Socioeconomic and demographic risk factors in COVID-19 hospitalization among immigrants and ethnic minorities

Abstract Background Immigrants and ethnic minorities have been shown to be at increased risk of hospitalization from COVID-19. Our aim was to analyse the contribution of socioeconomic and demographic risk factors on hospital admissions for COVID-19 among immigrants and ethnic minorities compared to the majority population. Methods We used nationwide register data on all hospitalized COVID-19 cases between February and June 2020 (N = 2232) and random controls from the general population (N = 498 117). We performed logistic regression analyses and adjusted for age, sex, comorbidity, and socioeconomic and demographic factors. The main outcome measure was hospitalization with COVID-19 and was estimated using odds ratios (OR) and 95% confidence intervals (95% CI). Results Among 2232 COVID-19 cases, the OR of hospitalization with COVID-19 among immigrants and descendants of non-Western origin was 2.5 times higher (95% CI: 2.23–2.89) compared with individuals of Danish origin with most pronounced results among individuals from Iraq, Morocco, Pakistan and Somalia. The OR was largely attributed to comorbidity and socioeconomic factors, especially household size, occupation, and population density. Conclusion There is a significantly higher OR of hospitalization with COVID-19 among non-Western immigrants and ethnic minorities compared with ethnic Danes. This knowledge is crucial for health policymakers and practitioners in both the current and future pandemics to identify more vulnerable groups and target prevention initiatives.


Introduction
M igrants and ethnic minorities are reported to be at increased risk of acquiring COVID-19 and to have worse clinical outcomes of COVID-19 compared with the majority population. [1][2][3][4][5] National statistics from Denmark and other Scandinavian countries correspondingly show that the incidence of COVID-19 cases is higher among individuals of non-Western origin, particularly of Middle Eastern origin, compared with individuals of Danish and other Western origin. 6,7 In 2020, individuals of non-Western origin made up 58% of immigrants in Denmark, and among descendants, 83% were of non-Western origin. 8 Therefore, the focus of this study is on individuals of non-Western origin, as they make up the majority of immigrants and have a higher incidence of COVID-19 than individuals of Western origin.
Research on previous pandemics, including the 1918 Influenza and the 2009 A/H1N1 Influenza, likewise shows immigrants and ethnic minorities as disproportionately affected compared with majority populations. 9,10 This is hypothetically explained by various factors related to socioeconomy, language barriers, cultural norms and comorbidity in combination with poorer access to healthcare. [9][10][11] First, lower income among immigrants and ethnic minorities may result in more people living together in closer proximity: data from Denmark show that the proportion of COVID-19 cases of non-Western origin more often live !5 individuals at the same address compared with ethnic Danes and individuals of Western origin. 6 Furthermore, some immigrants and ethnic minority groups may have a culture of multi-generational living, which challenges social distancing and exposes older individuals with comorbidities to be at higher risk of COVID-19. 11 In addition, multi-generational living may as well be a function of socioeconomic disadvantages besides being a cultural phenomenon. 12 Second, many immigrants and ethnic minorities are more vulnerable to cardiovascular disease, diabetes and hypertension, which are all associated with more severe [13][14][15] Third, compared to ethnic Danes, individuals of non-Western origin are more likely to work 'frontline' jobs within, e.g. transportation, hospitality, cleaning and operational services, and social welfare institutions; these jobs typically preclude working from home and are associated with a higher prevalence of COVID-19 across populations. 6,16 Finally, language barriers may account for higher vulnerability among immigrants and ethnic minorities if, for example, guidelines on COVID-19 from the national and regional health authorities are not adequately translated and communicated.
These associations all make immigrants and ethnic minorities more vulnerable to COVID-19. Some of these associations have already been documented in reports and scientific papers from England and the USA. [3][4][5]17 However, further studies are needed from other European immigration countries, which have different compositions of immigrants and ethnic minorities as well other contextual factors at play. Furthermore, adults of non-Western origin have a higher morbidity, but a lower mortality compared with individuals of Danish origin. 18 Yet, no national health initiatives have been implemented in order to promote the health status of individuals of non-Western origin. 18 Moreover, large-scale registerbased studies are needed to disentangle the contribution of socioeconomic, demographic and comorbidity factors on COVID-19 hospitalization. Therefore, the aim of this study is to analyse the contribution of different socioeconomic and demographic risk factors, including occupational status, income, education, household size and population density, on hospitalizations for COVID-19 among immigrants and ethnic minorities compared to the majority population.

Study population
The study population is based on individuals residing in Denmark aged >17 years on 1 January 2020. Cases are identified by being registered as inpatients (admitted for !24 h) with a COVID-19 diagnosis of B34. 2

Country of birth
The study population is categorized using country of birth (Danish/ Immigrant), or mother's country of birth (Descendant) as a proxy hereof, to define individuals' ethnicity. Ethnicity is defined in three ways as: (i) origin categorized as Danish (a person of Danish origin has at least one parent, who is a Danish citizen born in Denmark), immigrant (an immigrant is born in a foreign country and none of the parents are Danish citizens born in Denmark), or descendant (a descendant is born in Denmark and none of the parents are Danish citizens born in Denmark); (ii) grouped country of birth categorized as Danish (Denmark), Western (all EU countries, Andorra, Australia, Canada, Iceland, Liechtenstein, Monaco, New Zealand, Norway, San Marino, Switzerland, USA or the Vatican) and non-Western (Albania, Bosnia and Herzegovina, Belarus, Kosovo, North Macedonia, Moldova, Montenegro, Russia, Serbia, the Soviet Union, Turkey, Ukraine, former Yugoslavia, all countries in Africa, Southand Central America, Asia, Oceania with the exception of Australia and New Zealand, or stateless) 19 ; and (iii) seven specific countries of birth, where a minimum of 10 COVID-19 cases are identified to compare with individuals born in Denmark (Denmark, Iraq, Lebanon, Morocco, Pakistan, Somalia, Turkey or former Yugoslavia). Information on country of birth is obtained from Statistics Denmark. 20 For the descriptive analyses, immigrants and descendants are separated, and for the regression analyses, immigrants and descendants are considered as one group. Individuals are categorized into Western and non-Western countries of birth without distinguishing between being immigrant or descendant. The length of residence is also included for immigrants as a proxy for integration with a range from 0 to 34 years of residence and was categorized as 5; 6-15; 16-29; !30 years.

COVID-19
National register-based hospital data on cases of COVID-19 are obtained from The National Patient Register, which includes information on diagnoses whenever an individual is in contact with any hospital department by inpatient, outpatient, or emergency room visits. 21 Data are obtained on all individuals registered with the following ICD-10 COVID-19 diagnoses: coronavirus infection, unspecified site (ICD-10: B34.2; B34.2A), or coronavirus as the cause of diseases classified to other chapters (ICD-10: B97.2, B97.2A). 22 To increase the validity of COVID-19 diagnoses, COVID-19 cases are defined as inpatients hospitalized for at least 24 h and only with COVID-19 as the primary reason for hospitalization.

Socioeconomic and demographic factors
Data on socioeconomic and demographic factors are obtained from Statistics Denmark. 20 The covariates include: (i) marital status: unmarried vs. married; (ii) highest-attained educational level based on International Standard Classification of Education (ISCED) 23 : low (primary school; ISCED levels: 0-2); medium (upper secondary education, business high school, vocational education; ISCED levels: 3-4); and high (short-term further education, middle-range education, bachelor's, extended education, research degree; ISCED levels: 5-8); (iii) persons per household: range: 1 to !5 persons; (iv) occupation defined by either business industry or occupational status for those, who are unemployed or with unknown business industry: (a) business industry: farming, forestry and fishery; manufacture, raw material extraction and supply company; building and construction; trade and transportation; information and communication; financing and insurance; property trading and rental service; business service; public administration, education and health and social services; culture, leisure and other services; and (b) employed with unknown business industry (self-employed, collaborating spouse, salary earner); other occupational inactivity (retired persons, other economically inactive persons, who have no predominant income and welfare benefits are lower than the basic amount); and unemployed (unemployed person, unemployment benefit claimant,

Comorbidity
To understand the contribution of comorbidity on the OR of COVID-19 hospitalization among immigrants and ethnic minorities compared to the majority population, information is obtained on all ICD-10 diagnoses from The National Patient Register registered between 1980 and 2020. 21 Comorbidities are weighted by the Charlson Comorbidity Index (CCI) to create a continuous weighted comorbidity score. 24,25 The comorbidity scores assign weights from 1-6 to each comorbidity for every individual, which then receives a sum of weights based on all present comorbidities in an individual. A comorbidity score of 0 represents no comorbidities; higher scores indicate more severe comorbidities. The CCI is categorized into a score of: 0, 1, 2 or !3.

Statistical analyses
Descriptive analyses are conducted to investigate the distribution of comorbidity, socioeconomic and demographic factors among individuals with and without COVID-19.
For the main analyses, logistic regression models are used to calculate odds ratios (OR) of COVID-19 hospitalization and 95% confidence intervals (95% CI) among different ethnic groups. Different categorizations of ethnic groups are used to investigate whether the OR of hospitalization with COVID-19 differed depending on the definition of ethnicity. First, the OR of hospitalization with COVID-19 is investigated among individuals of immigrants and descendants, Western origin, non-Western origin and residence length of immigrants compared with individuals of Danish origin. Nine models are applied to adjust for several covariates, including age, sex, CCI, household size, occupation, population density, equivalized disposable income, and educational level. Further, based on the ethnic groups, who are most frequently hospitalized with COVID-19, the OR of hospitalization with COVID-19 is investigated among individuals of specific ethnic origins compared with Danish origin. In sensitivity analyses, it is investigated whether individuals of non-Western origin, compared with individuals of Danish origin, are at higher risk of hospitalization with COVID-19 by including interaction terms between origin and either occupational group (farming, forestry, fishery, manufacture, raw material extraction and supply company, business service, trade and transportation, public administration, education and social and health services, other occupational inactivity, unemployed), or household size (1 to !5 persons), respectively. In additional analyses, the OR of hospitalization with COVID-19 is investigated for different risk factors presented as unadjusted and adjusted OR estimates.
SAS Enterprise Guide version 7.1 is used to conduct all analyses including a statistical significance level of 0.05. The study is approved by the University of Copenhagen Ethics Board (reference number: 514-0231/18-3000). Individual consent is not required by Danish legislation for register-based studies.

Results
In this national register-based study, 2232 individuals are hospitalized due to COVID-19 in Denmark between February and June 2020. Table 1 shows the descriptive analyses. The median age at hospitalization with COVID-19 is 67 years with an overall equal distribution of men and women. More than 80% of hospitalized COVID-19 cases are of Danish origin. Yet, a higher number of COVID-19 cases per 10 000 persons are of non-Western origin or other origins than Danish. Out of the 2232 COVID-19 cases, 13 individuals (0.6%) are from the Nordic countries Finland, Norway and Sweden. Among immigrants, there is a higher number of COVID-19 cases where the individual's length of residence exceeds 30 years. Having a higher CCI shows a higher number of COVID-19 cases per 10 000 persons. The COVID-19 cases are not observed specifically in some business industries, but a higher proportion of COVID-19 cases is occupationally inactive, thus, either retired or economically inactive. There is also a higher number of COVID-19 cases in the provinces around Copenhagen.
The results of the main analyses ( The results of the analyses for specific countries show that the OR of hospitalization with COVID-19 is highest among individuals from Somalia, followed by Morocco, Iraq, Pakistan, Lebanon, Turkey and former Yugoslavia.
Overall, the results of the analyses on immigrants' residence length show that when comparing with immigrants who lived in Denmark for !30 years, the risk of COVID-19 hospitalization is lowest among immigrants who lived in Denmark 5 years, while the risk is higher among immigrants who lived in Denmark for 16-29 years. However, the results are not statistically significant for immigrants, who lived in Denmark between 16-29 years.
For the sensitivity analyses, specific groups of business industries or occupational activity are selected based on having prevalent COVID-19 cases among individuals of non-Western origin, and the analyses are grouped by these occupational groups using There are too few COVID-19 cases to be displayed among immigrants and descendants of Western origin. c: There are 6529 individuals without information on educational level. d: There are 139 671 individuals without information on business industry, but with information on occupational status (unemployed/ employed without known business industry/other occupational inactivity) and 3530 individuals without information on neither business industry or occupational status. e: Business industries of farming, forestry, fishery and manufacture, raw material extraction and supply company are combined into one group due to few cases in some industries. f: PD: population density.
Socioeconomic and demographic risk factors in COVID-19 hospitalization 305 Table 2 Odds ratio (OR) with 95% confidence intervals (95% CI) of hospitalization with COVID-19 among individuals grouped by regional origin, immigrants and descendants, and specific countries of birth living in Denmark, N      a: Farming, forestry, fishery, manufacture, raw material extraction and supply company are abbreviated as Farming and manufacture due to limited space in this table.
Socioeconomic and demographic risk factors in COVID-19 hospitalization 307 individuals of Danish origin as reference group (table 3). The results show that for all occupational groups, individuals of non-Western origin have a higher OR of hospitalization with COVID-19 in a range of ORs from 1.27-3.01 compared with individuals of Danish origin. In the analyses for household size, the OR of hospitalization with COVID-19 is investigated in individuals of non-Western origin compared with individuals of Danish origin (table 3). Among individuals of non-Western origin, the ORs of hospitalization with COVID-19 are observed to be 1.52-2.37 times higher compared with individuals of Danish origin. In the adjusted analyses where different risk factors are investigated separately in relation to the OR of hospitalization with COVID-19 (table 4), the highest OR's are observed in individuals of non-Western origin, the older age groups, those with higher CCI, those living five or more individuals together in one household, those working within public administration, education and health and social services, and those who are occupationally inactive.

Discussion
Using an exceptional nationwide cohort, findings of this study provide unique information on the contribution of socioeconomic and demographic factors to the higher COVID-19 hospitalization risk among immigrants and ethnic minorities compared to ethnic Danes. Immigrants and descendants of non-Western origin have a 2.5 times higher OR of hospitalization. About half of the excess risk is explained by comorbidity, socioeconomic and demographic factors.
This study complements emerging evidence from other European countries that immigrants and ethnic minorities are at higher risk of  31 Third, population density is an important factor to hospitalization risk, which is supported by a Swedish study, where neighbourhood population density together with income, employment and household members, explained COVID-19 mortality risk. 32 Finally, education and income do not markedly change the estimates, which could be because occupation, population density and household size work as a proxy hereof. As comorbidity, socioeconomic and demographic factors only explain some of the risk, it can only be speculated about the remaining contributory factors. Some explanations could be language barriers, 33 'newness', 34 limited health literacy 35 and genetic and biological susceptibility, 36 which may be driving mechanisms potentially interacting in complex, dynamic ways within groups of immigrants and ethnic minorities. The complex interplay between factors related to language barriers, health literacy, access in healthcare, migration background and changing migration patterns may as well explain why immigrants with shorter residence length are at lower risk of COVID-19 hospitalization, which was against our expectations, as residence length was considered a proxy for integration over the years.
The key strength of this study is its basis on unique nationwide data representing all hospitalized COVID-19 cases in Denmark from February to June 2020. Using national data made it possible to include a large study population and a random reference group, link different national registers and utilize retrospective and recent data on several comorbidity, socioeconomic and demographic confounders and link them to each individual. The validity of COVID-19 diagnoses is increased by only including cases, where COVID-19 is the primary reason for hospital contact.
This study also has several limitations. First, Denmark's management of the pandemic through restrictive policies resulted in a comparatively limited number of hospitalizations during the 'first wave' limiting the sample size of specific country groups and reducing statistical power in some analyses. Second, adjusting for CCI does not markedly change the results, yet, another Danish study finds an increased risk of COVID-19 among individuals with one or more comorbidities of the CCI. 37 However, using a comorbidity index constructed specifically for COVID-19 could be more relevant. Third, the study is limited by the availability of data of the basis for residence permit and missing data on business industry and educational level. Fourth, the study population only includes the most severe COVID-19 cases hospitalized for !24 h and is not representative of COVID-19 cases in the general population. Therefore, the findings of this study only apply to the most severe COVID-19 cases among immigrants and ethnic minorities.
In conclusion, there is a higher OR of hospitalization with COVID-19 among immigrants originating from non-Western countries compared to ethnic Danes. The excess risk is explained by socioeconomic and demographic factors. Further studies including qualitative approaches are needed to understand the underlying reasons for the observed disparities. The results also highlight immigrants and ethnic minorities as more vulnerable groups in terms of acquiring COVID-19, and that this tendency should be investigated among COVID-19 cases in the general population as well. This knowledge is crucial for health policymakers and practitioners in both the current and future pandemics to identify more vulnerable groups and target prevention initiatives.

Supplementary data
Supplementary data are available at EURPUB online.

Funding
This study was funded by University Hospital Hvidovre, University of Copenhagen and Statens Serum Institut.