How do social determinants and legal status affect the health of undocumented migrants in France?

Abstract

Background

The health of undocumented migrants (UMs) in France is poorly understood, especially outside of health care settings. This study aimed to (1) describe the health status of UMs in France, (2) to analyse the influence of the social determinants of health (SDoH) on four main reported conditions and self-perceived health and (3) to compare UMs with documented migrants (DMs) for those five outcomes.

Methods

The Premiers Pas study (PPS) is a cross-sectional survey that is representative of UMs aged 18 years or older living in France. The study was performed between February and April 2019 at sixty-three sites. The 2019 European Health Interview Survey (EHIS), representative of the French population, was subsequently used to compare UMs with DMs. Five outcomes were compared: diabetes, high blood pressure (HBP), asthma, depression (using the PHQ-9) and self-perceived health (SPH, using the Minimum European Health Module). Logistic regression models were used to explore associations between the five outcomes and (i) SDoH among UMs and (ii) administrative status (UMs vs DMs).

Results

Among UMs, 68.1% reported having a current health condition (n = 1188), and 33.5% reported having a chronic condition (n = 1189). Musculoskeletal (30.3%), mental (27.4%) and digestive (12.0%) conditions were among the most reported conditions. Several SDoH were risk factors for poor outcomes. For example, food insecurity was associated with a greater risk of depression (‘often’ aOR = 3.05 95% CI [1.58–5.89]). The prevalence of depression and poor SPH was significantly greater for UMs than for DMs for the same gender and age groups. After adjusting for age, gender, region of origin and job status, depression was higher (aOR = 2.35 95% CI [1.85–2.99], p < 0.001), and poor SPH was worse (aOR = 2.49 95% CI [1.89–3.29] p < 0.001) among UMs.

Conclusion

This study showed that living conditions play an important role in the health of UMs. UMs had higher risks of poor SPH and depression than DMs. These inequalities persisted when stratifying by duration of residence and gender. Living conditions could be improved through targeted public policies that promote migrants’ rights, social inclusion and legalization.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-27924-x.

Introduction

The International Organization for Migration defines irregular migration as “a movement that takes place outside of the regulatory norms of the sending, transit and receiving country” [1]. Migrants can shift in and out of irregularity depending on laws and policies. In France, one in five documented migrants (DMs) have been undocumented at some point [2]. Undocumented migrants (UMs) are those without residence permits. Their numbers are difficult to estimate, both in France and elsewhere. While some enter illegally, many become undocumented due to visa overstays, permit nonrenewals, or rejected asylum claims [2, 3].

Theories such as the "healthy migrant effect" suggest migrants are initially healthier than their peers at home due to selection (of healthier and younger individuals) and opportunity factors related to or associated with migration [4–6]. However, this benefit declines over time—a phenomenon known as the "exhausted migrant effect" [7], driven by poor working and housing conditions, food insecurity, stress, and other social factors [8, 9]. While this applies to DMs, it's unclear how it applies to UMs.

UMs in France are a diverse group. The Premiers Pas study (PPS) found over half entered without visas [10]. Those without visas face greater precarity and are less likely to have health insurance. In contrast, those who previously held visas tend to be more integrated and have better healthcare access, which improves with time spent in France [10, 11].

UMs' health remains poorly understood and differs from DMs due to unique social determinants. Because of the legal status they have restricted access to the labor market, and social services which contributes to worse health outcomes [12, 13, 14, 15, 16, 17, 18]. They are often in low-paid, high-risk jobs with long hours [19–22], face unstable housing, and are exposed to poor living conditions affecting both physical and mental health [23–26]. Compared with the general population, UMs show worse overall health, including mental health and self-perception [25, 27, 28].

Access to healthcare is also limited. UMs often delay or forgo care, including urgent care, due to lack of insurance [10, 11, 15, 27–29, 29–33]. In France, AME (State Medical Assistance) is available to UMs under specific conditions [34], but nearly half of eligible individuals don’t benefit from it [10, 33, 35–38]. Even insured UMs face discrimination and care refusal [39–42]. “Over time, the AME has undergone numerous reforms aimed at reducing its scope or tightening access terms. The conditions of eligibility and access were modified between 2020 and 2021 (this study was conducted prior to these changes). These included an obligation to submit a file and a waiting period for certain non-urgent care [43]. More recently, in November 2023, as part of an immigration bill, the French Senate voted to replace the AME with access to urgent care only. The law was later rejected [44]. These repeated threats against the AME are part of a broader context of an increasingly hostile political climate towards migrants [45].

Legal status impacts health through a) social determinants and b) stress, fear of arrest, and isolation. Studies in Switzerland show that obtaining legal residency improves mental health [46–48]. UMs often avoid seeking care due to fear of deportation, a phenomenon known as the “chilling effect” [49, 50], which contributes to poorer health outcomes [51, 52].

While DMs experience a health decline over time, UMs often begin migration with poorer health and face ongoing barriers, with little research on how their health evolves [15, 23, 26, 33, 47, 48, 53, 54]. Differences in social determinants may explain more than legal status alone [12, 14, 46, 48].Thus, the exhausted migrant concept may not fully capture the complexity of UMs’ health trajectories.

To our knowledge, no population-based study has compared health between UMs and DMs in France. At the European level, outside of Switzerland, we have not found any such studies [22, 46–48]. We then hypothesized that UMs have poorer health than DMs.

The first objective of this work was to describe health status and identify the most frequently reported conditions and symptoms among UMs in France.

The second objective was to analyse how SDoH influence five major health outcomes: diabetes, high blood pressure (HBP), asthma, depression and self-perceived health. These five outcomes were chosen because (1) the literature reports high prevalence rates among migrants and UMs and (2) SDoH have a significant effect on these outcomes [15, 25, 32, 46, 47, 54, 55].

The third objective was to compare UMs with DMs in terms of those five outcomes.

To meet these objectives, we drew on 2 studies: the PPS, which was conducted among UMs, and the European Health Interview Survey (EHIS), which is representative of the French general population, and DMs with longer residences in France.

Methods

Study design and population

PPS is a cross-sectional survey representative of UMs living in France. The study was carried out between February and April 2019. The inclusion criteria were status as an UM and an age greater than 18 years. Fourteen different languages were used to translate the questionnaire and interact with the participants. Individuals were randomly selected from 63 facilities that provide support or assistance to deprived or migrant persons in the Parisian and Bordeaux regions [33]. Individuals were recruited mainly through nonhealth care facilities whose missions varied: social services for homeless people (for example, daytime shelters or shower baths), support for legal or administrative issues, and local health insurance centres or NGOs. The few health care facilities were NGOs (for example, Doctors of the Word) or public facilities. A total of 1223 participants were included in the PPS. Details of the methodology can be found in the articles published previously [10, 33, 35, 56].

The EHIS is a health survey conducted every six years in the 28 countries of the European Union [57]. It is a representative survey of the French population aged 15 years or older living in ordinary housing. The population was randomly drawn from individuals registered with the tax authorities (for income tax, council tax or both) and those declared as having ordinary dwellings (excluding people housed in shelters, collective housing or homeless people). Given the sampling, the presence of undocumented individuals can be considered marginal, and the two samples can be considered distinct. Thus, migrants from EHIS can be considered documented migrants (DMs). To identify migrant individuals, the place of birth and nationality at birth were used. Those born without French nationality in a foreign country were considered migrants. A total of 14,192 individuals were included in the EHIS, and 1478 were migrants. For our study, we used 2019 data from migrants aged 18 years or older living in metropolitan France.

Data and variables

PPS

A range of socioeconomic indicators were collected and categorized into coherent groups according to the following: demographic characteristics (age and gender), migration characteristics (region of origin, time since arrival in France, motives for migration, administrative status at arrival in France), social characteristics (having a job, job in the country of origin, fluency in French), the family environment (isolation), material living conditions (housing type, food insecurity) and access to care (AME). The category "precarious housing" included unstable accommodations such as emergency hotel placements for the night, sleeping on the street, in public places, squats, shelters, and similar situations. The participants were asked if they were experiencing any of the following conditions: diabetes; HBP; cancer; cardiovascular, respiratory, infectious, skin, joint and muscle, liver, neurologic, or gynaecologic diseases; mental health conditions; or other conditions. The exact question was as follows: “Do you currently suffer from any of the following health problems [list of conditions]?”. If the individual answered yes to one of the organ-related conditions, more specific questions were asked to specify the disease or the symptom. In different questions, participants were asked if they had a chronic disease/condition, whether they had dental pain, missing or broken teeth or infected teeth (cavities), if they had eyesight disorders and if they wore glasses. Based on these two questions, we identified current and chronic conditions. Independent of the answer to the main question, all the respondents were surveyed by the Patient Health Questionnaire-9 (PHQ-9) [58].

A cut-off score of 10 or more was used to estimate moderate to severe depression [59]. Pregnancy status in the last 12 months was recorded, and data were collected. Self-perceived health was assessed by the Minimum European Health Module (MEHM) [60]. Self-perceived health is a global measure that encompasses different dimensions of health, such as “physical, social and emotional function as well as biomedical signs and symptoms” [61]. For the question “How is your health in general?”, the MEHM has five possible answers: “very good–good–fair–bad–very bad”. We built a binary variable as follows: “very good–good–fair” as good self-perceived health and “bad–very bad” as poor self-perceived health. In this paper, we used the category “poor self-perceived health” to describe perceived health to stay in line with the four other outcomes that represent a pathological status [60]. To ensure representativity and to correct for nonresponse, the data were weighted based on the probability of inclusion. Survey weights were constructed for each facilitiy, based on the number of UMs received during a representative week and the number of UMs present at the facility on the day of data collection. The weights were then adjusted to correct discrepancies between the reported weekly average and the attendance observed during the day of the interview.

EHIS

The data included in our analyses from EHIS were age, gender, place of birth, length of residence in France and having a job. Our outcome variables are measured using three questions (one question per outcome): “Over the last 12 months, have you had diabetes?”, “Over the last 12 months, have you had high blood pressure?”, and “Over the last 12 months, have you had asthma?”. Individuals who report having any of these conditions are coded as 1, and as 0 otherwise. We also include depression, measured using the PHQ-9 score, with a cut-off of 10 or higher to indicate moderate to severe depression. In addition, we use self-reported health, defined as a binary variable equal to 1 for individuals reporting very good, good, or fair health, and 0 for those reporting bad or very bad health. To ensure representativity and to correct for nonresponse, the data were weighted based on the probability of inclusion.

Statistical analysis

In the first step, we described the prevalence of declared conditions and symptoms among UMs from the PPS using percentages. Major diseases by organ were subsequently compared by gender using a chi-square test. Finally, we performed univariate and multivariate analyses on the PPS data for diabetes, HBP, asthma, depression and poor self-perceived health as outcomes and socioeconomic indicators as independent variables. When an association was significant in the univariate analysis, we performed multivariate analyses. A difference was considered significant when the p value of the crude odds ratio (OR) for the univariate result was < 0.200. A p‑value cutoff ranging from 0.15—0.20 is often used and recommended during initial variable selection because a stricter 0.05 can omit variables that may be important predictors [62, 63]. To address collinearity and avoid overadjustment in the multivariate analyses, we used a hierarchical regression method. A blockwise selection process was performed, and the different categories of socioeconomic indicators were added step by step. In this six-step process, we added each group of variables sequentially, monitoring how the coefficients changed to ensure that the models remained robust. To avoid weighing down the models and reducing their performance, we added only the variables with a p value of the crude OR of less than 0.200 to the model. Alternatively, within a group of variables for which all the crude ORs were greater than or equal to 0.200, we selected the variable with the smallest p value.

Because of power constraints, the age categories “’40–59 years” and “60–82 years” were grouped together into the category “40–82 years” in some analyses when there were small sample in one category.

In the second step, the results from the PPS (UM) were compared with those from the EHIS for diabetes, HBP, asthma, depression and poor self-perceived health. For this purpose, the two databases pertaining to the two samples (EHIS and PPS) were pooled. For each outcome, the two populations were compared based on age and gender, using percentages and chi-square tests. When differences were significant for age or gender, multivariate analyses were performed using logistic regression and adjusted for age, gender, region of origin and job status. Length of residence in France was not used to adjust the models because the distribution for this variable was too different for the two populations. Socioeconomic indicators such as housing, food insecurity or social support were not available in EHIS sample and thus were not used to adjust the models comparing UMs and DMs. The substantial disparities in length of residence and socioeconomic indicators between UMs and DMs was a major limitation of our study and limited the attribution of health differences solely to legal status.

To take into account the length of residence, sensitivity analyses were performed. The multivariate models comparing UMs and DMS were stratified by length of residence according to the two following categories: “less than 10 years” and “10 years and more”. The 10 years cut-off was chosen because it allowed to have the best power to compare the two populations of migrants. The models were adjusted for age, gender, region of origin and job status. DMs were the reference for all the multivariate analyses.

The threshold of significance was 0.05, and 95% confidence intervals were used. All analyses were performed with Stata© 15.1 software (StataCorp, College Station, TX, USA).

Results

Description of UMs from the PPS

A total of 1188 individuals answered the main question related to an ongoing condition, and 1189 answered the question related to a chronic condition. The characteristics of the sample have been described in previous publications and are detailed in Table 1 [10]. A total of 28.8% of the respondents were women, 36.3% were aged 18–29 years, 33.9% were aged 30–39 years, and 62.4% were from sub-Saharan Africa. Those who entered illegally represented 58.7% of the sample [10].

Table 1.

Description of the characteristics of the Premier Pas survey and EHIS, 2019

	Premiers Pas survey		EHIS		p value
	n	% (weighed)	n	% (weighed)
Total	1223		1478
Demographic characteristics
Gender	1220		1478		< 0.001
Woman	426	28.8	783	52.6
Man	794	71.2	695	47.4
Age class	1216		1478		< 0.001
18–29	333	36.3	146	10.3
30–39	443	33.9	269	19.3
40–59	384	27.6	624	40.8
60 and more	56	2.3	439	29.6
Migration characteristics
Region of origin	1209		1477
North-Africa	363	23.7	522	34.4	< 0.001
Sub-Saharan Africa	630	62.4	241	15.8
Asia	50	3.4	158	10.3
Europe European-Union	47	1.8	367	27.4
Europe outside European Union	39	2.3	143	9.1
Americas and Oceania	80	6.3	46	3.0
Arrival in France	1215		1471
Less than 3 months	132	15.3	1	0.1	< 0.001
[3 months −1 year[	324	22.5	5	0.3
[1 year −3 years]	314	28.3	43	3.1
[3 years −5 years[	147	11.7	71	5.9
More than 5 years	298	22.2	1351	90.6
Motive for migration	1036
Health	121	8.4	/	/	/
Not health	915	91.6	/	/	/
Administrative status upon arrival in France	1189
With visa	623	41.3	/	/	/
Without visa	566	58.7	/	/	/
Social characteristics
Current job	1212		1478		< 0.001
Yes	286	21.1	745	54.4	/
No	926	78.9	733	45.6	/
Job in country of origin	1214
Employee and self-employed	540	44.9	/	/	/
Manual worker	361	32.3	/	/	/
Students and other	107	8.1	/	/	/
Does not work/has never worked	206	14.7	/	/	/
Fluency French	1222
Very good/somewhat good	643	56.4	/	/	/
Not very well/poorly or very badly	277	21.6	/	/	/
Non-French speaking	302	22.0	/	/	/
Family environment
Isolation	1191
Alone	475	41.5	/	/	/
Alone with child	226	18.5	/	/	/
In a couple without children	132	10.7	/	/	/
In a couple with child(ren)	358	29.3	/	/	/
Material living conditions
Housing type	1214
Ordinary	536	38.1	/	/	/
Collective	219	19.4	/	/	/
Precarious	459	42.5	/	/	/
Food insecurity	1214
Often	321	30.7	/	/	/
Sometimes	448	36.6	/	/	/
Never	445	32.7	/	/	/
Access to care
Aide Médicale d'Etat	1218
Yes	526	43.8	/	/	/
No	692	56.2	/	/	/

/not available in the data base, p-value (Chi2), p-value using chi-square test

Health status: description

Among the respondents, 68.1% reported having an ongoing health condition (n = 1188), and 33.5% reported having a chronic condition (n = 1189). The main ongoing health conditions reported were musculoskeletal (30.3%), mental (27.4%), digestive (12.0%), respiratory (10.3%), skin (7.8%), infectious (7.2%), neurologic (4.2%), and cardiovascular (4.0%) diseases. The other categories of health conditions, including cancers, represented less than 3% of the total sample. Figure 1 presents the distribution of declared ongoing health conditions by gender (those with a prevalence of less than 3% are not displayed). In total, 14.3% of the individuals reported suffering from pain. Among the women, 12.1% reported having a gynaecological disease or symptoms (n = 418), 13.7% were pregnant at the time of the interview (n = 417), and 27.6% had a pregnancy in the last 12 months (n = 345). No significant difference was observed between men and women except for ophthalmologic disease (woman = 0.7% and man = 2.6%, p = 0.039) (Fig. 1).

Fig. 1.

Weighted prevalence of ongoing health conditions in order of frequency and according to gender, Premiers Pas survey, 2019, n = 1189. Women are represented in green or yellow, and men are represented in orange or purple

Figure 2 presents the most frequently reported diseases, with a prevalence > 3%. Dental problems were at the top of the list of reported conditions in order of frequency, followed by mental and cardiovascular diseases (HBP and diabetes).

Fig. 2.

Weighted prevalence of conditions by order of frequency, Premiers Pas survey, 2018, n = 1188. Only prevalences higher than or equal to 3% are displayed. HBP: high blood pressure. HIV: Human immunodeficiency virus

For perceived health (n = 1219), 18.8% described it as very good, 34.2% as good, 22.2% as fair, 19.1% as bad and 5.7% as very bad. According to the binary variable, 24.8% of the participants reported poor self-perceived health.

Comparative analyses with SDoH among Ums

The results from the multivariate analyses displayed in Table 2 revealed various influences of SDoH on the outcomes. Worse living conditions were associated with poorer health: food insecurity was associated with a greater risk of depression (‘often’ adjusted odds ratio, aOR = 2.66, ‘sometimes’ aOR = 2.34), asthma (‘often’ aOR = 3.55) and poor self-perceived health (‘often’ aOR = 2.86). Living in collective or precarious housing was associated with a greater risk of poor self-perceived health (aOR = 3.46 and aOR = 2.34, respectively).

Table 2.

Univariate and multivariate analyses for diabetes, high blood pressure, asthma, depression and poor self-perceived health, Premiers Pas Study, 2019

p-value (Chi2): p-value using chi-square test,

OR Crude odds-ratio, aOR adjusted odds-ratio, a: crude odd ratio, b: adjusted odd ratio ; p-value: °p < 0.20, *p < 0.05, **p < 0.01, ***p < 0.001, 95%CI: 95% Confidence interval; /: variable not included in the final model ref: reference category

(a): Univariate analysis (b): Multivariate analysis: Models after adjusting for socioeconomic related variables : gender, age, employment, job in the country of origin, fluency in French,region of origin, time since arrival in France, motive of migration, isolation, housing type, food insecurity, healthcare insurance with the Aide Médicale de l’Etat. Results are presented as follows: aOR 95%CI, p-value

An occupation and a better social environment were associated with better health: having a job was associated with a lower risk of depression (aOR = 0.46) and HBP (aOR = 0.18), whereas having children was also associated with a lower risk of depression (‘alone with child’ aOR = 0.44, and ‘in couple with child’ aOR = 0.50).

Migratory characteristics such as time spent in France and region of origin had a variable influence on health. The risk of asthma seemed to increase with time spent in France (aOR = 10.10 for those in France for more than five years when reference category was “[1 year −3 years]”). At the same time, those in France for three months to one year had a higher risk of asthma (aOR = 6.32) than those in France for one to three years (reference category). The category ‘other’ for the region of origin had a lower risk of diabetes (aOR = 0.19), of HBP (0.28) and a higher risk of asthma (aOR = 6.18) than did sub-Saharan Africa. For power purposes, this category included different groups of populations: the Americas (mostly Latin America), the European Union (EU) and Europe outside of the EU and Asia and Oceania. For diabetes, the weighted distributions were as follows: North Africa, 7.3% (33/356); sub-Saharan Africa, 4.3% (26/615); Americas and Oceania, 6.7% (5/75); EU, 4.3% (2/47); Europe outside EU, 0.0% (0/35); and Asia, 4.3% (2/47). As HBP, they were as follows: North Africa, 12.9% (35/356); sub-Saharan Africa, 6.5% (54/615); Americas and Oceania, 6.7% (5/75); EU, 8.5% (4/47); Europe outside EU, 8.6% (3/35); and Asia, 8.5% (4/47). The prevalence rates of asthma were as follows: North Africa, 3.9% (22/356); sub-Saharan Africa, 2.5% (26/615); Americas and Oceania, 6.7% (5/75); EU, 4.3% (2/47); Europe outside EU, 5.7% (2/35); and Asia, 4.3% (2/47).

Better fluency in French was associated with poorer health: a higher risk of diabetes (‘not very well/poorly’ aOR = 4.95) and a higher risk of depression (‘very good/somewhat good’ aOR = 2.29). Finally, entering in France with a visa was a protective factor for depression (aOR = 0.57).

Comparison of UMs with DMs

Among the migrant population in EHIS, 52.6% were women, and 47.4% were men (Table 1). DMs in EHIS were older than UMs in PPS. For the region of origin, DMs were more often from North Africa in EHIS (34.4%), Europe (EU 27.4%) and Europe outside EU (9.1%) and less often from sub-Saharan Africa (15.8%) than were UMs (PPS). DMs had a longer length of residence in France, with 90.6% declaring living in France for more than 5 years, whereas it was only 22.2% among UMs in PPS.

Table 3 provides comparisons between the PPS population (UM) and the EHIS migrant population (DM) stratified by gender and age, for the five outcomes (diabetes, HBP, asthma, depression and poor self-perceived health) and Table 4 shows comparisons by age stratified by gender between the two population and for the five outcomes.

Table 3.

Comparisons between the Premiers Pas population (undocumented migrants) and the EHIS migrant population (documented migrants) according to gender, age and region of origin for diabetes, high blood pressure, asthma, depression and poor self-perceived health, 2019

p-value (Chi2): p-value using chi-square test,

p-value : °p < 0.20, *p < 0.05,**p < 0.01, ***p < 0.001

Table 4.

Comparisons by age stratified by gender between the Premiers Pas population (undocumented migrants) and the EHIS migrant population (documented migrants) according to age for diabetes, high blood pressure, asthma, depression and poor self-perceived health, 2019

p-value (Chi2): p-value using chi-square test

p-value : °p < 0.20, *p < 0.05,**p < 0.01, ***p < 0.001

Multivariate analyses comparing UMs from PPS and DMs from EHIS, adjusting for age, gender, region of origin and job status, revealed a significantly greater risk of depression (aOR = 2.35) and poor self-perceived health (aOR = 2.49) among UMs. For asthma, the risk was lower for UMs (aOR = 0.58). Figure 3 synthetizes the results from the multivariate analyses for the outcomes.

Fig. 3.

Adjusted odds ratios for the five outcomes diabetes, high blood pressure, asthma, depression and poor self-perceived health adjusted for age, gender, region of origin and having a job. Comparison between undocumented migrants from Premiers Pas study and documented migrants from EHIS. The reference category is the documented migrants from the EHIS. aOR: adjusted odds ratio. The figures in square brackets are 95% confidence intervals

Sensitivity analyses

Sensitivity analyses were performed in order to compare the variable “self-perceived health” depending on the dichotomization. In this paper we used the following dichotomization: “very good–good–fair” as good self-perceived health and “bad–very bad” as poor self-perceived health. We compared the results of this dichotomization with those of the following: “very good–good” as good self-perceived health and “fair-bad–very bad” as poor self-perceived health. The results were similar and consistent between the two dichotomizations, and we did not identify any conflicting results among UMs (univariate and multivariate analyses) and when comparing UMs to DMs.

Multivariate analyses stratified by length of residence were performed to compare UMs and DMs.The detail of the analyses is presented in supplementary tables. For the length of residence “ten years or more”, a clear difference appeared in multivariate analyses for HBP with UMs having a lower risk of HBP compared to DMs: aOR = 0.45, 95%CI [0.25–0.81], p = 0.008. No difference appeared for diabetes and asthma probably due to the small sample size reducing the power. As for depression, the differences persisted for the length of residence “less than 10 years” aOR = 2.81, 95%CI [1.88–4.18], p < 0.001) as well as for “ten years or more” (aOR = 2.11, 95%CI [1.35–3.30], p = 0.010). Finally for poor perceived health, the differences also persisted and appeared even stronger for the length of stay “less than 10 years” (aOR = 4.18, 95%CI [2.37–7.35], p < 0.001). For “ten years or more” it was aOR = 2.11 [1.31–3.39], p = 0.002.

Sensitivity analyses were performed for univariate and multivariate analyses among UMs. The strength of the associations varied little, but their significance did not change when people recruited in health care facilities were excluded from the analyses.

Discussion

Main results

As reported in previous studies, musculoskeletal, mental, digestive conditions, and dental problems were the most frequently reported health conditions among UMs [31, 54, 64]. Pain was also commonly reported [55].

Our findings confirm the strong influence of SDoH on UMs’ outcomes. Poor living conditions, including food insecurity and insecure housing, were associated with depression, asthma, and poor self-perceived health, while social support and employment appeared protective. This aligns with existing literature [15, 25, 29, 31, 32]. Access to health care was not significantly associated with these outcomes in our models.

Comparison with DMs

Our second hypothesis was confirmed: UMs had significantly poorer mental and self-perceived health compared to DMs. Asthma was less prevalent among UMs, with risk increasing with longer residence in France. No significant differences were found for diabetes or HBP. In the stratified analyses, the risk of HBP was lower for UMs in France for ten years or more compared to DMs in France for ten years or more.

Differences in health status may stem from legal status, which contributes to fear, self-isolation, and stress, worsening mental health status [46, 47, 50]. Poorer living conditions, often tied to legal exclusion (e.g., no access to social housing or formal employment), further exacerbate disparities [22–24]. However, multivariate models comparing UMs and DMS could not account for several SDoH (such as housing, food insecurity, or length of stay) due to limitations in the EHIS data. Thus, some observed differences likely reflect the broader impact of living conditions and integration, beyond legal status alone.

Regarding asthma, lower prevalence among UMs may be due to their shorter stay in France. The literature suggests an "allergy catch-up" mechanism, where prolonged exposure to new environments reprograms immune responses, increasing asthma risk over time [65].

In the multivariate analyses stratified by length of residence, HBP appeared to be significantly lower among UMs than among DMs for the length of residence “ten years or more”. While trends suggest lower risk of diabetes among UMs, more robust studies are needed. Risk factors for HBP and diabetes include socioeconomic context, health behaviours, and access to preventive care (all influenced by post-migration conditions) [66, 67].

Mental health and self-perceived health

Over a quarter of participants reported mental health issues, including high rates of sleep disorders, anxiety, and depression. The PHQ-9 was consistently used across both UMs (PPS) and DMs (EHIS), minimizing measurement bias. Interviews were conducted in multiple languages, using validated tools [68–72].

Poor self-perceived health, reported by a quarter of UMs, reflects multidimensional health and correlates with mortality [61, 73, 74]. Though it may be influenced by non-medical factors (e.g., age, education, trust, interviewer characteristics) [75], our study showed clear associations between poor living conditions and worse self-perceived health. These findings are consistent with prior research linking this outcome to socioeconomic vulnerability and lack of social support [12–15].

Women had better self-perceived health than men but differences were not significant in comparisons and in multivariate analyses. This could be linked to a lower impact of the lack of a residence permit and the migration process on women. In our study women were less likely to be in a precarious situation than men: they were more likely to live in “ordinary” housing, they were more likely to be in a relationship, their reason for migration was more often family-related, and they were slightly less likely to experience food insecurity. Our findings are consistent with international literature showing that migrant men in high-income countries are more often exposed to visible forms of precarity (for example unstable employment or housing), whereas migrant women, more frequently arriving through family-related pathways and embedded in household structures, may experience relatively more stable living arrangements despite other forms of structural vulnerability such as gender-based violence [76]”.

UMs who had been in France for less than three months reported poorer self-perceived health. However, these differences did not persist in multivariate analyses. This could be due to a reporting bias related to the difference between expectations and reality upon arrival. It could also be linked to a genuine deterioration in health related to the migration journey and the difficulties encountered by recently arrived migrants on arrival (e.g. difficult access to employment and housing, administrative issues, food insecurity, isolation, etc.). These difficulties may be more significant initially and diminish over time. Contrary to some studies [16], we found no difference in self-perceived health between health insurance beneficiaries and non-beneficiaries [16]. However, our results are aligned with existing evidence that legal status and integration improve self-perceived and mental health [77].

Chronic conditions

Chronic disease prevalence (diabetes, HBP, asthma) was similar between UMs and DMs across age groups, except asthma was more common among younger DMs. Studies show chronic diseases are not negligible among UMs, especially when deprived of housing, insurance, or legal status [31, 54, 77–80]. UMs often face delayed care and limited medication access [15, 32, 81, 82].

Diabetes prevalence is particularly high among North African migrants, likely due to lifestyle factors [83, 84, 85]. Length of stay in the host country may influence the risk of developing diabetes, though no difference was found in our study, likely due to limited power [86–89]. Environmental and lifestyle factors post-migration contribute to diabetes development [90–93]. While obesity and overweight were more frequent among women and North Africans in our sample, they were not associated with diabetes in multivariate models.

Interestingly, better French fluency was linked to higher risk of diabetes and depression risk. This may reflect higher health literacy and diagnosis awareness, or other unmeasured confounders [94]. Although most studies report poorer mental health among migrants with lower language proficiency, some evidence suggests the opposite pattern, which may be linked to both improved access to healthcare and diagnosis [95]. These associations warrant further investigation.

Other health conditions

Skin and dental issues were also commonly reported. Skin conditions may sometimes arise from stress or environmental exposure in precarious housing [96]. Dental pain was the most reported symptom, aligning with other studies showing poor oral health among UMs [97, 98]. Social support may help mitigate dental health issues [99].

Communicable diseases

Prevalence rates for HIV, hepatitis B, and C were nonnegligible and discussed elsewhere [26]. Existing data suggest underestimation, particularly among UMs from sub-Saharan Africa [31, 100, 101].

Prenatal care

Two out of five UM women were pregnant or recently pregnant, raising concern about access to adequate prenatal care. Literature indicates higher risks for poor outcomes among UM women [52, 102].

Strengths

This study is the first in France to analyze the health of UMs beyond healthcare settings, using a diverse, randomly selected sample. The multilingual approach and varied recruitment sites ensured broader inclusion. Although the sample size was limited, the UM population is hard to reach, making these findings significant. The comparison with a representative DM sample adds value, especially with consistent measures like PHQ-9 and self-perceived health.

Limitations

Our study faced limited statistical power for some outcomes, particularly for diabetes and asthma. The sample, though randomized, covered only two regions and may not fully represent UMs across France. Still, most UMs reside in urban areas, especially around Paris, supporting generalizability [103].

Results were self-reported, which may lead to underreporting, especially for conditions requiring clinical diagnosis (e.g., diabetes, hypertension, viral infections) [11, 34]. Lack of regular checkups and insurance coverage may contribute to diagnostic gaps. This aligns with evidence of underdiagnosis of diabetes and HBP in the general French population [104, 105].

A key limitation was the inability to adjust for length of residence, which differed substantially between the groups (80% of UMs < 5 years vs. 10% of DMs), and for socioeconomic factors reflecting living conditions, which also differs substantially between UMs and DMs (such as housing, food security, social support). Thus, those disparities limit the attribution of health differences solely to legal status. Other unmeasured confounders may also influence findings. Therefore, differences in health status need to be further investigated and confirmed by other studies that take these factors into account. Despite these limitations, the differences observed between the two groups in terms of mental health and self-perceived health reflect reality and are relevant from a public health perspective.

Conclusion

This study underscores the negative impact of living conditions on UM’s health. Living conditions can be addressed and improved through targeted public policies. Policies that promote migrants’ rights, social inclusion, and legalization, are beneficial from a public health perspective [106–109]. In today’s increasingly hostile and polarizing political climate towards UMs, this research provides objective, evidence-based insights [45]. Health-related motives for migration were relatively low and did not explain health status to any great extent. This dispels common misconceptions and fear-mongering claims that health-related motives are a major driver of migration [110]. Compared to DMs of the same age, gender, region of origin, and employment status, UMs were generally younger and experienced fewer chronic conditions. However, they often endure physically demanding, hazardous and low-status occupations that are avoided by the general population [12, 19, 52, 111, 112] Despite their contribution to the labor force, they often remain socially marginalized and exposed to high levels of psychosocial stress.

Although health insurance did not explain health disparities in our findings, and while public policy is a strong structural determinant of health with the most powerful influence on health outcomes, the healthcare system also plays a critical role [113]. The Aide Médicale de l’État healthcare insurance program, although frequently targeted in political debates, is a cost- effective way to ensure basic healthcare coverage [114, 115]. Ensuring access to healthcare for all regardless of legal status is a matter of social justice and a cornerstone of an effective and inclusive public health system.”

Abbreviations

AME

Aide Médicale de l’Etat

DMs

Documented migrants

EHIS

European Health Interview Survey

EU

European Union

HBP

High blood pressure

PASS

Permanence d’Accès aux Soins de Santé

PPS

Premiers Pas study

SDoH

Social determinants of health

UMs

Undocumented migrants

Authors’ contributions

S.M., P.D. and J.W. made substantial contributions to the conception or design of the work, analysis and interpretation of data. S.M., P.D., J.W., N.V., A.M. and F.J. drafted the work or revised it critically for important intellectual content. S.M., P.D., J.W., N.V., A.M. and F.J. approved the version to be published. S.M., P.D., J.W., N.V., A.M. and F.J. agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding

The Premiers Pas study was funded by the National Research Agency (Agence nationale de la Recherche, AnR) in 2016 (grant number ANR-16-CE36-0008).

Data availability

All the data generated or analysed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

The studies were conducted in accordance with the principles of the Declaration of Helsinki and in compliance with French legislation. The Premiers Pas study involving human participants was reviewed and approved by the Institute for Research and Documentation in Health Economics. Informed consent for participation was obtained from each participant in accordance with national legislation and institutional requirements. The study was conducted under the control of the CNIL (Commission nationale de l'informatique et des libertés, registration number: 2203002 v 0). Participation in the survey was free and anonymous. It was conducted under the supervision of the CNIL, which guarantees respect for anonymity, privacy and the protection of personal data. The European Health Interview Survey is a mandatory survey (binding on questionnaire form and content) for EU member countries (European Commission Regulations n°141/2013 of February 19, 2013 and n°2018/255 of February 19, 2018), steered by Eurostat. The EHIS legal framework is governed by Regulation (EC) No. 1338/2008 of the European Parliament and the Council, dated 16 December 2008, which pertains to Community statistics on public health and health and safety at work.

This study involving human participants was reviewed and approved by the Institute for Research and Documentation in Health Economics. Informed consent for participation was obtained from each participant in accordance with the national legislation and the institutional requirements.

The study was conducted under the control of the CNIL (Commission nationale de l'informatique et des libertés, registration number: 2203002 v 0).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.