Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Global Public Health. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 26 February 2024

Convergence Theory and the Salmon Effect in Migrant Healthfree

Convergence Theory and the Salmon Effect in Migrant Healthfree

  • Yudit NamerYudit NamerDepartment of Epidemiology and International Public Health, Bielefeld University
  •  and Oliver RazumOliver RazumDepartment of Epidemiology and International Public Health, Bielefeld University


For decades, researchers have been puzzled by the finding that despite low socioeconomic status, fewer social mobility opportunities, and access barriers to health care, some migrant groups appear to experience lower mortality than the majority population of the respective host country (and possibly also of the country of origin). This phenomenon has been acknowledged as a paradox, and in turn, researchers attempted to explain this paradox through theoretical interpretations, innovative research designs, and methodological speculations.

Specific focus on the salmon effect/bias and the convergence theory may help characterize the past and current tendencies in migrant health research to explain the paradox of healthy migrants: the first examines whether the paradox reveals a real effect or is a reflection of methodological error, and the second suggests that even if migrants indeed have a mortality advantage, it may soon disappear due to acculturation. These discussions should encompass mental health in addition to physical health.

It is impossible to forecast the future trajectories of migration patterns and equally impossible to always accurately predict the physical and mental health outcomes migrants/refugees who cannot return to the country of origin in times of war, political conflict, and severe climate change. However, following individuals on their path to becoming acculturated to new societies will not only enrich our understanding of the relationship between migration and health but also contribute to the acculturation process by generating advocacy for inclusive health care.


  • Global Health
  • Special Populations
  • Theory and Methods

Healthy Migrants: A Paradox

As census data became increasingly available to researchers in the United States, a phenomenon called the Hispanic Paradox, the epidemiological finding that migrants of Latin American descent enjoy a mortality advantage compared to the majority white population began to be documented (Bradshaw & Fonner, 1978; Siegel & Passel, 1979).1 The phenomenon was first expressed as a paradox by Markides and Coreil (1986), who reviewed the existing literature and concluded that despite social disadvantages, the southwestern Latina/o population, the majority of whom were Mexican American, placed closer to white Americans in terms of infant mortality, cardiovascular mortality, and cancer morbidity. As a matter of fact, Mexican Americans had significantly lower mortality rates than white Americans. Only in cases of diabetes and infectious diseases were the Latinas/os more disadvantaged than the white population. This was a paradox because although the Latina/o population in the United States shared similar socioeconomic conditions as African Americans, Latina/o health appeared to be remarkably better. Other studies quickly followed to establish the existence of this phenomenon throughout the regions where Latina/o populations resided in the United States (for a systematic review of the Hispanic mortality paradox, see Ruiz, Steffen, & Smith, 2013).

A similar phenomenon was discovered in Britain by Swerdlow (1991). By following up Vietnamese “boat people” who arrived in Britain during and in the aftermath of the Vietnam War, Swerdlow (1991) was able to show that the refugee population had significantly lower overall mortality than the majority populations in England and Wales. More specifically, Vietnamese individuals had lower mortality from ischaemic heart disease, colorectal and breast cancers, and non-cerebrovascular circulatory disease. However, higher mortality from tuberculosis, stomach cancer, peptic ulcers (in women), cancer of the nasopharynx (in men), as well as cerebrovascular and chronic lower respiratory disease was observed relative to the general population of England and Wales.

Subsequent studies pursued to demonstrate that this was not a phenomenon native to the Latina/o population in the United States. For example, a Mediterranean paradox has also been described, although not as widely as the Hispanic paradox. Darmon and Khlat (2001), in a systematic review, presented a mortality advantage among North African migrants in France. Despite poorer health coverage, lower utilization of health services, and less likelihood to engage in preventative health measures, Moroccan migrant men in France had lower rates of death than the general population in France, as well as Morocco (Khlat & Courbage, 1996). On the other hand, migrants from Eastern Europe, West Africa, India, Pakistan, and Southeast Asia had higher rates of death compared to the general population: Eastern European migrants due to cardiovascular, digestive and respiratory diseases, and the other groups due to parasitic and infectious diseases (Wanner, Bouchardy, & Khlat, 1997). Similarly, in the Netherlands, Uitenbroek and Verheff (2002) showed that among the residents of Amsterdam, life expectancy was lowest for people of Dutch origin and other industrialized countries and highest for Mediterranean migrants. Turkish and Moroccan migrants in Brussels, Belgium, also had lower mortality despite discrimination and poor standards of living (Anson, 2004). In another Dutch study, however, focusing on national statistics, Turkish, Antillean/Aruban, and Surinamese men had higher rates, but Moroccan men had lower rates of mortality than Dutch men, hinting at the complexity of findings (Bos, Kunst, Keij-Deerenberg, Garssen, & Mackenbach, 2004).

From the 2000s onward, with the aid of systematic reviews and novel methodologies, the finding of healthy migrants has been more critically cogitated. A 2001 review reported variant descriptions of “Hispanic” in the literature: self-report, having a Spanish surname, mother’s ethnicity on birth certificate, nativity, and death certificate data were employed by different studies to verify the mortality advantage (Franzini, Ribble, & Keddie, 2001). Despite the range of descriptions, and the detailed consideration of the methodological differences, the review concluded that there is an all-cause mortality advantage among the Latina/o population, especially in middle and old age. Singh and Hiatt (2006) analyzed census data from the United States between the years 1979 and 2003 and interpreted a growing health advantage of foreign-born US residents over native born, not only for Latina/o migrants, denoting the heterogeneity of migrant health. More specifically, migrants had a life expectancy of 2.3 years higher than the US-born population between 1979 and 1981, and 3.4 years longer between 1999 and 2001. The mortality advantage also increased for cancers, diabetes, respiratory, and cardiovascular diseases and accidental as well as deliberate injuries despite poorer health-care access and utilization (Singh & Hiatt, 2006).

Another longitudinal approach, a meta-analysis and systematic review that looked at 58 longitudinal studies published between 1990 and 2010, and extracted data for 4,615,747 participants, further concluded that Hispanic ethnicity indeed conveys a mortality advantage. This advantage is found to be moderated by age, with the effect being stronger for older participants. The advantage also varied according to preexisting conditions, with the Hispanic population advantaged in the contexts of renal disease and cardiovascular disease but at equal risk in terms of cancer and HIV/AIDS with the non-Hispanic population. The Hispanic population, although at a mortality advantage compared to non-Hispanic white and non-Hispanic black populations, were at a minimal disadvantage compared to Asian Americans. The authors of the meta-analysis argued that the Hispanic mortality advantage was a robust effect and recommended scrutinizing over the causes of this resilience rather than its presence in the light of the data available (Ruiz, Steffen, & Smith, 2013).

It is important to note that most of the discussion in this article will revolve around mortality findings. When it comes to morbidity, especially self-reported symptomatology, the situation becomes more complex. For example, Razum and Zeeb (2000) argued that self-reported cardiovascular morbidity among Turkish migrants may be higher relative to the German population, even though cardiovascular mortality seemed to be lower. This may ultimately be less strange than it initially seems, as there is a parallel example in other groups: In most countries, women have a worse self-reported health than men, but a substantially lower mortality (see Benjamins, Hummer, Eberstein, & Nam, 2004 for a discussion of the relationship between self-reported health and mortality). We here discuss mortality differences between migrant and host populations and illustrate morbidity differences using the example of mental health.

Explanations of a Paradox: Better Health or Error?

Many hypotheses have been proposed to explain the finding of migrants living longer than the majority population. Some are specific to the Hispanic paradox; others are put forward in relation to other migrant populations. They range from the characteristics of migrants to methodological bias or study design problems. The salmon effect is considered to be a characteristic of migrant populations, which results in methodological bias: whereas the convergence theory tries to answer whether this effect persists as time is spent in the host country. The following summary of other explanations featured in the literature attempts to contextualize the discussions of the salmon effect and the convergence theory. These explanations are argued to be complementary rather than contradictory to each other (Domnich, Panatto, Gasparini, & Amicizia, 2012).

Healthy Migrant Effect/Selection Bias

This explanation is based on the hypothesis that only the healthier members of a population migrate.2 It reflects a bias of selection: either by the person, in the decision to migrate or flee; or by the state, which may specifically call for healthy workers or refuse to host ill individuals (Chiswick, Lee, & Miller, 2008; Gushulak, 2007; Vang, Sigouin, Flenon, & Gagnon, 2015). There is also an issue of survival: even if less than healthy individuals set out to migrate, they may not have survived the transit. To consider Swerdlow’s (1991) findings as an example, the Vietnamese refugees who left during the war traveled on unstable vessels, leaving the unwell members of their family behind. It is assumed that over 10% of the refugees lost their lives during the journey as a result of drowning, starvation, thirst, diseases, or battles with pirates (Wain, 1981).

One component of this selection bias is comparable to the healthy worker effect (McMichael, 1976). Workers in many occupations are selectively healthy since ill and physically limited individuals are not selected into the workforce. However, this effect is argued to be visible only immediately following selection to work, as years in the workforce may have detrimental consequences, reversing this effect (Juel, 1994). It may persist if workers maintain their health through health promoting policies, such as smoking bans and provision of a healthy diet, in certain industries (Li & Sung, 1999). Routine health screening and access to health care may also contribute to the persistence of better health of workers (Baillargeon, 2001). The example of the healthy worker highlights the complexity of behavioral and structural factors contributing to health in the life course.

Barrio Advantage/Ethnic Enclave Hypothesis

The health disadvantages of living in economically disadvantaged, segregated neighborhoods, or “ghettos” are significant for the African American population. Controlled for field center, age, and sex, living in racially segregated residential areas was found to be related to systolic blood pressure increase for the non-Hispanic black individuals in a recent study (Kershaw et al., 2017). Even though the Latina/o populations in the United States also live in enclosed neighborhoods, or “barrios,” where most residents are Mexican American, it is argued that they overcome the socioeconomic disadvantage through higher rates of residential stability, property ownership, labor participation, and close-knit family relationships (Moore, 1989; Moore & Pinderhuges, 1993). Eschbach, Ostir, Patel, Markides, and Goodwin (2004) investigated the impact of living in barrios on health by analyzing the prevalence of a number of illnesses and survival over seven years for Mexican Americans over the age of 65. They found lower levels of morbidity and mortality in neighborhoods dense with Mexican American residents and higher levels of morbidity and mortality in neighborhoods where Mexican Americans and non-Latina/o populations are integrated, providing support for this advantage (Eschbach et al., 2004). It is possible that the finding of higher mortality and morbidity in integrated populations is partly related to migrant populations’ acculturation and eventual convergence to the health of majority populations, which will be discussed later (see section “Convergence Theory: Do Migrants Stay Healthy?”).

Time Travel Metaphor/Health Transition

Drawing on the story of Oliver Twist, Razum and Twardella (2002) suggested that migrants coming from a country that has not adopted the dietary practices and sedentary lifestyle of industrialized societies associated with ischaemic heart disease and certain cancers have a lower risk of developing these chronic illnesses. To exemplify the mechanism, Young and Hopkins (2014) inferred an indirect protective effect of high-fiber legume consumption, which characteristically feature in Latina/o and Mediterranean diets, in surviving lung, colon, and breast cancer via a reduction of systemic inflammation. On the other hand, migrants benefit from the immediate effects of advanced health-care technologies of the industrialized host country that counter the negative health outcomes of poorer living conditions, such as epidemic infectious diseases. Razum and Twardella (2002) argued that migrants are on different continua in comparison to the majority populations: (1) They are less (or only for shorter time periods) exposed to lifestyle factors that lead to chronic diseases in late adulthood and old age; (2) the consequences of poorer health care in the country of origin are quickly reversible by the better availability of health care in the host country, thus reducing the mortality from infectious and perinatal causes. This is a process of “health transition.” This by no means connotes that migrants are immune to chronic illnesses; in fact, policies often neglect chronic illnesses in migrants. Migrants likely encounter the negative effects later in their lifetimes, manifesting a convergence (Razum & Twardella, 2002).

Data Artifact

Wallace and Kulu (2014), in their analysis of migrant mortality in England and Wales, summarize forms of data artifact that may lead to the presumably false conclusion that migrants are healthier than the general population: (1) Age can be misreported; (2) nationality and/or ethnicity may be misclassified; (3) different forms of return migration may lead to numerator-denominator error, causing returned migrants to indefinitely age in population registers (Wallace & Kulu, 2014). City-specific calculations further overlook the choice of native-born residents to move to rural areas toward older age, disappearing from the city registers (Uitenbroek & Verhoeff, 2002).

To illustrate one form of data artifact, Smith and Bradshaw (2006) took into account the changes in census description decisions, from using Spanish surnames to self- and informant-reported country of origin to determine Latina/o ethnicity in the United States. To resolve inconsistencies in mortality results that different classification methods may cause, they compared death counts and population growth rates in Latina/o and non-Latina/o ethnic backgrounds and reestimated non-Latina-white and Latina/o death rates. Their findings led them to conclude the difference between both populations’ life expectancies to be insignificant and to declare that “there is no Hispanic paradox” (Smith & Bradshaw, 2006, p. 1686). It was not a subtle declaration, yet this declaration has not halted the efforts for methodological innovations. A subsequent study investigated the issue of ethnic misclassification on death certificates by comparing census data and death certificates that are linked to the census in the United States: the study found the classification of Hispanic origin on death certificates to be sound (Arias, Eschbach, Schauman, Backlund, & Sorlie, 2010).

Three forms of return migration are additionally discussed by Wallace and Kulu (2014): (a) the mobility bias, (b) the unhealthy re-migration hypothesis, and (c) the salmon effect. The mobility bias reflects the multiple unregistered incidences of returning to one’s homeland, irrespective of health status, specifically when the host country and the country of origin are geographically proximal. Mobility bias would statistically contribute to risk time (Khlat & Darmon, 2003). The unhealthy re-migration hypothesis refers to an ongoing selection process that sees the unhealthy return and the healthy remain (Razum, Zeeb, Akgün, & Yilmaz, 1998). Along with the salmon effect, they will next be discussed in more detail.

Salmon Effect

The salmon effect (or the salmon bias as it is sometimes referred to in the literature) describes the possible trend of migrants returning to their country of origin when they are gravely ill. They may do so to spend their last days in their homeland or to be cared for by relatives (Pablos-Méndez, 1994). The metaphor echoes the lives of salmon, which begin their lives in rivers, spend adult lives in the ocean, and return to natal rivers to spawn toward the end of their lifetimes (Moyle & Cech, 1996). The argument holds that this pattern of return migration is both selective and not reflected in the census data, on which mortality studies are usually based, leading to the reported elevated lifespans of migrants. Shai and Rosenwaike (1987), among others, employed this explanation to partially explain a lower mortality among Mexican-born residents in the greater Chicago metropolitan area.

According to Abraido-Lanza, Dohrenwend, Ng-Mak, Turner (1999), however, the salmon effect cannot explain the Hispanic paradox. They tested the salmon hypothesis by analyzing National Longitudinal Mortality Study data between 1973 and 1985 from Cubans, who at the time of data collection were not able to return to their country of origin for political reasons; US-born individuals of Latin American ancestry, who have not migrated themselves and live in the country they are born in; and Puerto Ricans, the deaths of whom are already reflected in the US census. Abraido-Lanza et al. (1999) asserted that for Cubans and US-born Latina/os, since there was no question of return, a salmon effect would not have been possible. The results reflected lower mortality in Cubans and Puerto Ricans, as well as US-born Latina/os compared to non-Latina/o White population, indeed indicating no support for the salmon effect.

Turro and Elo (2008) nevertheless reported the existence of a salmon effect in the Latina/o population, based on newer data extraction opportunities. By using Social Security data, most notably, the beneficiary records between 1995 and 2000, they maintained to capture all deaths between these years. Turro and Elo (2008) argued that since the Social Security Administration performs stewardship duties with seriousness, near 100% ascertainment of deaths inside and outside of the United States is upheld for primary beneficiaries. Using this method, they found support for the salmon effect not only in the context of the Hispanic paradox but also for foreign-born non-Latina/o whites, indicating that this methodological error is not only seen in the documentation of immigrant populations but in emigrant deaths as well. They noted, however, that the magnitude of their finding did not warrant the full attribution of the Hispanic paradox to the salmon effect.

One other method to test the salmon hypothesis is to reach return migrants themselves. Palloni and Arias (2004) conducted a comparative study by using the dataset of the Mexican Health and Aging Study (MHAS), which asks residents of Mexico about migration status and history, and the United States–based National Health Interview Survey (NHIS). They were thus able to compare the self-reported health of Mexican-born individuals over the age of 50 residing in the United States and age-matched return migrants in Mexico, whose return occurred in the previous ten years. They extracted three samples: (a) all Mexican-born individuals in the NHIS, (b) all Mexican-born individuals in the NHIS not matched to death records before the follow-up midpoint, and (c) all Mexican-born individuals in the NHIS not matched to death records in the full follow-up period. The distribution of health in the third sample, which was assumed to include the highest number of return migrants, was expectedly better than other samples, providing further support for the salmon effect. Among the cases represented in the MHAS, return migrants had the worst self-reported health when compared to NHIS. Although Palloni and Arias (2004) cautioned against taking these results as decisive, the trend they illustrate indicated that national databases possibly lack information about return migrants, and comparative measurements are crucial in studying migrant mortality.

A recent study by Diaz, Koning, and Martinez-Donate (2016) attempted to illuminate the salmon effect by surveying Mexican-born men as they crossed the border from the United States to Mexico, that is, during return migration. They compared data from return migrants and deported individuals with Mexican-born individuals who remained in the United States. Their findings revealed higher level of reported stress and health restrictions among return migrants compared to migrants who remained. Overall bad health and chronic health conditions, however, were not less likely to be reported by return migrants and deported migrants, contradicting the salmon effect. The study also showed that migrants were more at risk to be deported if they are in good health, that is if they are “healthy migrants,” adding to the already abundant paradoxes in the literature. Diaz et al. (2016) concluded that since deportees are essentially healthier than those who remain or return voluntarily, the health advantage observed in the Latina/o population would even be stronger had there been no policy of deportations.

Razum et al. (1998) used death registry data and population estimates and found, in consistency with the U.S. literature, significantly lower all-cause mortality, compared to the majority population in Germany, as well as the urban population of Ankara, Turkey. The possibility of this finding being due to a healthy worker effect was ruled out as the Turkey-born residents in Germany were comprised of workers as well as their families. The mortality advantage was also too large to be explained by return migration, or over-registration (Razum, Zeeb & Rohrmann, 2000). Razum et al. (1998) then proposed another interpretation of the salmon effect in explaining their results, namely the “unhealthy re-migration hypothesis.” They suggested that there is an ongoing selection process: migrants who cope well with the socioeconomic conditions of the host country and the stresses of migration stay; migrants who have difficulty coping return migrate, perhaps before the long-term effects of stress manifest. Those who stay, then, are self-selectively healthier. Qualitative findings partly support this hypothesis: the perceptions that Germany’s climate is detrimental to one’s health, that the lack of social integration in Germany is a source of severe stress, and that the labor conditions are tough were all listed as reasons to return migrate to Turkey (Razum, Sahin-Hodoglugil, & Polit, 2005). This hypothesis, of course, warrants consideration of health to incorporate mental health and resilience, which will be discussed later on in the article (see section on “Mental Health.”).

The salmon effect at least partly explains the paradox of healthy migrants in some populations and in some datasets (but not in others) and is subject to unique selection patterns. In cases where the health advantage of migrants is confirmed, another question remains: among migrants who remain, does this trend of healthy migrants persist over time spent in the host country?

Convergence Theory: Do Migrants Stay Healthy?

Convergence theory holds that as migrants stay in the host country and become acculturated, they adopt the health behaviors of the majority population and their health erodes (or improves) to converge to that of the majority profile. Acculturation indicates a course of change whereby the contact between the host and migrant cultures initiates a dynamic that impacts the individuals’ attitudes and behaviours. The individual adopts new ways of being, while participating in the changes of their culture of origin (Berry, 1995). Razum and Twardella (2002), as indicated before, conceptualize this process as part of a “health transition” when the advantages of having arrived from healthy lifestyle practices are gradually overturned by introduction to “western” life practices such as sedentariness, consumption of processed food and cigarette smoking.

One of the first studies observing this process was the “Ni-Hon-San” epidemiological coronary heart disease study conducted with Japanese individuals in Japan (the “Ni” for Nippon), Japanese individuals who had migrated to Honolulu (the “Hon”; close to the country of origin), and Japanese migrants in San Francisco (the “San,” farther from the country of origin) (Marmot et al., 1975). Using data from this study, Marmot and Syme (1976) found a gradient for Japanese people, with the lowest coronary heart disease risk in Japan, intermediate risk in Hawaii, and highest risk for immigrants in San Francisco. It was possible to conclude that (1) migrants brought risks from countries of origin (i.e., past exposures in the country of origin were relevant); (2) health-related risks could change during the migration process; and that (3) this was not only an issue of laboratory values—social processes also played a role (Marmot & Syme, 1976). Although not reflected in this study, the health of migrants relative to the majority population of their new home country was possibly affected also by selection effects during migration and by access to health services.

Other studies have tried to explore convergence by observing weight gain and risk behaviors as markers of health deterioration. Goel, McCarthy, Phillips, and Wee (2004) tracked BMI and found that duration of residence in the United States was associated with an increase in BMI for migrant populations, especially after the 10-year mark. A similar effect was also observed in migrant children: in Canada, professionally measured annual BMI increase of second-generation migrant children was closer to native-born children, whereas first-generation children enjoyed a protective advantage (Maximova, O’Loughlin, & Gray-Donald, 2011). An important finding by Goel et al. (2004) was that despite the erosion of diet and gradual increase in weight, the migrant population was less likely to consult with their physicians on the issues of diet and exercise than their U.S.-born counterparts. This implies that dietary counseling with migrant populations may have been neglected, perhaps due to the assumption that migrant populations inherently have good dietary practices. In line with diet, a number of studies documented that smoking prevalence among Turkish migrants in Germany converged to that of the German majority populations with duration of stay: this meant a health disadvantage for women (whose smoking increased) and a health advantage for men (whose smoking frequency decreased) (Reeske, Spallek, Razum, 2009; Reiss, Breckenkamp, Borde, Brenne, David, & Razum, 2015; Reiss, Schunck, & Razum, 2015).

When the relationship between acculturation and heath behaviors is directly analyzed, a mix of increase in risky as well as preventative health behavior is likewise reported. In the Latina/o population in the United States, for example, acculturation was linked to an increase in the likelihood of high alcohol consumption and high BMI in both women and men, an increase in smoking, and also physical exercise in women (Abraido-Lanza, Chao, & Florez, 2005). Interviews on risk behaviors and acculturation experiences conducted both with Mexican women who migrated to the United States and women who remained in Mexico revealed that U.S. residency was associated with having more than one lifetime sexual partner and higher alcohol consumption. The effects were marginal, however, due to small sample sizes (Hennessy-Burt, Stoecklin-Marois, Meneses-González, & Schenker, 2011). Another interview study with pregnant Latina women reported acculturation to be related to sexually risky behavior, substance use, fast food consumption, and low fruit intake (Kasirye et al., 2004).

Yet residency is a not a static process as assumed in most studies; transnational individuals tend to have dynamic ties with both countries. Ortmeyer and Quinn (2015) considered cumulative migration duration as well as repeated single-trip migration experiences (which indicate different levels of acculturation) to arrive at a better understanding of convergence. They analyzed data from Mexican Americans who repeatedly visit their countries of origin. The analysis revealed that as the trips back and forth (i.e., cumulative experience in the United States) increased, migrants’ health deteriorated, supporting the convergence theory. However, the probability of having deteriorated health decreased the longer in duration the last trip was. Ortmeyer and Quinn (2015) attributed the latter finding as a recovery effect to compensate for the border-crossing experience. Furthermore, they reported that having non-Latina/o white friends was negatively associated with improved health outcomes, while having Latina/o friends was positively associated with improved health outcomes, providing support for the convergence theory, as well as for the Barrio advantage.

In a way, it is possible to say that convergence theory represents the loss of a Barrio advantage. Adopting the behaviors of the new population may indicate a gain of a community but at the same time, a loss of another community, hence the complexity of findings. Acculturation, manifested as psychological acculturation, will next be discussed in further detail in the section “Mental Health.”

Difficulties of Conceptualizing and Defining the Target Group “Migrants”

In discussions of the salmon effect, the convergence theory, or other explanations of the paradox of healthy migrants, what is inherent but not always explicitly articulated in migration and health literature is the variation within and between country practices regarding demographic decisions in describing migrants or individuals with a migration background (Villalonga-Olives & Kawachi, 2014). In the European Union, for example, nationality, citizenship, and country of birth have all been used to describe migrants (Rechel, Mladovsky, Ingleby, Mackenbach, & McKee, 2013). It is additionally not clear when one stops being a migrant: a person whose grandparents migrated to the country of residence could be considered part of the general population in one country and a citizen with a migration background in another.

Migrants, as a group, are neither homogenous nor do they face the same conditions in host countries they select or were selected into. Furthermore, each host country is unique in its selection process, rates of internal and external migration, history of (and in turn attitudes toward) migration, migrants’ socioeconomic mobility opportunities, health policies regarding migrants, to name a few. In a critical study, Malmusi (2015) documented better self-reported health of migrants living in countries with a “multicultural/individualistic-civic” orientation, with a path to citizenship through birth and residence, than those of migrants in countries with a “differential exclusionist/collectivistic-ethnic” orientation, which grounds citizenship on ancestry and features higher health inequalities between migrants and the native population. Compared to living in countries with an “assimilationist/collectivistic-civic” orientation, where citizenship through residency is possible (but experiences of culture of origin are seen as a private matter) living in multicultural countries also indicated a health advantage (Malmusi, 2015). The impact of country policies on health has also been shown to extend to mortality (Ikram, Malmusi, Juel, Rey, & Kunst, 2015). These findings justify distinct considerations of health of migrants exposed to different migration realities across diverse social and geographical contexts.

Impact of Selection: Refugee and Non-Refugee Migrants

Australia is a country with high immigration: 28.5% of the population was born overseas (ABS, 2016). The country offers a number of paths of migration, which makes it a good field for studying the relationship between health and forms of migration. Chiswick, Lee, and Miller (2008) considered the self-reported health outcomes of different forms of selection processes and found that the recipients of independent and business skill visas—that is, those selected for their potential—reported the best health, and the recipient of humanitarian visas reported the worst health among participants. The self-reported health of all visa recipients deteriorated over three years in Australia, regressing to the mean (i.e., approaching or converging to the group average), eliminating significant differences between visa types, except for recipients of humanitarian visas, whose self-reported health remained poor.

Canada is also a country characterized by high immigration; over 20% of the population is foreign born (Statistics Canada, 2013). An important national cohort study revealed that the type of migration was related to health outcomes, with different outcomes for non-refugee migrants and refugees who arrived in Canada between 1980 and 1990. Overall, although both non-refugee and refugee migrants enjoyed lower mortality than the general population, refugees had higher risk of death than non-refugee migrants. Furthermore, the health status of refugees did not change in the years spent in Canada in contrast to non-refugee migrants, for whom there was a risk increase, a convergence (DesMeules et al., 2005). More specifically, refugees had higher risks of mortality from cardiovascular, respiratory, and infectious diseases, and violence/accidents/poisoning but lower risk for diabetes than non-refugee migrants (DesMeules et al., 2004).

Internal and Within Europe Migration

Sweden is similarly a country with a substantial number of immigrants, with approximately 17% of the population foreign born and 27% of the population having a migration background (Statistics Sweden, 2017). Internal migration is also dynamic, with residents of larger but sparsely populated northern Sweden more likely to move to the more populated southern Sweden. Andersson and Drefahl (2017) analyzed movements of the outmigrants and return migrants from northern Sweden (Norrland) to southern Sweden by following all residents born in the north from 1954 onward. The findings provided evidence for a salmon effect in that return migrants to Norrland displayed higher mortality regardless of gender marital status, and level of education. The salmon effect was stronger for unmarried and less-educated men (Andersson & Drefahl, 2017).

A considerable proportion of the immigrant population in Sweden is from northern Europe, offering a unique perspective on within-EU migration health outcomes. Despite analogous health standards, a study comparing migrants’ mortality with the mortality rates of birth countries revealed that migrants from Germany and Denmark and migrant men from Norway enjoyed lower all-mortality risk than the populations of birth countries. Migrants from Finland and migrant women from Norway, however, were not significantly different from country of origin in terms of mortality. Furthermore, in migrants from Germany, and migrant men from Norway and Finland, coronary heart disease mortality risk was lower than the country averages of Germany, Norway, and Finland. Considering that among all the countries studied, Sweden had the lowest mortality and highest life expectancy, it is possible to conclude that a convergence to better health is more common in migration within industrialized countries (Gadd, Johansson, Sundquist, & Wändell, 2006).

Mental Health

The discussion so far indicates that physical health is significantly impacted by the migration experience. However, as the challenges of migration go beyond physical hardship, whether the paradox of healthy migrants extends to mental health is a valid question. Although epidemiological studies concerned with the paradox of healthy migrants often overlook mental health status of migrants, a limited number of findings suggest that individuals with a migration background may enjoy better mental health than the majority population.

In the United States, mental health difficulties are generally experienced at a higher rate by white individuals than people of color. Budhwani, Hearld, and Chavez-Yenter (2015) explored whether this phenomenon was a demonstration of a healthy migrant effect or rather the result of a healthier lifestyle. The results showed foreign-born Asian, Latina/o, and Afro-Caribbean individuals experiencing lower rates of lifetime major depressive disorder than American-born Asian, Latina/o, and Afro-Caribbean individuals. The effect remained after adjusting for socioeconomic and demographic variables. Among all variables in the study, nativity was the primary factor protecting against major depressive disorder, with foreign-born Afro-Caribbean individuals displaying the lowest level of depression, which suggested a healthy migrant effect for mental health (Budhwani et al., 2015). Such findings may of course suggest an underutilization of services, utilization of support networks instead, and services’ lack of cultural familiarity (Markides & Coreil, 1986).

This is by no means a universal finding. Other studies imply migration to be associated with psychosis, affective disorders, and addictive behavior (for an extensive discussion of the relationship between migration and health, see Bhugra, 2004). The type of migration (whether forced or not), pre-migration and post-migration experiences, documentation status, social and economic inequalities experienced in the host country, age group, sexual orientation and gender identity all impact migrant mental health within multilevel and complex interactions, just as they do physical health (Bhugra et al., 2014).

An illustration of this complexity is reflected in a systematic review of the mental health of war refugees in line with the physical health findings previously discussed. Bogic, Njoku, and Priebe (2015) reported that after five or more years following displacement, refugees displaced as a result of war or other organized violence had higher incidences of posttraumatic stress disorder (PTSD), other anxiety disorders, and depression than the general population, with exposure to traumatic incidences, post-migration stress, and poor post-migration socioeconomic status indicating poor mental health outcomes in terms of depression. Highest rates of PTSD, anxiety, and depression were observed in refugees from Cambodia and former Yugoslavia, whereas the lowest rates were observed in refugees from the Middle East, Sub-Saharan African countries, and Vietnam. In countries where refugee resettlement policies are generally more positive, such as Australia and Canada, the health paradox was observed, with mental health outcomes being more positive than the general population (but not in Europe or the United States, where policies tend to be stricter) (Bogic, Njoku & Priebe, 2015).

The paradox of healthy migrants, although partially extending to mental health, therefore does not universally point to migrants enjoying better mental health. A helpful concept to make sense of the differences between findings is psychological resilience: the capability to adapt to stressful and traumatic circumstances. Psychological resilience is believed to be a robust protective factor against developing stress-related mental health difficulties (Rutter, 1987). A recent systematic review of resilience in the context of forced migration found resilience to be strongly associated with high-quality social and familial support; and resilience, not only on individual but also community levels, was positively associated with positive mental health outcomes (Siriwardhana, Ali, Roberts, & Stewart, 2014). To illustrate, a study from Spain showed subjective well-being to be related to self-esteem, social support, and social community integration but not to sociodemographic characteristics among Latin American immigrants (Herrero, Fuente, & Gracia, 2011).

Berry’s (1995, 1997, 2001) theory of psychological acculturation attempts to contextualize how psychological resilience may be bolstered by the choices migrants make in negotiations with host and original cultures. He describes four psychological acculturation strategies that can be adopted by migrants: (a) assimilation, when individuals do not maintain cultural identity of origin and instead seek contact with the host culture; (b) separation, when contact with the host culture is avoided and cultural identity of origin is maintained; (c) marginalization, when neither contact with the host culture nor maintenance of culture of origin is sought; and (d) integration, when both cultures are incorporated into one’s identity (Berry, 1997). Congruence between a migrant’s expectation of acculturation and actual experiences, perception of control over acculturation experiences, pre-contact preparation regarding the host culture, and positive contact with the host culture, all indicate positive mental health outcomes (Berry, 1995). Integration, or biculturalism, appears to be the most protective strategy for mental health for Latina/o adolescents (Lawton & Gerdes, 2014). Separation and marginalization, on the other hand, are the strategies associated with poorer mental health for German adults with a Turkish migration background (Brand et al., 2017).

It is possible that familism (broadly defined, the practice/ideal of putting the family’s needs before that of the individual) inherent in the Latina/o and Mediterranean experiences further explain the mental health advantage of the Latina/o population in the United States and Mediterranean migrants in Europe. A psychoanalytic addition to cross-cultural studies helps deepen the relationship between familism, resilience, and acculturation. Roland (1987, 1988) argues for a supraordinate self-organizational system consisting of (a) the individualized self, a product of Western/Northern cultural and nuclear family experience; (b) the familial self, a product of Eastern/Southern cultural and extended family experience that allows little room for individuation; (c) the spiritual or transcendent self, the attempt toward which allows for some individuation through the private nature of some rituals while retaining a deep connection to the community; and (d) the expanding self, which is an adaptive reaction to being exposed to both Western and Eastern values. The familial self helps individuals manoeuvre the hierarchical relational systems as opposed to the Western individualized self, which negotiates egalitarian relational structures (Roland, 1980, 1987, 1988).

When the individual is exposed to both individualistic and interdependent cultures, usually as a result of urbanization and/or migration, as in the Latina/o or Mediterranean cultures, an expanding self may develop, which “represents a growing individuation of the self” while retaining familial values (Roland, 1988, p. 6). The expanding self promotes the psychological survival of the individual, assisting in adherence to individualistic values in contexts when such values are preferred and adherence to familial values when it is more advantageous. Fişek (2002) argued that being exposed to different cultures means being exposed to different relationship governing rules along with different ways of being-with-other. The individual is likely to emerge from different cultural exposure experiences with a sense of expanding self but perhaps also with feelings of ambivalence and contradiction. As families converge to Western norms, hierarchical control that the authority figures have over the members of the family will become less strong, autonomy and separation will be tolerated to a greater extent while emotional proximity is maintained (Fişek, 2002). In other words, capacity to extend allows integration to be cultivated as an acculturation strategy.

Given the available data and related theorizations, it can be argued that expanding self, acculturation, and psychological resilience may be possible links between migration experiences and health outcomes. In fact, Schunck, Reiss, and Razum (2015) showed that perceived discrimination had an effect on physical health through mental health. The relationship Malmusi (2015) documented between migration policies and self-reported health, for example, is most likely moderated by psychological perception of acceptance. DesMeules and colleagues’ (2004, 2005) documentation of health differences between refugee and non-refugee migrants, similarly cannot be considered as separate from Bogic et al. (2015) findings regarding refugee mental health. Culturally informed psychosocial support programs, reinforcing the psychological acculturation strategy of integration, or the development of expanding self—and at the same time targeting an increase in the capacity for psychological resilience—would then be expected to promote better physical and mental health for migrants.

Migrant and Refugee Movements in the Early 21st Century

The world is experiencing rapid human movements. This is an inevitable characteristic of a global era but also an acute outcome of several civil wars in the Middle Eastern and African regions, as well as major inequalities between nations. Data about health prospects of forced migrants to Europe, who became refugees and asylum seekers during the Syrian civil war, are yet to come. What we know from anecdotal sources and news stories might suggest a strong self-selection bias. The 2015–2016 patterns of refugee migration to Europe, faced with closed borders, involve crossing the Aegean and the Mediterranean seas in treacherous conditions and in some cases, passing through countries by foot for weeks. The ordeal with human smugglers, who may ask for fees that families may not be able to come up with for all members of the family, generates situations whereby the families pool all resources to fund the passage of the strongest members.

Going back to Swerdlow’s (1991) findings, the cohort he studied—the refugees from Vietnam who traveled to neighboring countries in Southeast Asia in treacherous conditions—closely resemble the refugees arriving in Europe from the Middle East and North Africa in the early 21st century. Both cohorts further experienced an isolating pattern of settling in refugee camps and then resettling in disparate parts of the host countries (Ostrand, 2015; Swerdlow, 1991). The similarity in patterns deems it possible to signal a similar mortality outcome. Yet other reported findings (Bogic et al., 2015; Chiswick et al., 2008; DesMeules et al., 2004, 2005) caution us against optimism. Although refugees may experience lower rates of mortality, the closed border policies and the negative arrival they encountered during their flight, which may lead to developing the acculturation strategies of separation or marginalization, may in turn translate into a disadvantage when compared to non-refugee migrants.

We will not know if and/or when the ongoing civil wars will quieten, so it is not possible to predict whether a salmon effect will be observed in current refugees (and hopefully, future residents of host countries )or whether their health with converge to the mean of host population. However, the unfortunate opportunity to follow individuals on their path to become acculturated to new societies will not only enrich our understanding of the relationship between migration and health, but it will also contribute to the acculturation process by generating advocacy for inclusive health care.

Remaining Questions

This article attempted to put the persistent finding of healthy migrants into perspective by focusing on various explanations of the phenomenon, with specific emphasis on the salmon effect and the convergence theory. The phenomenon was further situated within discussions of cross-cultural differences, variances in policies and migration orientations, mental health, and specificity of migrant and refugee movements. The authors recognize the geopolitical limitation of this article in the sense that it mostly focuses on immigration to or immigration within Western countries. Immigration between or within low-to-middle income countries are not reflected here. It should also be noted that only literature in English and Turkish were reviewed in preparing this manuscript. Given the historical emergence of the healthy migrant phenomenon, important findings in Spanish and Portuguese may have been neglected.

Future Directions for Migrant Health Research

Migrant populations frequently experience better health outcomes than the majority population of the receiving country, and this health advantage may pertain for several years—this is the most important message to policymakers and the general public of receiving countries. For public health researchers and epidemiologists, challenges remain. The salmon effect and the convergence theory, taken together, illuminate several gaps in research and policy: (1) It is highly challenging to compare migrant health data and arrive at an understanding of the paradoxes of migrant health due to varying definitions of migrants and diverse study designs to measure mortality and morbidity; (2) host countries do not always track immigrants’ and emigrants’ mobility patterns, which leads to significant errors in census data; (3) in the case of return migration, the motivations for going back to the country of origin are not fully understood, and elective transnational lifestyles are not always possible as a consequence of restrictive migration policies; (4) despite significant evidence for the loss of heath advantage and convergence to the host population’s health, policies do not promote the maintenance of protective health behaviors.

An ideal migrant research cohort would need to enroll participants in the country of origin and before the decision to migrate is taken. This ideal cohort, although possible to implement, may require focused deliberation and societal acceptance due to data protection issues (Razum, 2006). Comparisons, when pursued, should be made between migrant populations and their peers in the country of origin, as well as between migrants with different statuses (Rechel et al., 2013). A life-course approach, specifying a time axis—thus following individual pre-migration, arrival, post-migration, and if applicable, resettlement experiences and related health predictors and outcomes (and incorporating ecological exposures)—should nevertheless be the way forward (Spallek, Zeeb, & Razum, 2011). Qualitative studies, informed by sociological and psychological perspectives, should also be part of this approach to achieve deeper understanding of social, cognitive, and affective factors influencing health behaviors as time is spent in the host country (Khlat & Darmon, 2003). Roura (2017) advocated for a transdisciplinary research agenda whereby participatory, co-creative, iterative, and inductive practices inform and connect social determinants of health and migrant health research.

Perhaps in original research, systematic reviews, and meta-analyses it would make more epidemiological sense to examine the phenomenon of healthy migrants in terms of migration type and country policy. In order to resolve the paradox of health migrants, researchers should separately focus on the health outcomes of different categories of migrants situated against host countries’ migration orientation. As the available research suggests, the health outcomes of asylum seekers and refugees fleeing war and political instability is different than those of individuals migrating to further professional or financial attainment. Given the ever-changing political climates around the world, we propose two critical questions for researchers studying the paradox of healthy migrants with specific focus on the salmon effect and the convergence theory: Do forced migrants/refugees/asylum seekers/political exiles who cannot return to the country of origin show a healthy migrant effect? And what is the impact of access barriers and entitlement restrictions that are enforced in refugee health care on convergence to the population mean?

Further Reading



  • 1. When the phenomenon was first described, “Hispanic” was the preferred term to refer to Americans born in Spanish-speaking countries. The term was later replaced by the broader term “Latina/Latino” to include Americans of all Latin American ancestry. As the phenomenon of lower mortality among migrants was first described in the United States in the 1980s, when the term Hispanic was in use, the phenomenon is referred to as the “Hispanic paradox.” Palloni and Arias (2004) further argue against the use of “Hispanic” to describe this phenomenon, as most of the data used to arrive at this finding contain participants of “foreign-born Other Hispanics and foreign-born Mexicans—not of Puerto Ricans or Cubans” (p. 409). When the term Hispanic is used in the article, it is done so to accurately represent the study described.

  • 2. “Healthy migrant effect” and “healthy immigrant effect” are often used interchangeably in migration literature. The term is also used to refer to both the phenomenon that migrants have a mortality advantage in general (e.g., Domnich, Panatto, Gasparini, & Amicizia, 2012; Vang, Sigouin, Flenon, & Gagnon, 2015), and also to describe the self-selection explanation in that only the healthy individuals migrate (e.g., Franzini, Ribble, & Keddie, 2001; Razum, 2006). This article will use “healthy migrant effect” throughout for the purposes of consistency and will refer to the latter description.