Fragile gains: The impact of refugees and other permit immigrants on national life expectancy in Sweden before, during, and after the COVID-19 pandemic

Highlights

• •Immigrant contributions to Swedish life expectancy (2010–2022) vary by permit status and origin.
• •Refugees contributed positively to Swedish life expectancy before and after the COVID-19.
• •From 2019 to 2020, male refugees lost 3.4 years and females lost 2 years.
• •Immigrants’ positive effect on life expectancy recovered in 2021–2022 and rose in 2023.

Abstract

International migrants are increasingly shaping the population health of high-income countries, yet existing research has treated immigrants as a single, homogenous group when estimating their impact on national life expectancy. Moreover, European research has only studied the impact of immigrants up until 2019—before the COVID-19 pandemic. Using Swedish total population administrative register data from 2010 to 2023, we calculate period life expectancy at age 1 for immigrant subpopulations defined by permit status (refugee; other permit) and for permit status by region of birth (Nordic, Other European, and non-European), applying the Arriaga decomposition to estimate each subgroup's contribution to national life expectancy. From 2010 to 2019, refugees were the primary driver of the positive impact of immigrants on Sweden’s national life expectancy, exhibiting consistently higher life expectancy than both non-migrants and immigrants with other permits. In 2020, COVID-19 struck refugees disproportionately hard. They exhibited life expectancy losses up to four times higher than non-migrants and other permit migrants, and their positive effect on life expectancy was temporarily reversed. The reversal was most pronounced among non-European refugees. Life expectancy recovery among refugees was achieved by 2023 but lagged 2-years behind non-migrants and other permit migrants. Demographic and public health research should routinely disaggregate immigrant populations accordingly. COVID-19 showed that the positive contribution of refugees is fragile. Targeted policies are required to protect refugee populations, particularly during periods of national crisis, if the positive and growing contribution of immigrants to Sweden’s life expectancy is going to be sustained.

1. Introduction

The world today is ‘a world on the move’ (Abubakar et al., 2018). The rising volume, visibility, and diversification of today’s global migration flows is unprecedented (Hossin, 2020). Yet, while migrants are flowing out of an ever widening pool of origin countries for an increasingly disparate number of reasons, they are flowing into a narrowing pool of high-income host countries concentrated in Europe, Oceania, and North America—where they account for approximately 15 to 30% of entire resident populations (United Nations Department of Economic and Social Affairs: Population Division, 2025). This represents a seismic shift in global population dynamics, and international migration is now reshaping populations, societies, national health, and health systems the world over (Wallace et al., 2025).

One way that existing studies have assessed this impact of immigrants is by examining how their life expectancy compares to and influences overall life expectancy in the host country (Hendi and Ho, 2021; Page et al., 2007; Wallace et al., 2022; Wallace and Drefahl, 2024). Period life expectancy is one of the world’s most widely used population health metrics to summarise and compare the mortality situation of countries, forming the basis for a wide range of public health, life insurance, retirement, and broader social policies (Luy et al., 2020). This is because a country's life expectancy reflects, among other things, its existing socio-economic conditions and inequalities and the quality of its public health and health infrastructure (Hendi and Ho, 2021). Variation in life expectancy across subpopulations within a country reflects the uneven distribution of the same conditions that shape national health metrics.

These studies reported a positive influence of immigrants on national life expectancy in Australia, the United States, and most of the Nordic region. Immigrants’ lower mortality risks have helped to elevate national life expectancy, and increasingly so over time (Hendi and Ho, 2021; Page et al., 2007; Wallace et al., 2022). This reflects the well-known “migrant mortality advantage”, the observation that migrants have lower mortality risks compared to non-migrants in the host country owing to factors like positive in-selection in migration and the retention of health-protective cultural norms from the birth country (Abraído-Lanza et al., 2006; Guillot et al., 2018; Jasso et al., 2004). The strength and timing of this advantage are not uniform, however, varying across host countries and over time, as the Swedish case will illustrate. Importantly, these studies only estimate life expectancy for immigrants as a single, homogenous group (i.e., all foreign-born). Aggregating life expectancy across immigrant origins obscures substantial variation by reason for arrival and country of birth, risking the concealment of key health inequalities relative to non-migrants and varying impacts on wider population health relative to other immigrant groups that disaggregated analyses would otherwise reveal.

Understanding how migrants belonging to different resident permit and/or country of birth groups might differentially influence national population health metrics is crucial to understanding future developments and improvements in national life expectancy. We know that the mortality risks vary by reason for arrival. Refugees encounter specific risks (e.g., elevated mortality risks from cardiovascular, communicable diseases, and mental disorders) and tend to exhibit weaker mortality advantages than migrants who arrive for other reasons (e.g., work) (DesMeules et al., 2005; Hollander et al., 2012; Norredam et al., 2012). We also know that a migrant mortality advantage does not present universally across all origins; it is strongest and most consistent among people born in low and middle-income countries that move to high-income countries and weaker (and often absent) among those migrants that move between high-income countries (Shor and Roelfs, 2021). These compositional differences take on particular weight where the aggregate migrant advantage emerged late, or was preceded by a period of disadvantage: such trajectories suggest that shifts in the mix of permits and origins, rather than any uniform advantage, may be driving the national-level effect in ways that averages cannot reveal.

Moreover, the above studies that included refugees used data that ended on or before 2019. The evidence tells us that immigrants, who are often sidelined in pandemic response and preparedness plans, were disproportionately affected by mortality from COVID-19, particularly if they were born in low and middle-income countries and/or arrived as refugees (Hintermeier et al., 2024). It’s now been concluded that COVID-19 pandemic only served to exacerbate pre-pandemic health inequities and heighten the vulnerability of immigrants compared to non-migrants (Mengesha et al., 2022). However, while studies have quantified the impact of COVID-19 on general populations using changes in life expectancy pre-, peri‑and post-pandemic (Aburto et al., 2022a; Schöley et al., 2022) or on specific ethnic groups (Aburto et al., 2022b; Aldridge et al., 2020), no study has done this by nativity status and permit status. Thus, it is relevant to understand how the life expectancy of specific immigrant origin groups and permits groups, and their contribution to national life expectancy, was impacted by COVID-19, and how their life expectancy developed in the years after the pandemic.

Here, we aim to progress beyond averages and explore variation in the life expectancy and effects on national life expectancy of immigrants by permit type and permit type by region of birth in Sweden before (2010–2019), during (2020–22), and after (2023) the COVID-19 pandemic.

We ask three research questions to help us achieve our aim:

(RQ1) how does the male and female life expectancy of immigrants vary according to permit status compared to non-migrants in Sweden before the onset of COVID-19 and how does this variation affect the contributions of these groups to wider life expectancy?

(RQ2) does understanding of the magnitude and direction of effects reported in RQ1 change when further disaggregating permit status into immigrants’ regions of birth?

(RQ3) what kind of impact did the COVID-19 pandemic have on the life expectancy of refugee and other permit immigrants and their contributions to wider life expectancy, disaggregated by both permit type and region of birth?

We do this study in Sweden because it is a country with one of the largest relative numbers of immigrants in Europe (1 in 5 residents in 2025) that is renowned globally for accepting substantial numbers of refugees from both within and outside Europe relative to its total population size (Karlsdottir et al., 2018). Additionally, a positive effect of immigrants on national life expectancy emerged comparatively later in Sweden (in 2010 among women and 2016 among men) (Wallace and Drefahl, 2024) than it did in Australia, the United States, Denmark, Finland and Norway—where immigrants have been exerting a positive impact since the early 1990s. In Sweden in the 1990s and 2000s, immigrants negatively affected national life expectancy—and the life expectancy of immigrants was lower than that of the native-born population (Wallace and Drefahl, 2024). This late and initially negative trajectory makes Sweden a critical case for our questions: it implies that the aggregate advantage observed elsewhere may here be the product of a shifting compositional mix rather than a uniform migrant advantage. One has to wonder what role refugees—an especially vulnerable, excluded population when it comes to health—might play. Sweden is also one of only a few countries in the world that provides large enough data (on its entire registered resident population) and rich enough data (on both the country of birth and permit type) data to be able to conduct this kind of study.

1.1. Migrant mortality around the world

The Migrant Mortality Advantage (MMA) refers to migrants’ lower mortality rates and higher life expectancy than natives (Guillot et al., 2018; Juárez et al., 2019). Two reviews and meta-analyses reported globally lower mortality among migrant men and women compared to non-migrant men and women (Aldridge et al., 2018; Shor and Roelfs, 2021), including for all major cause-of-death categories except communicable diseases (Aldridge et al., 2018). These global estimates mask heterogeneity in immigrants’ mortality risks. The MMA has been observed among immigrants from Northern Africa, South America, Asia, and Southern Europe to high-income countries, while little evidence of a mortality advantage—and even some evidence of a mortality disadvantage—in other destinations, notably among immigrants moving between high-income countries (Shor and Roelfs, 2021). Other studies report substantial variation by age (Guillot et al., 2018; Wallace and Wilson, 2022), length of stay (Juárez et al., 2018; Wallace et al., 2019), and age-at-arrival (Wallace, 2022).

However, as Aldridge and colleagues underline in their recent review, few studies have reported the mortality for migrants according to reason for arrival (Aldridge et al., 2018). A study focusing on Canada between 1990 and 1998 found that refugees had lower mortality than native-born but higher than other immigrants (DesMeules et al., 2005). Two Danish and Swedish studies found similar results (Hollander et al., 2012; Norredam et al., 2012). Refugees are considered more at risk than other migrants for adverse health outcomes and a higher mortality risk, since the entire migration process—including leaving the country, travelling, and requesting asylum in the new destination country— is associated with very stressful and risky circumstances (Juárez et al., 2019; Oostrum et al., 2011).

1.2. Impact of immigrants on national life expectancy

Few studies have estimated the effect of international migrant mortality on national population health metrics, such as life expectancy. The first study was conducted in Australia, where migrants increasingly enhanced national life expectancy at birth up to 0.4 years among women and 0.6 years among men (Page et al., 2007). In the U.S., a surge in life expectancy at birth in New York between 1990 and 2010, compared to the U.S. national average, was attributed to a higher share of international migrants living in New York (37.8%) compared to the U.S. national average (13.6%) (Preston and Elo, 2014). Later, Mehta et al., 2016 estimated a positive impact of older migrants on life expectancy at age 65+ in the U.S. of 0.2 years among men and 0.2 years among women. The most recent U.S. study found that migrants increased life expectancy at age 1 between 1990 and 2017, when migrants’ contribution rose from 0.3 years to 0.9 years among men and from 0.3 years to 0.8 years among women (Hendi and Ho, 2021). In Europe, a comparative study of the Nordic countries’ region found a growing effect of migrants on life expectancy at age 1 in Denmark, Finland and Norway from 1990 to 2019. By 2019, migrants enhanced national life expectancy at age 1 by 0.1 to 0.2 years in all three countries (Wallace et al., 2022). In the same study, the authors found a negative impact of migrants on national life expectancy at age 1 in Sweden starting in 1990. But this effect gradually diminished over time and a negligible positive effect in the 2010s, reflecting a faster increase in longevity among migrants compared to non-migrants in the decades between (Wallace et al., 2022). A follow-up research focusing on Sweden found that this transformation in the effect on life expectancy at age 1 was due to changes in the composition of Sweden's migrant population from predominantly (low life expectancy) Nordic migrants to (high life expectancy) non-Western migrants at young adult and mid-adult ages (Wallace and Drefahl, 2024).

1.3. COVID-19 and its impact on the mortality and life expectancy of immigrants

The COVID-19 pandemic hit the world in 2020, and life expectancy declined across many countries (Aburto et al., 2022a). Despite the widely reported MMA, the pandemic negatively affected immigrant mortality in Western countries, reducing or even reversing this phenomenon. In the U.S.A., life expectancy decreased for the entire population, with larger losses among Black and Latino Americans than among White Americans, reducing the usual advantage in mortality for the Hispanic population (Andrasfay and Goldman, 2021). The loss was heterogeneous among Hispanic subgroups (Arias and Tejada-Vera, 2023). Drawing on the full 2020–2022 period, Paglino and Elo (2025) show that life expectancy in 2020 fell more steeply among every foreign-born subgroup than among their US-born counterparts, narrowing the foreign-born advantage by 1.0 year among women and 2.4 years among men; even so, because the foreign-born retained lower mortality, overall US life expectancy would have declined further still in their absence. This advantage then recovered unevenly, with the foreign-born contribution to national life expectancy returning to (women) or remaining below (men) its pre-pandemic size by 2022.

In Europe, several studies demonstrated high mortality increases among migrants during COVID-19. In Italy, an early study found the MMA vanished completely (Canevelli et al., 2020). In France, non-Western migrants, particularly from Africa, experienced a disproportionate mortality burden (Khlat et al., 2022). In Belgium, migrants experienced increased overall mortality, with only Sub-Saharan Africans experiencing excess mortality above that of natives (Vanthomme et al., 2021). In Spain, higher mortality risks were reported for migrants coming from South America, Sub-Saharan Africa and Asia (Aldea, 2022).

1.4. The swedish context

Sweden is a country based in Northern Europe with a socio-democratic welfare state and an integration policy of inclusive multiculturalism. Sweden's migration history began with European refugees fleeing the Second World War, followed by labour migration (chiefly from Finland, but also from Greece, Turkey and Yugoslavia) through the 1950s and 60 s. The 1970s brought family reunification for settled labour migrants, alongside the start of a sustained asylum-seeker inflow that would continue for many decades, from Chile and Lebanon initially, then Iran, Iraq and Eritrea in the 1980s, former Yugoslavia, Somalia and Ethiopia in the 1990s, and Syria in the 2010s. Since the 2000s, EU free movement has added a further layer of intra-European migration for work and study (Bevelander et al., 2013; Borevi, 2014; Holloway et al., 2021). Finnish immigrants have consistently been found to exhibit high mortality compared to native-born Swedes; immigrants from other Nordic, Eastern European countries and former Yugoslavia typically exhibit higher mortality risks too. In contrast, immigrants born in Western and Southern Europe, Iran, Iraq, Turkey, Central and South America, and Asia consistently report lower mortality than native-born (Honkaniemi et al., 2017; Juárez et al., 2018; Oksuzyan et al., 2019; Wallace, 2022, 2026). While studies in Sweden have shown that refugees exhibited higher mortality from all-cause and certain causes (like cardiovascular diseases) relative to non-refugee migrants from earlier data (Hollander et al., 2012; Thordardottir et al., 2020), studies based on more recent data show comparably lower mortality risks among refugee and non-refugee migrants relative to native-born Swedes (Helgesson et al., 2022; Stratmann et al., 2022). Immigrants born in low and middle-income countries were repeatedly found to have been disproportionately impacted by mortality from COVID-19 (Drefahl et al., 2020; Mussino et al., 2024; Rostila et al., 2021, 2023). Concerning the impacts of immigrants as a single aggregated population on Sweden’s life expectancy, studies reported an initial negative impact in 1990 that gradually diminished and then reversed over time in the early 2010s (Wallace et al., 2022).

We have identified three important gaps in the existing literature examining the impact of international migrants on national life expectancy that we directly address in this study: the unexplored role of permit status, the intersection with permit status with region of birth, and the effect that COVID-19 had on the contributions of immigrants to national life expectancy.

2. Materials and methods

2.1. Data

We used data from the research project REFU-GEN, which is anonymised and provided by Statistics Sweden for research purposes. The project is coordinated by Stockholm University. Data is accessible for research under ethical approval from the Swedish Ethical Review Authority (no 2025–04,797-02). We combined data from different administrative registers: the total population register, death register, international migration register, and residence permit register. From these registers, we took information on gender, date of birth, date of death, date of arrival in Sweden, country of birth, type of residence permit and date of issue of the permit. The residence permits register records all permits issued to the same person. We took a stock approach to refugee status, defining immigrants as refugees in a given year if their residence permit value for that year indicated that they were refugees. Thus, we calculate life expectancy based on refugee status in a given year. All immigrants born in European Union (EU) countries are coded as non-refugees as they can move to Sweden without a residence permit due to freedom of movement within the EU.

For our analysis, we first classified individuals according to resident permit into born in Sweden (no permit required), holds a refugee permit, and holds any other permit (e.g., labour, education, family reunification). We then further subdivided residence permit by immigrants’ region of origin. We defined the following regions based on their country of birth: Sweden, Nordic (people born in Norway, Denmark, Finland and Iceland), Other European (those born in Europe outside of the Nordic region) and non-European (people who were born in the rest of the world). The total Swedish resident population in 2010 was 9505,982, and at the end of the period observed in 2023, the resident population was 10, 646,059.

2.2. Method

To compare immigrants and native Swedes, we estimated period life expectancy (PLE), defined as the average number of additional years a person would live if they experienced age-specific mortality rates for the rest of their life (Wallace et al., 2022). PLE is widely used to assess and compare mortality rates across separate subpopulations. Because our focus is the immigrant population, we calculated the PLE from age 1 (PLE1). Mortality within the first year is concentrated in the initial days of life, and using life expectancy at birth would unrealistically assume immigration immediately after birth (Hendi and Ho, 2021). Wallace & Drefahl (2024) provides a lengthy discussion of and rationale behind this choice in Box S2 of the online supplementary materials of their article.

To compute PLE1, we created mid-year estimates and age-specific death rates for each year (2010–2023), gender, and age (1 - 95+) across different subgroups. We used mid-year estimates, rather than person-years-at-risk, to maximise comparability of our findings with two previous studies within which we want to contextualise and discuss our results (Wallace et al., 2022; Wallace and Drefahl, 2024). We calculated the 95% bootstrap confidence intervals (Andreev and Shkolnikov, 2010).

We calculated the PLE1 of the total Swedish resident population, separately for men and women. Then, we recalculated the PLE1 for the total Swedish resident population, excluding one specific immigrant group, and assessed the group’s contribution to overall PLE1 (Hendi and Ho, 2021). We then calculated the Arriaga decomposition using these two partial life expectancies to estimate the effect of the specific migrant population on the total Swedish resident population's partial life expectancy (Arriaga, 1984). The decomposition allows us to control for age effects (Auger et al., 2014).

First, we calculated the decomposition based solely on permit status to separate the contributions of refugees and other-permit immigrants. Afterwards, we calculated contributions by permit status by region of origin. Finally, we calculate the contribution for Other European refugees, Other European other-permit immigrants, non-European refugees and non-European other-permit immigrants. The Nordics are the only group we define as solely for other-permit immigrants because the refugee population is incredibly low.

We used R to calculate the PLE1 (particularly the library DemoTools (Riffe et al., 2019)) and the Arriaga decomposition.

3. Results

3.1. Description of the immigrant population in sweden

We present the share of immigrants by age and permit type (Fig. 1) and the percentage of refugees by origin region (Fig. 2). As sex differences were minimal, males and females were combined.

Fig. 1.

Share of Swedish-born, immigrants with refugee permits, and immigrants with other-permit types in the Swedish population, 2010 to 2023, by age class.

Fig. 2.

Share of refugees and other- permit migrants with other-permit types, divided by origin area (Nordic, non-European, Other European) from 2010 to 2023.

Other-permit immigrants grew from 9.2% of the Swedish population in 2010 to 12.3% in 2023; refugees rose from 4.3% in 2010 to 7.1% in 2021 (Supplementary Materials, Figure S.1). Fig. 1 presents population shares in Sweden from 2010 to 2023 by age and permit type for the immigrants. Among children aged 1–14, migrant shares increased slightly, peaking after 2016 before declining toward the end of the period. In the 15–34 and 35–64 age groups, both refugee and other-permit populations increased steadily. Among those aged 65+, migrants accounted 10% of the population, almost entirely other-permit migrants; refugees increased from 2% in 2010 to over 3% in 2023.

Fig. 2 shows the percentage of refugees for each immigrant origin group. There are no refugees in the Nordic group, and the percentage of residents born in the Nordic countries decreased from 2010 to 2023. Conversely, the number of immigrants from other European countries increased over the period (Supplementary Materials, Figure S.1). In this group, the refugee share fell from 24.6% to 16.3% between 2010 and 2023. Over half of non-European migrants held refugee permits in 1990. The percentage of non-European refugees rose between 2010 and 2023, peaking at 60.7% in 2016–2017, declining to 55.1% after.

3.2. Life expectancy from 2010 to 2023

Fig. 3 reports male and female PLE1 with 95% confidence intervals for the Swedish-born population, refugees, and other-permit immigrants between 2010 and 2023 (numeric values are presented in Supplementary Materials, Table S.1). Refugee men have consistently higher PLE1 than both other-permit migrants and non-migrants between 2012 and 2019 (refugees’ PLE1 = 81.9 years in 2019, compared with 80.0 and 80.6, respectively), but the difference becomes significant based on the confidence intervals only in 2014. In 2020, male refugees’ PLE1 is disproportionately affected by the COVID-19 pandemic, falling by 3.4 years (PLE1 = 78.5) compared to 1.2 years among other-permit migrants and 0.6 years among natives relative to 2019. This is the lowest PLE1 value for refugee men across the entire period. Their PLE1 is significantly lower than that of the Swedish-born, while the confidence intervals overlap for refugees and other-permit immigrants. In 2021, there is a slow recovery, and male refugees’ PLE1 is 79.9 years, similar to non-refugee men. In 2022, the PLE1 of refugee men is again significantly higher than that of non-migrants and other-permit migrant men. In 2023, their PLE1 reaches the maximum for the period of analysis: 82.6.

Fig. 3.

Life expectancy at age 1 with 95% confidence intervals, from 2010 to 2023 for Swedish natives, refugee migrants and other permit migrants, divided by sex. The black line represents the refugees PLE1 in 2020.

Male other-permit immigrants consistently exhibit lower PLE1 than non-migrants, but experienced steady gains over time, rising from 78.1 in 2010 to 80.6 in 2023. This group is affected by the COVID-19 pandemic, but to a much lesser extent than refugee men; there was a decrease in 2020 (PLE1: 79.2 years), but already a recovery in 2021 (PLE1: 79.6 years), returning to a similar level as 2018. In 2022 and 2023, their values are very close to those of the Swedish-born and they are not significantly different.

Among women, refugees have consistently higher PLE1 compared to other-permit immigrants and non-migrants from 2012 to 2019. The values are significantly different for all years except for 2012 and 2018. Their PLE1 is 85.0 in 2019, but they are severely impacted by COVID-19, and their PLE1 drops below the other groups' levels in 2020 (female refugees PLE1 = 83.0 years). This compares to a drop of just 0.5 years both for other-permit migrant women (PLE1 in 2019 = 83.4 years and PLE1 in 2020 = 83.9 years) and non-migrant women (PLE1 in 2019 = 83.5 years and PLE1 in 2020 = 84.0 years). The female refugees’ PLE1 value of 2020 is similar to that of 2011, while their PLE1 in 2022 (84.1) surpasses the other groups, even if it is not significantly so, and in 2023 reaches the maximum of the analysis period, 85.1, being significantly higher than the other groups. Female other-permit migrants have a PLE1 similar to non-migrants from 2010 to 2023. From 2020 to 2022, the other-permit PLE1 is slightly lower than that of Swedish natives, but not actually significantly different. In 2023, the PLE1s of these two groups are very close (other-permit migrants PLE1:84.2, Swedish natives PLE1: 84.1).

3.3. Decomposition

Fig. 4 reports the contributions to the Swedish population's PLE1 of refugees and other-permit migrant between 2010 and 2023 (numeric values are presented in Supplementary Materials, Table S.2). For male PLE1 until 2019, refugees’ contribution has always been positive with a maximum of 0.058 in 2019 and a minimum of 0.016 in 2011 and 2012. In 2020, when the COVID-19 pandemic hit refugees especially hard, the male refugees’ contribution becomes negative (−0.041), and it remains negative in 2021 (−0.035). In 2022 and 2023, it becomes positive again, returning to pre-pandemic levels (2023: 0.054). The other-permit immigrants’ contribution is consistently negative, with a maximum negative contribution in 2010 (−0.074), but, in 2022, the other-permit immigrant group has a small positive contribution (0.001), increasing further in 2023 (0.011). The male refugees’ contribution is sufficiently high to counteract the negative impact of other-permit migrants, resulting in an overall positive male migrant contribution from 2014 to 2019 and again in 2022–2023.

Fig. 4.

Contributions to the Swedish population's PLE1 of refugee permit and other-permit immigrants from 2010 to 2023.

The situation differs for the female population. The female refugees’ contribution has always been positive, except for 2020 (−0.007). In 2021, it is smaller than before the pandemic but positive (0.023) and reaches the maximum over the period considered in 2023 (0.059). Other-permit female immigrants contribute positively to the general population's PLE1 in 2010, 2012, 2017, 2018, and, after the pandemic, in 2023. They contribute negatively to the other years. Female refugees almost always make a higher absolute contribution to national PLE1, making the overall female migrant contribution positive, except in 2011 and 2020.

Fig. 5 shows the contributions of immigrants grouped by origin regions and permit status to the Swedish population's PLE1 between 2010 and 2023 (numeric values are presented in Supplementary Materials, Table S.3). Among men, non-European other-permit migrants always contribute positively, with a pre-pandemic maximum in 2010 (0.063). In 2020 and 2021, the contributions decline slightly but remain positive, increasing again in 2022 and 2023, reaching a new maximum of 0.084. Non-European refugees also contribute positively from 2010 to 2019, but it turns negative during COVID-19 (2020: −0.02, 2021: −0.008). In 2022 and 2023, the contributions are again positive and similar to the pre-pandemic size. The divergent COVID-19 patterns between these two groups might highlight the heightened vulnerability of refugees in this period. At the same time, it might also reflect compositional differences in origin countries within this broad non-European group.

Fig. 5.

Contributions to the Swedish PLE1 of refugees and other-permit immigrants, by origin area, from 2010 to 2023.

Other European other-permit immigrants positively contribute to PLE1 from 2010 to 2019, though their contributions are always smaller than those of non-European immigrants. In 2020, the contribution is minimal at 0.001; in 2021, it is negative (PLE1: −0.012), and positive again in 2022–2023, similar to pre-pandemic size. Other European refugees always make a small negative contribution to the PLE1. While this might reflect a heightened vulnerability of refugees during COVID-19, it may also reflect sizeable compositional differences in origin countries between the two groups. Nordic immigrants’ contribution is consistently negative. Among the female population, non-European other-permit immigrants, as for males, always contribute positively to the PLE1, with a maximum in 2020 (0.044). Female non-European refugees also contribute positively throughout, even if the contribution is small in 2020 (PLE1: 0.05), but already increases in 2021 (PLE1: 0.032), and returns to pre-pandemic levels. Other-permit Other European immigrants contribute positively from 2011 to 2021, with a maximum in 2012 (0.056). The contribution stays stable in 2020 but is quite reduced for 2021 (PLE1: 0.011). In 2022 and 2023, the contributions return to pre-pandemic levels. Other European refugees’ contribution is small. For most of the period, it is negative, except for 2015, 2017 and 2023. In 2020, the contribution is the largest and negative (PLE1: −0.012). Female Nordic immigrants' contribution is consistently negative, with a maximum of −0.096 in both 2014 and 2021.

4. Discussion

This study has explored the impact of immigrants’ region of birth and permit status on Swedish national life expectancy between 2010 and 2023. Our work builds upon a small number of existing papers that have investigated the impact of immigrants on national life expectancy in several Western countries, including Australia (Page et al., 2007), Denmark, Finland, Norway, and Sweden (Wallace and Drefahl, 2024), and the U.S.A. (Hendi and Ho, 2021) in two main ways. First, we investigated variation in life expectancy and impacts on national life expectancy by immigrants’ permit status and their permit status by region of birth—extending a body of work that has so far conceptualised the impact of immigrants on national life expectancy as one single, homogenous population (i.e., all foreign-born). Second, we extended the evidence beyond 2019 to 2023 to incorporate the COVID-19 pandemic and so help to understand how the life expectancy of refugee and other permit migrants and their broader effects on national life expectancy were impacted during the pandemic.

Our first research question was “How does the male and female life expectancy of immigrants vary according to permit status compared to non-migrants in Sweden before the onset of COVID-19 and how does this variation affect the contributions of these groups to wider life expectancy?”. Male and female refugees had higher life expectancy than Swedish natives and migrants holding other permits between 2010 and 2019, except in 2012 (Fig. 3), and these two groups made a positive contribution to national life expectancy during this period (Fig. 4). Migrant women that held other permits had comparable life expectancy to Swedish natives between 2010 and 2019; male migrants that held other permits had consistently lower life expectancy than natives in the same period (Fig. 3) and had a small negative effect on life expectancy in Sweden (Fig. 4). In the context of previous results that show that refugees have lower mortality than natives (Aldridge et al., 2018; DesMeules et al., 2005; Norredam et al., 2012), but higher mortality than other permit types (e.g., labour, family reunification), our results reveal that refugees had a larger mortality advantage than both migrants that held other permits and natives and played a defining role in the emergence of an overall positive influence of immigrants on life expectancy in Sweden.

Our second research question was “Does understanding of the magnitude and direction of effects reported in RQ1 change when further disaggregating permit status into immigrants’ regions of birth?” When we decomposed the impact of refugees by their birth region, we reported differences between the stable positive impact of non-European refugees on national life expectancy between 2010 and 2019 and stable negative impact of Other European refugees. In contrast, Other European and non-European migrants that held other permits reported a positive impact on wider life expectancy (Fig. 5). Intra-Nordic migrants made the largest, most consistent negative contribution to Sweden’s life expectancy (Fig. 5). While are unable to formally test drivers of the contributions and/or disaggregate further into more specific countries of birth, the literature shows some important context worth considering for future work. A higher mortality risk among intra-Nordic (and particularly Finnish) migrants has been well-established in the Swedish literature (Juárez et al., 2018; Lundgren et al., 2019; Östergren et al., 2023), owing to factors like health behaviours (higher rates of smoking and drinking) and the initial large-scale negative in-selection of disadvantaged, unemployed, blue-collar Finns into Sweden in the 1950s and 1960s (Hjern and Allebeck, 2002, Hjern and Allebeck, 2004; Tollosa et al., 2024; Wahlbeck, 2015) following the Nordic Common Labour Market Agreement in 1954 (Pedersen et al., 2008). A large majority of Other European refugees (>80%; authors’ calculations based on the Swedish registers) were born in Former Yugoslavia—a population that has also been shown to exhibit raised all-cause mortality risks relative to natives in Sweden driven by elevated cardiovascular disease mortality and external causes (Hollander, 2013; Hollander et al., 2012; Oksuzyan et al., 2019; Thordardottir et al., 2020). Interestingly, non-European immigrants contribute positively irrespective of permit status. This may point more towards factors associated with birth region. For example, health-promoting behavioural norms learned the country of birth, the protective impact of denser social and familiar networks in Sweden, and stronger health selection effects owing to the increased migration barriers (Aldea et al., 2023; Jasso et al., 2004; Roelfs and Shor, 2022). As stated, however, given we cannot disaggregate region of birth and/or test specific explanations, these studies should be treated as offering possible potential explanations, rather than confirmed explanations for the descriptive trends that we found here.

Our third research question was “what kind of impact did the COVID-19 pandemic have on the life expectancy of refugee and other permit immigrants and their contributions to wider life expectancy, disaggregated by both permit type and region of birth?” Between 2019 and 2020 all subpopulations experienced decreases in their life expectancy. However, the size of the loss was greatest among refugees. Compared to 2019, male refugees lost 3.4 years of life expectancy at age one (returning to levels not seen since 2012), while female refugees lost 2.0 years (returning to levels not seen since 2015). This contrasted with Swedish natives and migrants who held other permits, who lost around 0.5 to 0.75 years (Fig. 3). However, 2019 was an especially positive year for life expectancy gains in Sweden, particularly among refugees who gained 1.0 and 1.5 years between 2018 and 2019 respectively among women and men. If we instead compare the results of 2020 with 2018, whose life expectancy estimates are more in line with the previous year (i.e., 2017), losses are smaller across all subpopulations, but still greatest among refugee women (−1.0 year) and men (−1.6 years) (Fig. 3). In light of these losses, refugees contributed negatively to national life expectancy levels in Sweden in 2019—particularly men (Fig. 4.). From 2021 onward, we reported a recovery across all subpopulations, with life expectancy in 2023 recovering to or surpassing 2019 levels (Fig. 3). At the same time, it is important to highlight that while life expectancy among natives and migrants who held other permits had recovered within a single year (i.e., by 2021), the recovery of refugees took longer—men only returned to pre-pandemic levels in 2022 (2-years later; and resultantly continued contributing negatively to national life expectancy in 2021; Fig. 4); women returned to pre-pandemic levels in 2023 (3-years later). Our results are consistent with prior international and Swedish research that demonstrated a disproportionate impact of the pandemic on the mortality of immigrants as compared to natives in high-income countries (Andersson et al., 2021; Drefahl et al., 2020; Hiam et al., 2024; Hintermeier et al., 2024; Mussino et al., 2024).

When decomposing the effect of the COVID-19-related losses by permit status and region of origin we identified group-specific effects. Several groups maintained a status quo, with no real change in contribution in 2020 and 2021 compared to 2010–19. Specifically, Other European refugees and intra-Nordic migrants maintained a negative contribution and non-European migrants with other permits maintained a positive contribution to national life expectancy. Conversely, the contribution of refugees born in non-European countries reversed from positive to negative (especially among men) in 2020 and 2021 (Fig. 5). This heterogeneity outlines the uneven impact of the pandemic. It is fascinating in so much as the existing literature (which has focused on country of birth) reported a disproportionate impact of COVID-19 on the mortality risks of immigrants—particularly if they were born in non-European countries (Andersson et al., 2021; Drefahl et al., 2020; Hiam et al., 2024; Hintermeier et al., 2024; Mazzalai et al., 2023; Mussino et al., 2024; Paglino and Elo, 2024). Here, however, our results show that this effect was limited to refugees born in non-European countries. Their heightened vulnerability to mortality from COVID-19 might reflect pre-existing health vulnerabilities specific to refugee journeys that render some refugees more vulnerable to the health impacts of COVID-19, barriers to accessing healthcare, differences in health literacy, language proficiency and institutional trust that influenced abilities to follow public health guidelines, and social inequality (e.g., lower quality housing, over-crowded housing, over-representation in high-risk, high-exposure jobs that might have prevented the ability to minimize contact, isolate, or work from home) (Hintermeier et al., 2024).

This study has many strengths. It is only possible due to the availability of comprehensive, high-quality total population administrative register data from Sweden that can link longitudinal information on age, sex, region of birth, and permit status with mortality over time. In many countries of the world, the calculation of life expectancy by permit status (specifically among refugees) is simply not possible. What’s more, we have also examined variations in life expectancy, and how the impact of these variations differ at national level, according to permit status and permit status by region of origin. This represents a major contribution to a body of work that has so far treated immigrants as a single, homogenous group. To the best of our knowledge, our study is also one of the first to quantify the actual impact of COVID-19 on the mortality of immigrants using life expectancy in Europe. In the context of COVID-19, life expectancy provides a timely description of current mortality patterns, is age-standardised (permitting comparisons across subgroups), enables us to compare the cumulative impact of the pandemic against mortality trends leading up to the pandemic, and is routinely monitored/used by national and global agencies to capture differences in mortality.

Nevertheless, although we have benefitted from the use of total population administrative data to answer our research questions—and the data itself exhibits many strengths—there are still a range of limitations associated it that we must identify and outline to the readers. First, since the immigrants are classified by permit type of the year in question they may belong to different groups in different years. However, we calculated annual changes that are always <0.20% of the migrant population. Second, some immigrants do not have registered permits. For certain countries, such as other Nordic countries or countries within the European Union, that can move freely to Sweden, this merely reflects Sweden's visa system (i.e., they do not have to apply for a permit and therefore cannot have a permit type). For migrants born in these countries, we attribute a non-refugee permit. However, some very old refugees did arrive from parts of Europe during the Second World War. This share of older migrants is likely to be very small, and it is unlikely that they would still hold refugee status in Sweden during the period we investigate. We could not make this assumption for non-European migrants with missing permit data. The highest number of non-registered permits in this group was in 2010, when almost 13% of non-European migrants lacked a registered permit. Third, while we would have liked (and did attempt) to move beyond the broad intra-Nordic, other European and non-European categories to more granular countries of birth, the population sizes and death counts became too small to reliably estimate life expectancies. Fourth, the registers only have data on the registered population. A report by the European Migration Network and Swedish Migration Agency estimates that there are currently between 20,000 and 50,000 unregistered migrants living in Sweden (Santesson, 2022). A substantial (and inestimable) share of them are those who had asylum claims rejected and are "would-be" refugees. These are likely to be a high-risk group, prone to exploitation and lacking access to healthcare in Sweden. We wonder how our refugee-specific estimates would be influenced by the inclusion of this vulnerable sub-group.

This study makes a unique contribution to the growing body of work quantifying the impact of international migrants on national life expectancy in high-income countries, and to migrant health research more widely. Our findings demonstrate that it is non-European refugees, specifically, who were the principal driver of the positive impact of immigrants on Swedish national life expectancy that emerged in the 2010s onwards, bringing Sweden more in line with Australia, the United States, and the other Nordic countries where a positive effect has been established for longer. Simultaneously, our results expose the fragility of this contribution and the heightened vulnerability of the subpopulation driving it. COVID-19 struck refugee migrants disproportionately hard. The pandemic obliterated the positive impact of all immigrants combined to national life expectancy that had been gradually building for a decade (Wallace and Drefahl, 2024). Crucially, the pandemic's impact was not uniform across all immigrant groups: it was non-European refugees, not non-European other-permit migrants, who’s contribution to national life expectancy reversed during peak pandemic years, a finding that would have been invisible had we, like previous studies, treated immigrants as a single, homogenous population. Although refugee life expectancy has since recovered to or beyond pre-pandemic levels by 2023, this recovery took one to two years longer than among non-refugees and non-migrants. These findings have implications for social scientists and public health researchers, who should routinely disaggregate immigrant populations by permit status and the region of origin to avoid concealing health inequalities. They also present an important message to policy makers, not least the Swedish government and Public Health Agency of Sweden, that targeted social and public health interventions are needed to protect refugee populations, both in ordinary times (to help maintain their mortality advantage) and during periods of national crisis (to help prevent a disproportionate mortality burden), if the positive and growing contribution of immigrants to Swedish national life expectancy levels is to be sustained.

Funding

This work was funded by the UKRI HEu Guarantee: ERC Starting Grants 2023: APP44827 ‘Living longer in poorer health? Understanding the Immigrant Morbidity Mortality Paradox’ (PARA-MOR). This research has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research andinnovation programme (grant agreement No 948727).

CRediT authorship contribution statement

Daniela Foresta: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Formal analysis, Conceptualization. Matthew Wallace: Writing – review & editing, Writing – original draft, Supervision, Methodology, Conceptualization. Eleonora Mussino: Writing – review & editing, Writing – original draft, Supervision, Methodology, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.