The Migrant Mortality Penalty in a Pandemic: Disparities in COVID-19 Mortality Among Foreign Residents in Switzerland, 2020

Abstract

The COVID-19 pandemic has widened the gap in mortality between different population groups. While socioeconomic status has been shown to be an important determinant of mortality, the relationship between migration status and mortality risk remains unclear. The objective of this paper is to explain whether and why foreign populations had a higher risk of mortality than Swiss nationals during the initial outbreak of the COVID-19 pandemic in Switzerland. We use original linked data to measure the risk of COVID-19 mortality for different groups of foreigners, compared to Swiss nationals, using logistic regression. We find that the risk of death from COVID-19 in 2020 was significantly higher for some groups of foreign nationals—but not all groups—compared to Swiss nationals. Increased mortality is observed among foreign communities who have been living in Switzerland for more than 20 years, notably, Italians, people of Balkan origin, and Latin Americans. For these three communities, we suggest that high mortality is due to a combination of overexposure, for socioeconomic reasons, and reduced access to health systems. These findings contribute to the understanding of migrant health disparities during pandemics and inform future public health interventions.

1. Introduction

In most European countries, foreign populations generally have lower mortality rates than individuals who hold the nationality of the country where they live. In the literature, this phenomenon is referred to as the ‘migrant mortality advantage’. It is attributed to three main reasons: first, the ‘healthy migrant effect’, which suggests that immigrants arrive in the country in a better health condition than those of native-born populations [1,2,3]; second ‘out-migration selection’, the tendency of foreign-born individuals to return to their country of nationality upon falling seriously ill [4,5]; and third, ‘cultural effects’, wherein migrant populations maintain lifestyle habits from their countries of origin that are often healthier of those of populations in the countries where they live [6,7]. While such a ‘migrant mortality advantage’ has been well documented, we know little about its resilience in the context of a public health emergency, namely, what happens, for example, in the face of a pandemic.

This is important because existing studies show that the COVID-19 pandemic is associated with increased differentials in mortality between social groups, with some individuals being more exposed than others and experiencing greater difficulties in accessing the health system [8,9,10]. While socioeconomic status has been found to be an important determinant of mortality during the pandemic, findings on the association between migration status and mortality risk remain only partially conclusive, and call for greater research. The existing evidence suggests that the COVID-19 pandemic has impacted pre-existing mortality differentials between foreign and national populations [11,12,13,14,15,16], but the sizes of these differentials and their distributions across age groups, nationalities, and gender remain unclear.

We examine the case of Switzerland to understand whether foreign populations had a higher risk of death than Swiss nationals during the initial outbreak of the COVID-19 pandemic and identify the reasons that may explain these differentials in mortality. We use original linked data from the Swiss Statistical Office to measure the mortality risks for different groups of foreign individuals, compared to Swiss nationals.

There are three reasons why the Swiss context is particularly well suited for analysing the impact of the COVID-19 pandemic on foreign populations. First, with 25% of the population having been born abroad, Switzerland hosts a large and heterogeneous foreign population in terms of nationalities, qualifications, and age groups [17,18]. This heterogeneity makes it possible to study differentials in mortality across different groups of the foreign population. Second, Switzerland experienced particularly high levels of mortality during the COVID-19 pandemic [19], and it is important to understand in which groups of the population mortality has been especially strong. Finally, Switzerland is characterised by having accurate individual data on mortality, due to due to its comprehensive civil registration system, mandatory reporting, and standardised coding. Indeed, mortality in foreign populations in Switzerland has been the subject of various studies [20,21,22,23]. Although some groups, in particular undocumented immigrants [24] and foreign children [25], exhibit risk profiles different from those of Swiss nationals, overall, the rate of mortality in foreign populations is slightly lower than among Swiss nationals, in line with the above-mentioned ‘migrant mortality advantage’ theory.

We contribute to the existing research on the ‘migrant mortality advantage’ by showing that in Switzerland, some groups of foreigners—but not all groups—had a higher mortality rate from COVID-19 than Swiss nationals. We call this the ‘migrant mortality penalty in a pandemic’. In the Swiss case, this penalty characterises foreign communities who have been living in the country for more than 20 years, notably, Italians, people of Balkan origin, and those from Latin America (i.e., Central America, South America, and the Caribbeans). This finding invites a broader reflection on the equality of opportunity with respect to protection in a public health emergency situation for different groups of society, contributing to the understanding of migrant health disparities during pandemics and informing future public health interventions.

2. Background: The Mortality Risk of Migrants During the COVID-19 Pandemic

Several studies contribute to explaining the uneven impacts of the COVID-19 pandemic on different groups of the population. Early analyses estimated the incidence of the disease, its modes of transmission, and associated levels of mortality. However, it soon became apparent that the available data were incomplete. The number of cases of COVID-19, for instance, was underestimated [26,27]. A few months after the first wave of the pandemic, the availability of mortality data collected in hospitals allowed epidemiologists to better understand the phenomenon, but the number of deaths attributed to the virus continued to be underestimated [28]. As the pandemic progressed, the production of data from statistical offices allowed more a fine-grained analysis with information that was more exhaustive on causes of death and precise criteria, which are important for a virus that has particularly strong effects on elderly populations with co-morbidities.

In this paper, we focus on differences in mortality across different groups of the population, divided by nationality. In Switzerland, this indicator of migratory status is generally considered to be the most pertinent [21]. The justification for this paper is the observation that such differences require reflection on public health measures to ensure equal opportunities.

Historically, migrant communities have shown especially high increases in their risks of mortality, compared to native populations, during public health emergencies. For example, epidemic outbreaks in the United States at the beginning of the 20th century led to worse health conditions and higher rates of mortality for European immigrants, compared to native populations [29]. Migrant communities also exhibited a higher mortality risk than natives during the 1918 influenza epidemic [30]. Similar studies conducted during the COVID-19 pandemic suggest a higher mortality risk among foreign populations, compared to natives, during this public health emergency. Yet, while available evidence suggests that the COVID-19 pandemic has strongly affected—and possibly reversed—mortality differentials between foreign and native populations, the sizes of these differentials and their distributions across different age groups, nationalities, and gender remain unclear.

The first important study on higher mortality risks for different groups of the population during COVID-19 in high-income countries was published just after the first wave of COVID-19 in Italy [31]. This study finds a similar mortality rate during the pandemic for migrant populations compared to native populations, thus invalidating the healthy migrant effect. Subsequent studies suggest that during the pandemic, foreign populations experienced a higher rate of mortality compared to native populations. Based on a meta-analysis of articles published in high-income countries, another study [32] suggests that in 2020, mortality due to both COVID-19 and other causes of death was significantly higher for foreign populations than for natives. The authors observe that undocumented migrants, migrants living in collective households, and migrants working in health and care occupations were particularly concerned by an increased rate of mortality. In Milan, Italy, an increased rate of mortality among Latin American migrants was observed based on data referring to patients in hospital [12]. The same higher mortality risk in foreign populations compared to natives has been observed in Sweden [13,14]; in France, especially for sub-Saharan migrants [15,16]; in the United States [11]; and in England [25,33,34,35]. High levels of mortality were also recorded in US counties with a high proportion of Hispanic or Latino residents [36,37].

These studies show that mortality differentials during the pandemic were generally greater for migrants who came from distant countries, who are often the most vulnerable. These studies also propose five specific mechanisms that can explain these differentials and the uneven effects on different groups of foreigners.

First, larger differences were observed in France between foreigners and nationals in working-age populations [16]. This suggests that occupational exposure, given the fact that some groups of migrant populations find occupations in essential sectors, led to an increase in mortality risk during the pandemic. In particular, migrants are often overrepresented among workers in low-paying paramedical jobs, who are especially exposed to infection during their work and daily commuting.

Second, some groups of migrants belong to the most precarious socioeconomic groups in society. The link between socioeconomic status and the risk of mortality from COVID-19 is well documented [38,39,40]. People with low economic status are affected by more negative impacts of COVID-19 than those with high economic status. This can be explained by factors related to both immaterial and material resources, such as knowledge and money. Another study observed that in the countries with high levels of income inequality, mortality due to COVID-19 increased, compared to countries with lower levels of inequality [41], which can also be related to the fact that the most disadvantaged socioeconomic segment is characterized as having a greatly increased risk of mortality.

A third factor pertains to housing conditions. Ref. [16] notes the impact of population density in neighborhoods occupied by African populations in France, which are the most likely to experience a higher mortality risk, compared to native populations. Social distancing is more difficult among groups with a low socioeconomic status, who generally live in dense neighborhoods and housing structures. This hypothesis is confirmed by the fact that a higher mortality risk among migrants has also been observed in another country with a high concentration of African populations, Belgium [42].

The fourth important factor is the lack or difficulty of access to health for groups with low socioeconomic status, a phenomenon which is well known in the context of public health [43]. This leads to a situation in which some groups of migrants may have limited or delayed access to appropriate care for different reasons, such as language or financial barriers [44], race-related medical perceptions, and cultural and legal barriers [45]. These differentials were particularly observed in the United States shortly after the beginning of the pandemic [37,46].

Finally, due to the mode of transmission of COVID-19, a large portion of the literature has measured the effects of family composition, particularly with respect to intergenerational families, on COVID-19-related risks [47,48]. The number of persons sharing the same household is a risk marker, with more numerous households having a higher risk of mortality for COVID-19. Marital status is also considered a risk marker in some studies, with people who have never been married having higher mortality rates than people who were married and people who were divorced [14]. Given the possible differences in household size and marital status associated with nationality, we consider it important to take these variables into account.

3. Material and Methods

We use the data on individual mortality and causes of death for 2020 (BEVNAT statistics) provided by the Swiss Federal Statistical Office (SFSO). These data were matched with two other sources: the population statistics (STATPOP) which provide demographic information for the entire resident population of Switzerland (as of 31 December 2019 and 31 December 2020), and the Structural Survey, a non-exhaustive annual survey which was introduced in 2010, and replaced the Swiss Federal Census. The Structural Survey reaches out to more than 200,000 residents on 31 December of each year, enquiring about their socio-professional characteristics. The linkage was determinist, as all of these statistics are associated with the social security number used in Switzerland (the so-called AHV13), an anonymous identifier introduced in 2008 and since then commonly used in Switzerland.

From this, we created two databases with different criteria of inclusion. The first database exhaustively covers all the resident population of Switzerland present on 31 December 2019. This database only includes variables available on the BEVNAT and STATPOP statistics (Population I) and then excludes the socio-professional characteristics. The second database is limited to the persons who responded at least one time to the Structural Survey during the period 2010 to 2020 (Population II). It includes information on professional and socioeconomic positions. Most deaths due to COVID-19 involve people of retirement age, and for those people we do not have information relating to occupation or sector of activity. Therefore, the level of education attained remains the only variable estimating socioeconomic position that is available for all people in the sample.

We consider COVID-19 deaths those deaths in which the main cause (‘endgultig’ in the SFSO documentation) of death was coded as U071, according to the ICD-10 coding of the physician’s statement. We do not take into account deaths for which COVID-19 appeared as a concomitant cause according to the physician’s declaration. This decision is explained by the fact that the declaration of COVID-19 as a concomitant cause of death seems too imprecise in the context of a pandemic during which the level of screening was low, especially among people presenting lethal pathologies.

In our analyses, we only consider deaths that can be associated with persons living in Switzerland with a residence permit at the end of 2019 (31 December 2019). We therefore exclude from the analysis undocumented migrants and persons who arrived in Switzerland during the course of 2020. We also exclude persons who left Switzerland in the course of 2020. The number of deaths in our database differs slightly from the Swiss vital statistics, which include all deaths registered among the resident population, regardless of legal resident status, at the end of 2019.

After applying these criteria, foreigners make up 25% of the population considered in this study. The main region of origin of the foreign population is the European Union and European Free Trade Association countries (EU/EFTA—66% of foreigners have a nationality from these countries), followed by other European countries (17.2%), Asia and Oceania (7.9%), Africa (5.2%), North America (2.4%), and South America (1.3%).

We run logistic regressions to estimate the association of different factors with the risk of death of COVID-19, using the software SAS 9.4. Logistic regressions were intended to explain the probability (p) of death according to the dimensions under study and different control variables [49]. The formula is as follows:

logit(p) = ln(p/(1 − p)) = β₀ + β₁ x_(i,1) + β₂ x_(i,2) + ⋯

where β₀ is a constant and β_(1,…n) are the coefficients of the explanatory variables x_{(1,… n)}. We present the exponential values of the coefficients of the explanatory variables (odds ratio) and, for all models, we include the levels of significance (* p < 0.10; ** p < 0.05; *** p < 0.01) to facilitate the interpretation of the results.

As COVID-19 mortality was prevalent among adults, we limit the age categories to those above 24 years of age. Only two deaths were recorded in persons under the age of 25 in Switzerland during the year 2020.

Two different models are presented here.

• Model I: only controlling for age. Logistic regression estimates provide the odds ratio without further control. The model is run on the whole population observed in the STATPOP statistics (Population 1).
• Model II: controlling for age and socio-demographic variables (marital status, size of the household), as well as the region of residence and the level of education. The model is applied on a sample of people drawn from those answering the Structural Survey (Population II).

Given that the link between age and risk is not linear, age is introduced into the model using dummies by five-year age group.

Our data and methods have three important limitations.

First, the analysis is based on people who are registered as residents in Switzerland. As such, it does not account for the deaths of foreign individuals who reside irregularly in the country—undocumented or irregular migrants. In other words, we only compare the mortality risk of foreigners who live regularly in Switzerland to the mortality risk of Swiss nationals. The inclusion of undocumented migrants may yield different results, as these individuals are generally more likely to be exposed to specific hazards due to their precarious occupations and living conditions, and the difficulties they encounter in accessing health care systems [24].

Second, focusing on deaths does not account for other direct or indirect consequences of the pandemic. For example, it is now well known that long-term effects of COVID-19 infection in convalescent persons in all age groups will continue to be a burden. This goes beyond the scope of our analysis, which is limited to the effects of COVID-19 on mortality.

Third, death due to COVID-19 is often caused by a combination of different causes. We considered only the cases of COVID-19 for which this was indicated as the main cause of death as codified by the Federal Statistical Office and excluded cases in which COVID-19 was designated as a concomitant disease. We assume that the indications on the death certificate used by the FSO for coding are the most accurate identification of the main cause of death and that they do not vary according to nationality, as we have no reason to believe that there is a reporting bias, but there remains a small margin of error due to the possibility that different physicians applied different criteria when reporting.

4. Results

4.1. Descriptive Results

Table 1. Distribution (percentages) of the population groups living in Switzerland on 31 December 2019, according to demographic characteristics.

	Switzerland	EU/EFTA/UK	Other European	Africa	North America	South America	Asia	All
Sex
Men	48.5	54.9	49.6	53.2	33.8	43.9	49.1	49.6
Women	51.5	45.1	50.4	46.8	66.2	56.1	50.9	50.4
Age
0–24	25.8	22.7	29.8	38.8	22.1	21.4	33.2	25.7
25–64	52.6	67.9	63.4	59.4	75.5	74.1	63.8	56.1
65–79	15.8	6.8	6.3	1.6	2.0	3.7	2.6	13.3
80+	5.9	2.6	0.5	0.2	0.4	0.8	0.5	4.9
Area
Lake Geneva Region	17.2	26.5	13.6	33.7	41.8	40.2	23.1	19.1
Swiss Plateau	23.7	15.7	17.4	25.3	14.0	12.2	20.4	21.9
Northwestern Switzerland	13.4	13.3	21.0	10.0	10.8	12.0	14.2	13.6
Zurich	17.5	19.2	18.1	14.7	18.5	21.4	21.9	17.9
Eastern Switzerland	14.1	12.6	18.1	8.2	5.5	5.0	10.1	13.8
Central Switzerland	10.2	7.2	9.7	6.2	5.8	5.2	8.4	9.5
Ticino	4.0	5.5	2.2	2.0	3.7	4.0	1.9	4.1
Number	6,349,123	1,405,141	366,491	109,971	27,713	51,169	169,062	8,478,670

Source: STATPOP, persons living in Switzerland on 31 December 2019. We removed from the counts in this table any persons who emigrated during the year 2020, as they could not have been observed during the entirety of 2020.

shows the demographic characteristics of the populations in our databases. The foreign population is significantly younger than the population of Swiss nationals. For the population included in the analysis, the proportion of people aged 80+ is 4.9% (5.9% for Swiss nationals, less than 1% for most foreign-born groups, except for those from the EU/EFTA/UK which are at almost 3%). Overall, the proportion of people aged 65+ is 22% among Swiss nationals, compared to 9% for EU nationals and less than 3% among Africans and North Americans. For this reason, the expected number of deaths due to COVID-19 in foreign-born populations is significantly lower than for Swiss-born populations. There are different gender patterns, with most individuals from North America and Latin America being women, and most Africans and EU/EFTA/UK Europeans being male. In terms of place of residence, it is also important to note that Americans (from both North America and Latin America) and Africans live mainly in the Lake Geneva area, while people from the rest of Europe tend to live in the German-speaking regions of Switzerland. This can also impact the number of deaths, as the Lake Geneva area was more affected by the pandemics.

Due to this relatively young age structure, foreigners account for only 14% of the deaths due to COVID-19. Of the deaths among foreigners, 79% occurred among people from the EU/EFTA and 15% among those from other European countries.

Table 2. Deaths in 2020 according to the cause of death, in the context of the population living in Switzerland on 31 December 2019.

	Population 31 December 2019 *	Status at the End of 2020
		Alive	Dead
			COVID-19	Other Causes
Gender
Men	4,238,011	4,200,849	4780	32,382
Women	4,306,970	4,268,766	4275	33,929
Age Group
0–24	2,179,417	2,179,010	2	405
25–49	2,969,368	2,967,405	36	1927
50–64	1,796,512	1,789,775	303	6434
65–79	1,147,171	1,127,365	2263	17,543
80+	452,513	406,060	6451	40,002
Region of Nationality
Switzerland	6,408,608	6,341,195	7928	59,485
EU/EFTA (including UK)	1,410,720	1,404,255	886	5579
Other European	367,269	366,318	173	778
Africa	110,085	109,957	14	114
North America	51,267	51,150	19	98
Latin America	27,740	27,710	3	27
Asia/Oceania	169,292	169,030	32	230
All	8,544,981	8,469,615	9055	66,311

Sources: SFSO, Bevnat, and Statpop. * After the exclusion of persons emigrating during the year 2020 or not resident on 31 December 2019.

, below, provides descriptive statistics related to our sample.

The data in Table 2 offer some preliminary observations on the relationships between migration status, socioeconomic conditions, and health outcomes. Age is the strongest determinant of mortality, with the 80+ group experiencing the highest rate of deaths, particularly as to those from COVID-19. Men had a slightly higher COVID-19 mortality than women, reflecting known biological and behavioral health vulnerabilities. Regarding migration status, Swiss nationals accounted for most deaths, but there are important variations among the foreign-born populations. EU/EFTA nationals experienced higher mortality than non-European migrants, possibly due to age distribution. Lower mortality among migrants from Africa, Asia, and the Americas may reflect their younger age structure.

4.2. Logistic Models

The results are presented in

Table 3. Results of logistic regressions on the probability of death due to COVID-19 in Switzerland, in 2020, for those aged 25 and older (whole population).

	Aged 25 and Older—Model I
	Men			Women
Citizenship	OR	CI 95%	p	OR	CI 95%	p
Switzerland	1.00			1.00
EU/EFTA	1.19	{1.09–1.31}	***	1.18	{1.06–1.32}	**
Other European	2.36	{1.94–2.86}	***	2.48	{1.93–3.18}	***
Africa	2.31	{1.30–4.10}	**	0.95	{0.24–3.82}
North America	1.69	{0.54–5.33}		…	…
Latin America	2.16	{1.42–3.28}	***	2.24	{1.34–3.74}	**
Asia/Oceania	1.18	{0.72–1.94}		1.90	{1.15–3.12}	*
N	4778			4275
Wald Test	9224.45	20	<0.0001	7371.55	20	<0.0001

Sources: SFSO, Bevnat, and Statpop. Persons aged 25 and older; after control for age. OR stands for odds ratio and CI stands for confidence interval.

for the first model and

Table 4. Results of logistic regressions on the probability of death due to COVID-19 in Switzerland, in 2020, for those aged 25 and older. Population: participants in the Swiss Structural Survey.

	Aged 25 and Older—Model II
	Men				Women
	OR	CI 95%		p	OR	CI 95%		p
Citizenship
Switzerland	1.00				1.00
EU/EFTA	0.95	{0.81–1.11}			1.21	{1.01–1.46}		*
Other European	2.00	{1.36–2.92}		***	2.42	{1.49–3.90}		***
Africa	0.48	{0.07–3.45}			…	…
North America	…	…			…	…
Latin America	2.89	{1.47–5.69}		**	3.14	{1.42–6.93}		**
Asia/Oceania	1.11	{0.45–2.70}			2.59	{1.15–5.87}		*
Region
Lake Geneva region	1.00				1.00
Swiss Plateau	0.55	{0.48–0.63}		***	0.67	{0.57–0.78}		***
Northwestern Switzerland	0.48	{0.40–0.57}		***	0.53	{0.44–0.64}		***
Zürich	0.42	{0.35-0.50}		***	0.57	{0.47–0.69}		***
Eastern Switzerland	0.51	{0.43–0.60}		***	0.61	{0.50–0.74}		***
Western Switzerland	0.38	{0.31–0.47}		***	0.38	{0.30–0.49}		***
Ticino	0.90	{0.76–1.08}			0.96	{0.79–1.15}
Marital Status
Married	1.00				1.00
Other	1.34	{1.10–1.61}		**	1.15	{0.95–1.40}
Household Members
One	1.35	{1.20–1.51}		***	1.54	{1.29–1.84}		***
Two	1.00				1.00
Three and more	0.91	{0.65–1.27}			1.96	{1.35–2.84}		***
Collective households	7.94	{3.81–16.58}		**	6.45	{3.68–11.30}		***
Level of Education
Secondary I	1.35	{1.20–1.51}		***	1.36	{1.21–1.53}		***
Secondary II	1.00				1.00
Tertiary	0.68	{0.60–0.77}		***	0.75	{0.59–0.96}		*
N	1667				1368
Wald Test	3431.7	32		<0.0001	2386.58	32		<0.0001

Sources: SFSO, Bevnat, Statpop, and Structural Survey. Persons aged 25 and older; after control for age. OR stands for odds ratio and CI stands for confidence interval.

for the second one. The two models converge with regard to the association between region of nationality and COVID-19 mortality, with some limitations that are explained by the small sample size (model II) and the relatively small number of older-aged people of non-European origin in Switzerland.

The results show a significantly higher risk of death from COVID-19 in the year 2020 for some—but not all—nationality-based groups, compared to Swiss nationals. Specifically, there are increases of about 18% in mortality risk among EU/EFTA/UK citizens for both men and women, when only controlling for age (model I). However, this mortality pattern disappears after controlling for demographic factors other than age and the level of education (model II—Table 4).

Table 4, below, shows that, compared to Swiss nationals, individuals from ‘Other European’ countries and Latin America face significantly higher mortality risks (e.g., men from Other European countries: OR 2.00***; women from Latin America: OR 3.14**). Interestingly, migrants from Africa and Asia/Oceania show no clear pattern, with some subgroups having lower or non-significant risk levels. Being unmarried is associated with a higher mortality risk, especially for men (OR 1.34**). Finally, higher education is associated with lower mortality risk, with tertiary-educated individuals having significantly lower odds of death compared to those with secondary education (OR 0.68*** for men, OR 0.75 for women*).

However, examining European nationalities as a single, cohesive group makes it difficult to interpret the unique characteristics and experiences of each migrant nationality. For this reason, we propose a model identifying mortality risks for those nationalities having recorded at least 15 deaths due to COVID-19 during the year (

Table 5. Results of logistic regressions on the probability of death due to COVID-19 in Switzerland, in 2020, according to nationality.

	Men			Women
	OR	CI 95%	p	OR	CI 95%	p
Switzerland	1.00			1.00
Germany	0.59	{0.45–0.78}	***	0.64	{0.45–0.91}	*
France	0.92	{0.69–1.22}		0.81	{0.57–1.16}
United Kingdom	0.77	{0.45–1.34}		0.64	{0.29–1.44}
Italy	1.32	{1.18–1.49}	***	1.29	{1.11–1.49}	***
The Netherlands	0.92	{0.49–1.72}		0.60	{0.22–1.60}
Austria	0.97	{0.66–1.43}		0.70	{0.38–1.31}
Portugal	0.81	{0.46–1.44}		1.10	{0.54–2.22}
Spain	0.92	{0.63–1.34}		0.92	{0.60–1.41}
Turkey	2.21	{1.43–3.41}	***	1.44	{0.72–2.90}
Serbia	2.83	{1.87–4.28}	***	2.09	{1.15–3.81}	*
Bosnia	2.33	{1.28–4.25}	**	4.08	{2.29–7.26}	***
Macedonia	3.52	{2.18–5.71}	***	3.60	{1.86–6.97}	***
Kosovo	3.25	{2.32–4.56}	***	4.79	{3.15–7.29}	***
Sri Lanka	3.67	{1.88–7.13}	***	3.87	{1.59–9.40}	**
Somalia	10.92	{3.40–35.03}	***	…(a)	…
Other countries	0.89	{0.68–1.15}		1.03	{0.76–1.39}

Sources: Statpop and BEVNAT. Persons aged 25 and older. After control for age, region of residence, number of persons in the household, and marital status. (a): no cases. OR stands for odds ratio and CI stands for confidence interval.

). In this model the risk of mortality is significantly higher for Italians (both men—OR 1.32***—and women—OR 1.29***), but it is lower for Germans (both men—OR 0.59***—and women—OR 0.64*). The results hold after controlling for demographic confounders.

We observe a significantly higher mortality risk among non-EU/EFTA European populations, one which remains at the same level after controlling for demographic confounders and education (OR 2.00*** for men and 2.42*** for women). All Balkan communities are affected by an increased mortality risk which concerns both men and women. The risk can be more than three times higher for Macedonian and Kosovar men and women and for Bosnian women. The fact that risk differentials remain significant after controlling for confounding variables suggests that factors other than socioeconomic status and household size determine this increased risk (e.g., increased professional exposure or sub-optimal health care, as discussed above). We do not find significant differences in mortality for North American citizens compared to the Swiss, but we do find a higher risk of mortality for South Americans, who exhibit a high risk both before and after controlling for confounding factors (OR 2.89** for men and 3.14** for women).

The African community in Switzerland is relatively heterogeneous, in the sense that it includes asylum seekers, refugees, regular immigrants, and undocumented immigrants. Within this diverse group, there are both people with a precarious residence status and a low or intermediate level of qualification, and highly qualified persons who have obtained entry to Switzerland through their professional qualifications. Given the low number of cases of death among African nationals, it is not possible to distinguish these two categories. Despite this limitation, rates of COVID-19 mortality are slightly higher for African men who hold a regular permit, compared to Swiss nationals, before controlling for confounding, but the difference compared to Swiss men is not significant after controlling for demographic variables. In particular, a significantly higher rate of mortality is observed among the men from Somalia (Table 5).

Finally, there is a slightly higher mortality among women of an Asian or Oceanian nationality, which does not decrease after controlling for confounding factors. The interpretation of this higher mortality is difficult: the analysis by nationality (Table 5) shows that the mortality rate is very high for Sri Lankans of both sexes (men 3.67***, women 3.87**), a nationality characterised as comprising a high number of asylum seekers and refugees working in the restaurant industry or in the health sector.

4.3. Other Risk Factors

Firstly, the region of residence marks significant differences. Compared to the region considered as a reference (Lake Geneva region, which has the highest population density in Switzerland), the rate of mortality is, without exception, lower for all the other regions except for Ticino. The differences in rates of mortality between Latin Switzerland (composed of Ticino and the Lake Geneva region) and Central and North-Eastern Switzerland (in the broad sense, including certain French-speaking cantons and the whole of German-speaking Switzerland) can be explained by the much higher incidence of COVID-19 in these regions. If we limit ourselves to the year 2020, it has been established that the geographical proximity of Ticino to northern Italy played an important role in the high incidence of COVID-19 in this Italian-speaking canton of southern Switzerland, particularly during the first wave of spring 2020; this incidence resulted in a large increase in mortality levels [50]. Similarly, probably due to the numerous social contacts between Italy and the Lake Geneva region, the latter has experienced higher incidence rates than the rest of Switzerland, and higher mortality [50]. On the other hand, there is little explanation for the differences observed in the second wave.

In addition, living as a couple is statistically associated with a decrease in risk, compared to living alone or in a household of three or more people, a result that confirms the findings in the literature [14]. These results are coherent with the hypothesis that married people pay more attention to preventive messages than do unmarried people. This higher level of attention may counterbalance the fact that an infectious disease is often transmitted between members of the same household.

With regard to household size, we also included a modality referring to non-family households, such as hotels, homes for elderly persons, homes for orphans, and clinics and hospitals, which are often collective households. According to the model, the risk of death from COVID-19 for those who live in collective households increased by a factor of between six and eight compared to people living in a two-person household. This result is to be expected, in the sense that collective households were more often affected by the virus, which spreads more rapidly in such settings. Moreover, the fragility of people who are placed within collective households is likely to have also played an important role in this context.

The last control variable introduced in model II is the level of education, a marker of socioeconomic status. Unsurprisingly, the probability of dying from COVID-19 is increased among people with an education level of secondary I or less and decreased among those with a tertiary-level education, compared to the median level (secondary II). While we do not comment further on this result, it is important to note that it confirms the findings in the existing literature, but, once introduced in the modelling, does not significantly change the observed results according to nationality. In other words, nationality is associated with the level of risk even after this socioeconomic determinant is considered.

5. Discussion

Because death represents the end of a process of illness, differences in mortality among groups can express different things: either a variation in the proportion of people affected by the disease in each group, or a difference in the survival of the infected people [10]. Here we consider the data for 2020, when Switzerland was hit by two waves of COVID-19: the first in spring, the second at the end of the year. The proportion of people who had been affected by COVID-19 by the end of 2020 in Switzerland was estimated to be at around 21%, with an increased risk of contagion for middle-aged persons compared to the youngest and the oldest ones [51]. However, the authors of this previous study found no differential in prevalence between natives and foreigners in Switzerland, contrary to other countries like England [33]. Due to the lack of data for specific nationalities, it is impossible to estimate whether the differences between nationality-based groups that are observed in mortality are due to a variation in incidence or in survival.

Three remarks should be made before discussing the main results. First, as mentioned before, COVID-19 mortality mainly impacted elderly populations. In 71% of cases recorded in Switzerland, the person who died because of COVID-19 was aged 80+ years. This is important to note, because it indicates that our results do not exclusively refer to specific behaviours and exposures observed in the context of professional activity. Additionally, if the incidence rate of COVID-19 can vary according to the status of the person on the labour market, this does not mean that these variations automatically translate into differential mortality rates. Moreover, as migratory flows mainly concern young people, foreigners were not among those most immediately exposed to the risk of death.

Second, mortality due to COVID-19 mainly concerns people already suffering from respiratory or other pathologies (diabetes, etc.). The Swiss Health Survey 2017, which was the last national health survey organised in Switzerland before the COVID-19 pandemics, suggests that foreigners may be more affected by potential co-morbidity factors compared to Swiss citizens. In particular, 48% of foreigners aged 15 and above were overweight or obese (Swiss citizens, 40%). Tobacco consumption is also more frequent among foreigners aged 15 and above (32% versus 26%). However, this survey does not provide a precise overview of the differential in health according to nationality for persons who are aged 80+. For this reason, we cannot establish causal relationships between the state of health, the presence of comorbidities, and the increased risk of death in certain categories of foreigners. This is an important limitation, as our results only show significant associations between nationality and the risk of death.

Third, we emphasise again that these findings only apply to the population of migrants in Switzerland with a regular permit. The impossibility of taking into account the population without a residence permit, explained by the absence of reliable data, may hide important differences between Swiss nationals and certain groups of migrants, particularly those from the Balkans and African countries.

The results obtained for Switzerland confirm studies undertaken in other high-income countries, but also contribute to the identification of the specific groups among the foreign population that were the most affected. Our findings reveal that certain groups of foreigners residing in Switzerland, particularly those from most EU countries and African nations, retained the ‘migrant mortality advantage’ during the pandemic. However, other foreign groups exhibited unexpectedly high mortality rates, suggesting that some segments of the migrant population faced a ‘migrant mortality penalty’, one likely driven by specific transmission pathways and heightened exposure risks.

We see this clearly among people of Italian nationality, across both men and women, a group which exhibits significantly higher rates of mortality compared to Swiss nationals as well as to other EU/EFTA/UK foreigners. The specific case of the Italian group might be due to four characteristic elements of this group of the population. First, Italian migration to Switzerland is long-standing, dating from the post-war period; historically, this migration has been characterised by low-skilled individuals who maintain strong intra-community relationships and work in frontline sectors, such as healthcare, hospitality, construction, and manufacturing [52]. Second, the Italian population in Switzerland includes a high proportion of individuals aged 65+, who are more vulnerable to severe COVID-19 outcomes. Third, the proximity of Switzerland to Italy makes it easy for people of Italian nationality to return home. Indeed, Italians are the foreign community in Switzerland with the highest rate of journeys back to their home country, with an estimated 22% of them going to Italy at least once a month [53]. This is particularly the case for northern Italians: they are less than four hours by train from Milan, which was the gateway to COVID-19 in Europe, with a very high mortality rate at the start of the first wave of the pandemic. It is likely that connections between Switzerland and northern Italy contributed to the spread of the virus in both Ticino and French-speaking Switzerland between February and March 2020 [54]. Fourth, the abrupt halt in cross-border mobility between Italy and Switzerland at the end of March 2020 [55] following the introduction of travel restrictions [56,57,58] may have had severe consequences for Italians who regularly sought healthcare in their home country due to trust and familiarity with the health system [59,60]. The closure of the borders forced this group of the population to postpone or reduce their use of medical services, potentially impacting their health outcomes. These combined factors suggest that Italians in Switzerland faced a unique convergence of high exposure, increased vulnerability, and restricted healthcare utilisation, which likely contributed to their disproportionately high mortality rates during the pandemic.

The other two groups of nationals affected by a significantly higher mortality due to COVID-19 compared to the Swiss, both men and women, are those of people of a non-EU/EFTA/UK European nationality, mainly nationals of the Balkans, and people from Latin America. The higher mortality observed in these communities also suggests possible interpretations. Migration from the Balkans to Switzerland consists of two main groups: labor migrants, who arrived in the 1970s and 1980s, and asylum seekers, who fled to Switzerland during the Balkan wars of the 1990s. Migrants from Latin America are mainly settled through a work permit. These are populations that are engaged in so-called essential jobs, jobs which continued at the time of the first confinement [61]. Furthermore, as was the case with the Italians, individuals from the Balkans and South American countries maintained strong personal contacts, which can lead to an increased risk of transmission. The higher mortality in this group may be linked to the fact that elderly persons in these communities were overexposed to the risks of contracting COVID-19 because of a strong communal and intergenerational lifestyle.

Different levels of mortality due to COVID-19 across various groups of foreigners are not due to the congenital frailty of the immigrants themselves, but rather to a set of social and structural conditions that make some specific groups more exposed to those diseases in times of public health emergencies. In the end, the increased risks of mortality from COVID-19 observed among these populations suggest that higher mortality can be driven by social factors such as the demographic and socioeconomic profiles of some groups of migrants, international travel, and occupational exposure.

The results also point to survival differentials among people who have contracted COVID-19; a phenomenon which can be explained in two ways. First, episodes of co-morbidities are more frequent in certain groups; these are only loosely documented by national health surveys, but they should be analysed in more detail through medical records. Second, medical care may be different for different groups of the population, either because of a reduced use of care in the event of a deterioration in the health situation in a low-income population, or because of a lack of understanding of preventive messages. According to Ruedin et al. [62], who studied health literacy at the time of COVID-19 in Switzerland, migrants felt well informed. However, the authors found “a moderate correlation between COVID-19-related health literacy and socioeconomic vulnerability”, meaning that it was more difficult for socioeconomically marginalised migrants to find and understand information regarding COVID-19. Moreover, according to the authors, the most socioeconomically vulnerable migrants adhered more to unscientific theses that were not part of official communications.

Concerning health care access in Switzerland, health insurance is mandatory and covers most of the cost of consultation and hospitalisation. However, in cases of illness, part of the cost is to be covered by the insured person. It is difficult to estimate the number of people who forego health care because of budgetary reasons in Switzerland, as surveys provide results that vary according to the methodology used. According to a recent study, this foregoing of care may extend to up to 20% of the population, with an increased risk among foreigners compared to Swiss citizens [63]. According to recent data, foreigners in Switzerland are 1.75 to 2.5 times more likely to ignore their health insurance plan, as well as the level of their deductible, compared to Swiss nationals [64]. In Switzerland, the issue of discrimination in access to health care has received little attention in a context in which health insurance is compulsory. However, recent studies show that access may be limited for certain groups, mainly because of racism [65], but also because of a lack of attention to the cultural specificities of health care [66]. In this context, our results show that in a society in which access to health care is guaranteed by compulsory health insurance, it would be important to strengthen the measures developed to promote equitable access to health care for different segments of the population.

6. Conclusions

The analysis of mortality in Switzerland during the pandemic in 2020 exhibits a significantly higher mortality due to COVID-19 in some groups of foreign nationals compared to Swiss nationals. This reverses the ‘migrant mortality advantage’, or the generally observed low levels of mortality among groups of foreign individuals. During the COVID-19 pandemic, this effect was turned upside-down, exacerbating health inequalities based on migration status [67]. We call this ‘the migrant mortality penalty in a pandemic’ and we argue that common explanations for the ‘migrant mortality advantage’ disappear in a context of high risk from infectious disease due to specific transmission pathways that affect some foreign communities more than others.

The higher mortality, compared to Swiss citizens, is observed among foreign individuals, both men and women, who have been living in Switzerland for more than 20 years, notably Italians, people of Balkan origin, and those from Latin America. One can argue that these groups have been most acutely affected because they are characterised as having an older population, one which is more at risk of death from COVID-19. However, our models control for the age effect. For these three communities, we suggest that the mortality level was driven by occupational exposure, strong community relations, and the sharing of socially dense living areas, as well as important intergenerational relations which facilitated contact with people who had medium to low working qualifications and who were, in turn, highly exposed to the infection due to their work. In addition to these social factors, we suggest that there may have been differentials in the use of healthcare for people infected with COVID-19 due to linguistic and financial barriers. Foreign individuals would occasionally seek care in the country where they hold nationality, but during the pandemic they were no longer able to do so due to travel restrictions and lockdowns. Overall, these specific social conditions and hurdles to accessing health treatment are likely to have contributed to the significantly higher rates of mortality for COVID-19 for some groups of foreign nationals, compared to Swiss nationals.

The elements presented in this article should be of concern to public health officials. An impressive body of literature shows that the COVID-19 pandemic has strongly affected the most disadvantaged socioeconomic groups. Our analysis of the Swiss case suggests that more can be done to address the specific situations encountered by foreign populations during the COVID-19 pandemic, considering their living and working conditions, promoting more equitable access to health systems, and reducing communication barriers.