous employment, including child labour and survival prostitution. This, in turn, increased the exposure to and risk of a SARS-CoV-2 infection, but also other health issues such as injuries, sexually transmittable diseases, violence, unwanted pregnancies, and ill mental health. Adverse social consequences include descent into extreme poverty, food insecurity, impacts on children’s education, stigma, and tensions within families over changing gender and intergenerational roles. Access to social and health services, including tailored services such as walk-in clinics for uninsured patients, were also compromised due to the pandemic. The only available source of material and psychosocial support were often informal networks based on kinship, diaspora and/or religious communities.
(Synthesis based on Babuc 2021, Da Mosto et al. 2021, Knights et al. 2021, Filippi & Giliberti 2021, Im & George 2021, Martuscelli 2021, Sabar et al. 2021, Sanò & Della Puppa 2021, Sohel et al. 2021, Thomas et al. 2021, Yee et al. 2021, Zambrano-Barragan et al. 2021)
Vignette b): “Migrant domestic care workers”
Domestic care workers faced particularly strict pandemic measures and related changes in their working conditions: With their work permit and visa often bound to a specific client/employer, many had to accept severe mobility restrictions, namely cohabitation with their client in combination with strict curfews, and increased workloads (e.g. more cleaning and washing), alongside obligatory quarantine and testing measures, in order to keep their job. Cohabitation with the client meant a loss of personal freedom and privacy, longer working hours, constant availability, less rest and breaks, and more stress. Non-compliance with health and safety protection such as a lack of protective equipment or of paid sick leave was frequently noted. This increased health risks for domestic workers, including burnout and other mental health problems. Domestic workers were very concerned about being terminated due to a SARS-CoV-2 infection or for other health reasons, and in some studies they describe their work situation during the pandemic as a “prison” or as being trapped in a climate of constant control, abuse and fear, with no options to leave an employer without also losing their work permit and visa.
In studies which described their situation in positive terms (e.g., as having a safe job and income and a safe place to stay), such framings depended on the quality of the relationship with the employer/client, and decent employment practices (e.g., ensuring the worker’s day off and participation in familial decision-making). Being able to communicate with the family back home during the pandemic, e.g. via social media, was described as another source of resilience – and at the same time as potential stressor. Many migrant domestic care workers continued being responsible for their (left behind) families in the home country. This double responsibility of care translated into a “dual-country experience of the pandemic”; i.e. worries for the health and well-being of the family back home, increased pressure to send remittances, alongside the (often sole) responsibility for the care and health of the client in the host country.
(Synthesis based on: de Diego-Cordero et al. 2022, Lui et al. 2021, Kaur-Gill et al. 2021, Kuhlmann et al. 2020, Nasol & Francisco-Menchavez 2021, Sabar et al. 2021, Sanna 2021)
Among all 241 studies (Supplementary File, Table S6; p. 35ff) we identified seven studies (2·9% of all studies) that investigated vaccination coverage among migrants either empirically or anticipated by statistical models.37, 78, 110–114 Two modelling studies with medium and low quality recommended to include migrants in the national vaccination strategy to prevent COVID-19 incidence and ensure cost-effectiveness.111, 113 One medium quality study reported vaccination coverage within a Spanish prison but focussed on vaccination effectiveness of the respective vaccine.37 Three studies of high and medium quality from the European context reported lower vaccination rates among migrants (i.e., foreign-born) in comparison to the non-foreign-born population.78, 110, 112 The opposite, namely higher vaccination uptake among migrants compared to local residents was found in a medium quality study conducted in Nanjing and Chizhou, China.114
DISCUSSION
Our meta-analysis quantified inequalities between migrants and non-migrants in risk of COVID-19 infection, hospitalisation, ICU-admission, and death. We found that the incidence risk was more than twice the risk among migrants than among non-migrants, and was about seven percentage-points higher on average. Inequalities tended to be more pronounced in Northern Europe and North America, and lower in Southern Europe. These differences could be related to differences in access to health systems (e.g., based on legal status), in testing policies and management of SARS-CoV-2 cases, or in environmental risks (e.g., occupation and accommodation). Migrants in camps or other forms of institutionalised/shared accommodation appeared to be at highest risk of infection, albeit PIs for the subgroup analysis by migrant categories were extremely broad and overlapped (indicating high heterogeneity between studies within subgroups), so that differences should be interpreted with care.
We found no evidence for inequalities in risk of hospitalisation, which stands in contrast to inequalities observed between ethnic minorities and majority populations, indicating that different mechanisms are at play in the pathway between health and ethnicity versus migration.115 However, our meta-analysis indicates that, once infected and hospitalised, migrants may have experienced more severe courses of disease: the risk of being admitted to ICU was 36% higher in relative, and 4 percentage-points higher in absolute terms among migrants compared to non- migrants (albeit with broad PIs). While the share of migrants in deaths among hospitalised cases was lower (likely due to age differences), their share among deaths at population-level measured by excess-mortality was 31% higher in relative, and three percentage-points higher in absolute terms, compared to resident populations in respective studies.
Intersections among COVID-19-related exposures, risks, and impacts of pandemic measures in migrant populations compromised their physical and mental health in severe and sometimes unique ways. Cascading effects and feedback loops became evident, highlighting the syndemic nature of the COVID-19 pandemic.116, 117 Yet, we also found (potential) sources of resilience, indicating entry points for measurements (e.g., support from non-governmental organisations, peers or the local community) to improve migrants’ conditions in health emergencies. Overall, the qualitative results highlight the key role of socio-economic and work-related precarity, in combination with legal precarity.67, 86, 91 This is further underscored by studies, which describe that different States (i.e., Austria, Canada, and Italy) approved special policies to recruit migrant workers amid lockdown to work as fruit pickers and in other agricultural sectors, often in precarious labour contexts.43, 86, 98 Asylum seekers and other undocumented migrants were also absorbed in “essential” sectors to meet labour shortages.118 These examples illustrate the reliance of global economies of production on certain migrants as a captive and cheap workforce. And at the same time the COVID-19 pandemic has also shown that the same business models systematically shift the social costs of labour, including health risks, to the migrant workers themselves.119 The qualitative results thus underscore the importance of a political economy-perspective to understand the structures that generate and sustain social and health inequities for migrants. Sub-regional and regional governance structures that promote trade and mobility for economic cooperation, need to ensure migrant health as a key requisite of development. There is a critical need for governments, international organisations, civil society and migrant communities to engage with private sector, employers and industries in which migrant workers are overrepresented in precarious work contexts, with low-wages and with an irregular status. The underutilisation of the domain of work, including occupational health programs, in addressing the health of migrants is indeed truly a missed opportunity for global health.120
We found a high degree of heterogeneity among studies regarding the design, migrant populations, quality of data, and countries of studies. Poor reporting in primary studies raised considerable challenges in meta-analysis. This was most pronounced for mortality due to the use of different denominators: hospitalised SARS-CoV-2 cases or all-cause deaths in the total population. These two ways of reporting mortality have been identified by Karanikolos and McKee as a factor which substantially limits the comparability of COVID-19-related mortality between countries.121 Our review underlines the need for harmonised reporting in future health emergencies to ensure comparability of mortality estimates across countries and studies. The trend to lower mortality in hospitalised cases is dependent on testing strategies, and adjustment for age and comorbidities, which was not always performed in primary studies yielding estimates prone to confounding. In contrast, excess rates of age- and sex-standardised mortality are more likely to reflect the true mortality risks. In view of this, our results show that migrants tended to be at considerably higher risk of death associated with COVID-19 than non-migrants during the pandemic (until 11/2021).
The strength of this review is the synthesis of all (high to moderate quality) evidence from qualitative and quantitative studies on health-related impacts of the COVID-19 pandemic on migrants worldwide. The analysis of the interconnectedness of exposure, risk, and impact of pandemic measures at micro-, meso-, and macro-levels benefitted from qualitative research, which captures social phenomena in their complexity and within their particular context, thereby allowing us to contextualise findings. Despite an overwhelmingly broad landscape of literature, however, our knowledge of some groups and migration phases still remains scarce and sketchy, for example, related to labour migrants, undocumented migrants, migrants with disabilities, elderly migrants, returnees, migrants in transit or in pre-migration phases. Countries of studies were mainly destination countries so that evidence is skewed toward COVID-19-related migrant health in the post-migration phase. Further efforts are required to better reflect health-related aspects of the complex migration trajectory.
The meta-analysis provides valuable evidence of the magnitude of COVID-19-related inequalities, despite being limited by the striking heterogeneity of underlying primary studies. Cautious interpretation is required, however, due to poor quality of reporting in primary quantitative studies. We could not adjust the RR for age and sex because the data were either not reported or even unavailable to the authors of primary studies upon request. This highlights the urgent need to improve the reporting quality and primary data in studies presenting migration-stratified outcomes. The broad search enabled to include studies from different research fields, but this posed a challenge to the applicability of the quality appraisal tools. For 15 studies, only a few questions of the JBI checklists were applicable. We included English, German and Spanish articles, so that other languages, e.g., French, Chinese or Arabic, could not be considered and thus studies from respective countries may not have been included.
Our review sheds a glaring light on the lack of evidence on important measures of pandemic response, such as vaccination coverage among migrants, not only until the end of 2021. Updating our search in the WHO COVID- 19 research database on vaccination coverage among migrants for the years 2022 and 2023 resulted in 204 hits (April 2023). Of these, only about 15 studies addressing this topic could be identified after screening titles to report vaccination coverage/uptake/rates among the respective migrant populations. As with the search for this review, the focus of the research appears to be on vaccination hesitancy, knowledge and attitudes toward COVID-19 vaccination, or different ethnic groups, rather than on coverage rates in migrants as defined previously. The scarcity and poor quality of reporting of studies points to the weakness of Health Information Systems (HIS) to provide reliable data on key aspects of migrant health such as vaccination coverage, by the end of 2021, and even in 2023. The lack of harmonisation across different migrant categories and outcome measures puts validity and comparability of findings at risk. Our findings hence call for urgent implementation of recommendations and WHO technical guidance to enhance the integration and availability of migrant indicators in HIS.122, 123 This requires urgent action in the post-COVID-19 recovery phase to bridge the divides between health and migration governance by means of effective collaboration structures at all levels of government (local, national, international).124 Monitoring migrant health should be considered an essential component of pandemic preparedness, and future national plans must secure an adequate inclusion of migrant groups which promote social and health equity. This includes anticipation and prevention of unequal effects or unintended (negative) consequences of the pandemic on migrant health that are exacerbated by language barriers, stigma and discrimination, and by financial, administrative and legal barriers to health systems.
Contributors
MH: Conceptualization, Methodology, Investigation, Data curation, Formal analysis – statistical analysis, Writing – Original Draft, Visualization, Project administration
NG: Investigation, Formal analysis – qualitative analysis, Data curation, Visualization, Writing –Review & Editing
SR: Methodology, Investigation, Formal analysis – statistical analysis, Data curation, Visualization, Writing – Review & Editing
JO: Investigation, Data curation, Formal analysis, Visualization, Writing – Review & Editing MB: Investigation, Formal analysis – qualitative analysis, Writing – Review & Editing
SA: Methodology Investigation, Formal analysis – qualitative analysis, Writing – Review & Editing
JL: Methodology, Investigation, Formal analysis – qualitative analysis, Writing – Review & Editing
SF: Investigation, Writing – Review & Editing
AM: Investigation, Writing – Review & Editing
KW: Conceived the study, Conceptualization, Methodology, Writing – Review & Editing
KB: Conceived the study, Conceptualization, Methodology, Investigation, Formal analysis, Validation, Writing – Review & Editing, Supervision, Funding acquisition
All authors had full access to all data in the study and had final responsibility for the decision to submit for publication.
Declarations of interest
AM, KB, MH, SF received an individual honorarium from IOM. All authors declare no competing interests.
Data sharing
A full list of studies identified in the search as well as the full data extracted from included studies are available for academic research projects by request to the corresponding author: Kayvan.bozorgmehr{at}uni-bielefeld.de
Acknowledgments
No funding received.
References
