Migrant health fades after arrival, but women pay the steepest price

For decades, researchers have puzzled over one of the most counterintuitive findings in population health: migrants often arrive in their new countries healthier than the people already living there. This phenomenon, known as the healthy immigrant effect, has been documented across Europe, North America and beyond, and is usually explained by positive health selection, the idea that only those fit enough to endure the costs and risks of migration actually make the journey. But a growing body of evidence suggests that this advantage is borrowed, not owned. With time spent in the host country, migrants’ health tends to erode, a process researchers have called the exhausted migrant effect. What has remained stubbornly unclear is whether this decline strikes all migrants equally, or whether it is patterned by the same social inequalities that shape health in the rest of society.

A new study by Eleonora Trappolini, Elisa Barbiano di Belgiojoso and Stefania M.L. Rimoldi, published in SSM – Population Health, tackles this question with an unusually direct research design. Drawing on the 2025 MIGHTY Survey of migrants in Lombardy, Italy, the authors asked respondents to rate their health twice: once as it was before they left their country of origin, and once as it stood at the time of interview. This retrospective pre-migration measure, rare in European datasets, allowed the researchers to compare each individual with their own past self, rather than comparing different groups of people who may differ in countless unobserved ways. The analytical sample comprised 2,331 foreign-born residents of Lombardy, roughly half of them women, including undocumented migrants and naturalised citizens who are typically invisible to residence-based statistics.

The methodological core of the study is a correlated random-effects ordinal logistic regression, a statistical framework that separates within-individual change from between-individual differences. Each respondent contributed two observations, one pre-migration and one post-migration, with age treated as a time-varying covariate and supplemented by Mundlak terms, which capture the individual-level means of time-varying variables and allow the model to distinguish ageing from the effect of migration itself. Because the outcome, self-rated health, is measured on a five-point scale from very good to very bad, the authors used ordinal models and, when diagnostic tests revealed that the proportional-odds assumption was violated, they relaxed it, presenting category-specific marginal effects rather than forcing a single summary coefficient onto a process that operates differently at different points of the health distribution.

The headline finding is stark: once ageing is accounted for, post-migration health deterioration is concentrated among women. Among female respondents, the odds of reporting poorer self-rated health after migration were roughly four times higher than before, an association that remained stable across every model specification. Among men, the corresponding coefficient was smaller and not statistically significant. Before migration, men and women reported almost identical probabilities of being in very good health, 38.5 percent and 38.0 percent respectively. After migration, those probabilities fell to 29.1 percent among men and 24.6 percent among women, opening a gender gap where none had existed. The decline was driven primarily by the loss of very good health and its redistribution into the fair category, with women gaining seven percentage points in the fair category compared with four points among men. Movement into the worst health categories remained modest for both groups.

This gendered pattern is consistent with the double burden hypothesis. Migrant women in Italy are disproportionately employed in domestic and care work, sectors characterised by social isolation, emotional labour, irregular schedules, blurred boundaries between work and rest, and weak labour protections. On top of paid employment, many carry disproportionate unpaid and transnational caregiving responsibilities, juggling the roles of worker, mother, partner and breadwinner simultaneously. The authors caution that their survey does not directly measure time use or caregiving, so these mechanisms remain plausible interpretations rather than demonstrated pathways. But the fact that the gender gap emerges only after migration, from a baseline of near-identical health, strongly suggests that it reflects post-migration experiences rather than differential selection into migration.

Occupational status told a more complicated story, and one that differed sharply by gender. Among men, the decline in very good health varied significantly across occupational positions, ranging from under five percentage points among employees in non-physically demanding jobs to nearly twenty points among those in marginal or non-standard work arrangements. Self-employed men in non-physically demanding occupations also showed pronounced deterioration, a finding the authors interpret in light of the Italian context, where self-employment often functions as a fallback for those shut out of standard employment and can entail economic insecurity, long hours and weak social protections. Among women, by contrast, the decline was both larger and remarkably uniform across occupational categories; the interaction between migration timing and occupational status was not statistically significant in this group. This asymmetry, confirmed by a significant three-way interaction between migration timing, gender and occupation, complicates any simple double-burden narrative. One reading is that the burdens migrant women face, combining paid work with unpaid care, cut across occupational positions, elevating deterioration everywhere and leaving little room for occupation to differentiate the decline. For men, health consequences appear more tightly bound to specific working conditions.

The authors are careful about what their design can and cannot establish. Occupational status was measured at interview, after the health change had occurred, so the associations cannot be read causally; migrants whose health has deteriorated may also drift into self-employment or out of employment altogether, a selection process that could partly generate the observed gradients. Unemployed and inactive respondents showed smaller losses in the best health categories but larger increases in poor health, a pattern consistent with reverse causation, whereby illness pushes people out of work. The authors also acknowledge that their occupational classification, which distinguishes physically demanding from non-physically demanding jobs, may understate the physical and psychosocial strain of domestic and care work, the very sectors where migrant women are concentrated, potentially masking heterogeneity among women.

Retrospective self-reports carry their own hazards, and the study confronts them head-on. Recall bias may lead some respondents to idealise their pre-migration health, a phenomenon known as rosy retrospection, while response shift, a change in the internal standards against which people judge their health, may alter how the same objective condition is reported before and after years of life in a new society. Robustness analyses suggest these mechanisms may inflate the magnitude of the estimated decline but do not manufacture it: deterioration is present even among the most recently arrived migrants, the gender difference survives fixed-effects specifications, and a bounding analysis indicates that only a large, gender-specific recall bias could overturn the result. Other limitations include the restriction to Lombardy, the observational equivalence of cumulative exposure and ageing in a two-observation design, and the possibility of selective return migration, the so-called salmon bias, which could bias estimates in either direction.

What the study ultimately delivers is a reframing of migrant health inequality as a dynamic process rather than a static comparison. The probability of reporting bad or very bad health remained low both before and after migration, confirming that migrants are a positively selected population, but the erosion of that advantage is neither uniform nor random. It is steeper for women than for men, and for men it tracks occupational position, with the sharpest declines among those at the margins of the labour market. The authors argue that this stratification calls for differentiated policy: for men, stronger labour protections and occupational health services extended to non-standard and undeclared work; for women, interventions that reach beyond the workplace to address the combined weight of paid and unpaid care, including extending health surveillance into private households and improving physicians’ capacity to identify work-related illness among migrant home-care workers. As Europe’s workforce ages and its dependence on migrant labour deepens, the finding that the price of migration is paid unevenly, and most heavily by women, is one that policymakers cannot afford to ignore.

Subject of Research: Gender and occupational differences in post-migration health decline among migrants in Italy

Article Title: Do all migrants decline equally? Gender and occupational status differences in post-migration health in Italy

Article References: Trappolini, E., Barbiano di Belgiojoso, E., & Rimoldi, S. M. (2026). Do all migrants decline equally? Gender and occupational status differences in post-migration health in Italy. SSM – Population Health, Article 101971. https://doi.org/10.1016/j.ssmph.2026.101971

Image Credits: AI Generated

DOI: 10.1016/j.ssmph.2026.101971

Keywords: migrant health, healthy immigrant effect, exhausted migrant effect, self-rated health, gender inequality, occupational status, Italy, Lombardy, longitudinal analysis, cumulative disadvantage, care work, labour market segmentation