Drinking Cultures and Socioeconomic Risk Factors for Alcohol and Drug Use Disorders among First- and Second-Generation Immigrants: A Longitudinal Analysis of Swedish Population Data

Abstract

Background

Few longitudinal studies investigate predictors of substance use incidence among immigrants. The current study describes substance use disorders in immigrants to Sweden, focusing on drinking culture in the country of origin and socioeconomic status (SES), and how these intersect with generational status to influence risk.

Methods

Using pseudonymized Swedish population registry data, we track onset of alcohol use disorder and drug use disorder in a longitudinal study of 815,778 first-generation immigrants and 674,757 second-generation immigrants from 64 countries over a 6-year period. Cox regression analysis estimated risks of alcohol and drug use disorders in second-generation immigrants compared to first-generation, and moderation analyses assessed interactions of generational status with country-of-origin per capita alcohol consumption and SES.

Results

Immigrants and second-generation immigrants originating from countries with high levels of alcohol consumption had higher risks for alcohol and drug use disorders. Immigrants with high SES had lower risks for alcohol and drug use disorders. The interaction between generational status and country-of-origin alcohol consumption was significant for drug use disorder (not for alcohol use disorder), with drug use disorder risk for second-generation immigrants being highest for those from countries with the lowest level of country-of-origin per capita alcohol consumption. The interaction between generational status and SES was significant for alcohol use disorder, with low-SES second-generation immigrants showing markedly higher risk than first-generation immigrants with comparable SES.

Conclusions

Among immigrants in Sweden, second-generation immigrants are at increased risk of developing alcohol and drug use disorders, particularly if they have lower SES. Policy and community attention to these high-risk subgroups in immigrant communities is warranted.

1. Introduction

Past research in the United States and Europe has mostly shown the prevalence of substance use disorders is lower among immigrants than among native-born populations (Markkula et al., 2017; Salas-Wright et al., 2014), albeit with some variations. Notably, differences from native-born populations are typically more marked for foreign-born (or first-generation) immigrants than second-generation immigrants. In the United States, prevalence of substance use disorders is highest among native-born Americans, slightly lower among second-generation immigrants, and considerably lower among first-generation immigrants (Mancini et al., 2015; Salas-Wright et al., 2014). Similarly, in Finland, alcohol use disorder incidence is lower among immigrants than among native Finns (Markkula et al., 2017). With longer time in Sweden, alcohol use disorder rates among young adult migrants converged with rates among native Swedes (Harris et al., 2019).

There are some exceptions to these patterns, however. second-generation immigrants had higher alcohol use disorder risk than the native population in France (Rolland et al., 2017). In Sweden, risk for alcohol-related hospitalization was higher among immigrants from Finland (Leao et al., 2006) but lower among those from other countries (Hjern and Allebeck, 2004), compared to the Swedish majority population. Immigrants to Nordic countries and North America from non-Western or other European countries with lower drinking levels show lower alcohol use disorder risk than in their new country (Markkula et al., 2017). Immigrants to Israel from the former Soviet Union were more likely to have alcohol use disorder than other Israelis (Shmulewitz et al., 2012). Migrants to Sweden from Asia and the Middle East had lower substance use disorder risks than migrants from Eastern Europe and Russia (Harris et al., 2019). Island Puerto Ricans had higher rates of alcohol and drug use disorders than Puerto Ricans in the US and the general US population (Canino et al., 2019). These studies suggest immigrants are highly heterogeneous, and their risks for alcohol and drug use disorders may vary depending upon drinking cultures both in their country of origin and their new country of residence.

In the current study, we investigate country-of-origin influence on alcohol and drug use disorders in immigrants, building on research that has demonstrated enduring influence of country-of-origin drinking cultures on drinking patterns among Asians (Cook et al., 2013; Cook et al., 2015; Cook et al., 2012) and Hispanics (Cook and Caetano, 2014) in the United States, immigrants to Taiwan (Chen and Chien, 2018) and from the former Soviet Union to Israel (Shmulewitz et al., 2012). These studies are mostly grounded in transnationalism theories, which suggest immigrants often maintain socioeconomic ties with their homelands and retain their cultural heritage, which may also appeal to second-generation immigrants (Portes et al., 1999; Schiller et al., 1995). Unlike socially-sanctioned alcohol consumption, cultural practices may not develop around illicit drug use, and we are unaware of credible data for country-level drug cultures for many countries. Still, drug use may be more tolerated in countries with high levels of alcohol consumption, and thus immigrants from such countries may have higher risk for drug use disorder as well. We test these hypotheses.

We also examine the role of socioeconomic disadvantage in alcohol and drug use disorders in immigrants. Although immigrants in high-income countries might be less disadvantaged than the native-born population overall, immigrants are diverse socioeconomically. Country-of-origin living standards and circumstances of immigration vary widely, with some immigrant groups having greater economic and human capital, and others arriving as refugees with scant resources. Socioeconomic resources ethnic groups command are integral to the well-being of immigrants and their offspring (Zhou and Xiong, 2005). The role of socioeconomic disadvantage in alcohol and drug use disorders has been documented in the broader literature. In the United States, drug use disorder is more prevalent among individuals with lower education and income (Grant et al., 2016) and those living in disadvantaged neighborhoods (Boardman et al., 2001; Karriker-Jaffe, 2011); Swedish data support a causal role of neighborhood disadvantage in drug use disorder (Kendler et al., 2014). Though not unequivocal, there is also evidence of increased alcohol use disorder risk due to disadvantage (Collins, 2016; Mulia et al., 2014). As the alcohol harm paradox thesis (Bellis et al., 2016; Lewer et al., 2016) suggests, although individuals with higher socioeconomic status (SES) tend to consume a greater volume of alcohol, individuals with low SES tend to have more harmful drinking patterns, potentially bearing greater burden of alcohol use disorder and other alcohol-related harms. Even with a substantial body of literature on links of lower SES to alcohol and drug use disorders, limited attention has been paid to these links for immigrants specifically. To address this gap, we test the hypothesis that immigrants with low SES will have higher risk for alcohol and drug use disorder onset compared to immigrants with higher SES.

We also investigate generational differences in risks for alcohol and drug use disorders among immigrants. We expect the influence of country-of-origin alcohol consumption on these risks will be greater for first-generation immigrants (likely more familiar with country-of-origin cultural practices) than for second-generation. Conversely, we expect the influence of disadvantage on alcohol and drug use disorders will be greater for second- than for first-generation immigrants. We also explore whether similar patterns are observed for men and women, given stark gender differences in substance use (McHugh et al., 2018; Zemore, 2007; Zilberman et al., 2003).

We use population registry data that include all immigrants of diverse origins to Sweden, a distinct advantage given that prior studies on immigrant drinking or drug use have mostly focused on immigrants from a small number of countries (Salas-Wright et al., 2014). These data also allow us to track the onset of alcohol and drug use disorders, rendering the current study to be of a rare longitudinal design in immigrant substance use research.

2. Methods

2.1. Data

Data were collated from several national registers including the Total Population Census, hospital discharge registers (Inpatient Register, 1964–2015; and Outpatient Register, 2001–2015), Prescription Register (2005–2015), Crime Register (1973–2015), and Crime Suspicion Register (1998–2005). Data were linked using the 10-digit civic registration number assigned to each person in Sweden upon birth or immigration. The entries were pseudonymized by replacing this number with a serial number to ensure anonymity. The study was covered by ethical approval from the Regional Ethical Review Board in Lund (Dnr: 2012/795). Census data were linked to the Multi-Generational Registry that identifies each individual’s mother and father (if known) and their place of birth.

The study population included everyone ages 15+ years in 2010 who was either a first-generation or second-generation immigrant. Individuals with unknown neighborhood information were excluded, as were individuals with alcohol or drug registrations prior to 2010. In total, 815,778 first-generation and 674,757 second-generation immigrants from 64 countries of origin were followed from January 1, 2010 until the first registration(s) for the focal outcome (alcohol or drug use disorder), death, emigration, or the end of the study period on December 31, 2015— whichever came first.

2.2. Measures

Alcohol use disorder was identified in the Inpatient and Outpatient registries by the International Classification of Diseases 10th version (ICD-10) alcohol-related codes (E24.4 pseudo-Cushing syndrome; G31.2 degeneration of the nervous system and brain; G62.1 polyneuropathy; G72.1 myopathy; I42.6 cardiomyopathy; K29.2 gastritis; K70 liver diseases, K85.2 acute pancreatitis, and K86.0 chronic pancreatitis, all caused by alcohol; O35.4 treatment of pregnant alcoholic woman; T51 toxic effects of alcohol; and F10 mental and behavioral disorders due to alcohol use, excluding F10.0 acute alcohol intoxication); in the Crime and Crime Suspicion registers by two codes reflecting alcohol use while driving or boating; and in the Prescribed Drug Register by prescriptions for disulfiram, acamprosate, or naltrexone. These represent relatively severe cases of alcohol use disorder.

Drug use disorder was identified in the hospital registries by relevant ICD-10 codes (F10–F19 mental/behavioral disorders due to psychoactive substance use excluding alcohol or tobacco); in the Suspicion Register by codes indicating arrests for driving under the influence of drugs and drug possession/use; in the Crime Register by convictions for narcotics and drug-related driving offenses; and in the Prescribed Drug Register by prescriptions for hypnotics, sedatives, or opioids in dosages consistent with drug use disorder (i.e., more than defined daily doses for 12 months), excluding cancer patients. About 23.5% of the alcohol use disorder cases also had registrations for drug use disorder.

Individual-level demographic variables

Country of origin was identified by using the father’s (or the mother’s if father’s was unknown). The country of origin of the foreign-born parent was used in the case of partial Swedish heritage.

Immigrant generational status was coded as first-generation if the individual was born outside of Sweden, and as second-generation if born in Sweden to at least one foreign-born parent.

Age was used as a continuous variable.

Educational attainment is a categorical variable based upon the years of formal education: ≤9 years, 10–12 years, and 12+ years.

Family income was based on the annual family income divided by the number of family members. Also accounting for the ages of people in the family, the sum of all family members’ incomes was multiplied by the individual’s age-based consumption weight (with lower weights for small children given than for adolescents and adults) and then divided by the family members’ total age-based consumption weight.

A summary SES measure, used in moderation models (see Statistical Analysis section), was an ordinal variable ranging from 1 (the lowest SES) to 6 (the highest) based upon the sum of the values of educational attainment (coded as 0–2), family income (0–3), and neighborhood deprivation (0–2, described below).

As a proxy for country-of-origin drinking cultures, per capita alcohol consumption indicates liters of ethanol consumed per adult aged 15+ years in the immigrant’s country of origin in 2010 (World Health Organization, 2018). Capturing the varying degree of ‘wetness’ of a drinking culture (Room and Mäkelä, 2000), or prevalence of drinking in a society, this measure is based on past research on the influence of immigrant drinking cultures on drinking patterns (Cook et al., 2013; Cook and Caetano, 2014; Cook et al., 2015; Cook et al., 2012).

As recorded in the Multi-Generation Registry, reason for immigration had the categories of labor market (or employment) opportunity, refugee, student, other, or unknown.

Neighborhood-level variables

Based upon geocoded home addresses, small geographic units with an average of 1,000–2,000 residents were used as proxies for neighborhoods (Karriker-Jaffe et al., 2017; Sundquist et al., 2004).

Neighborhood alcohol outlet density was a dichotomous variable indicating whether a neighborhood had one or more on- or off-premise alcohol outlets (versus none).

Neighborhood deprivation was a summary measure based on four indicators for residents aged 25–64 (Winkleby et al., 2007): proportion of individuals with low education (<10 years); proportion of low income (< 50% of median); proportion of the unemployed (excluding full-time students, those in military service, and early retirees); and proportion of individuals receiving social welfare benefits. The variables loaded on one principal component, and a z-score, weighted by the coefficients for the eigenvectors, was calculated for each indicator. These were summed to create the neighborhood deprivation index (Gilthorpe, 1995). We used a categorical variable of low- (< 1 standard deviation (SD) from the mean), medium- (within 1 SD), and high-deprivation (>1 SD).

To adjust for regional differences in hospital admissions, region of residence was included: large cities (three cities with a population of >200,000; Stockholm, Gothenburg and Malmö), Southern Sweden, and northern Sweden.

2.3. Statistical Analysis

Cox regression analysis was conducted to estimate adjusted hazard ratios with 95% confidence intervals for alcohol and drug use disorders in second-generation immigrants compared to first-generation. Moderation tests used an interaction term between being second-generation with country-of-origin alcohol consumption and another with the SES summary score. Interactions were tested separately, then simultaneously. To illustrate how alcohol and drug use disorder rates differed between first- and second-generation immigrants of different cultural and socioeconomic profiles, raw rates were graphed for each group across country-of-origin alcohol consumption levels and SES (separately). Supplemental analyses then generated adjusted hazard ratios for second- versus first-generation immigrants from models stratified by the levels of each of the two moderators.

3. Results

3.1. Characteristics of Immigrants to Sweden

The majority (57.0%) of first-generation immigrants and an even higher proportion of second-generation immigrants (81.3%) originated from other European countries (Table 1). While a higher proportion of second- than first-generation had received college education, a higher proportion of second- than first-generation lived in the most-deprived neighborhoods. Alcohol and drug use disorder rates were higher among second- than among first-generation.

Table 1.

Characteristics of immigrants in Sweden, ages 15+ years in 2010

	First generation (born outside Sweden), N=815,778		Second generation (born in Sweden), N=674,757
	No.	%	No.	%
Male sex	383,142	47.0	345,560	51.2
Age, mean (SD)	48.3 ± 16.1		36.2 ± 15.0
Region of origin
Nordic countries	191,343	23.5	342,785	50.8
Middle East/North Africa	170,260	20.9	43,008	6.4
Eastern Europe	121,471	14.9	43,001	6.4
Asia (excluding Middle East)	89,057	10.9	35,464	5.3
Central Europe	51,939	6.4	36,805	5.5
Western Europe	51,605	6.3	74,568	11.1
Africa (excluding North Africa)	40,373	4.9	10,464	1.6
South America	37,471	4.6	17,104	2.5
Southern Europe	24,179	3.0	30,808	4.6
Russia	12,725	1.6	5,339	0.8
Baltic countries	10,816	1.3	15,741	2.3
North America	10,186	1.2	17,830	2.6
Others	4,353	0.5	1,840	0.3
Proportion with one Swedish parent	10,407	1.2	448,743	66.5
Reason for immigration			N/A
Unknown	504,453	61.8
Relatives to EU	149,771	18.4
Humanitarian reason	88,743	10.9
Protective status	47,185	5.8
Other refugees	12,374	1.5
Labor market	7,781	1.0
Students	2,882	0.4
Others	2,589	0.3
Educational attainment, % at highest level	409,943	50.3	398,938	59.1
Family income (quartiles), % at highest level	204,008	25.0	168,804	25.0
Region of residence, % in large cities	368,603	45.2	410,387	60.8
Neighborhood deprivation, % highest level	86,196	10.6	114,594	17.0
Neighborhood alcohol outlet density, % with any outlets	107,757	13.2	91,641	13.6
Alcohol abuse, %	16,495	2.0	24,362	3.6
Drug abuse, %	14,024	1.7	31,081	4.6

Note. List of countries included in each region here.

Nordic countries: Denmark, Finland, Norway and Iceland.

Middle East/North Africa: Afghanistan, Algeria, Egypt, Indonesia, Iraq, Iran, Jordan, Lebanon, Morocco, Pakistan, Palestine, Saudi Arabia, Syria, Tunis, and Uzbekistan.

Eastern Europe: Bosnia, Yugoslavia, Croatia, Poland, Rumania, Bulgaria, and other Eastern countries.

Asia (excluding Middle East): Bangladesh, Philippines, India, Japan, China, South Korea, Turkey, Sri Lanka, Vietnam, and other Asia countries.

Central Europe: Poland, Czechoslovakia, Hungary, and Moldovia

Western Europe: Netherland, England and Ireland, Germany, Austria, and other European countries.

Africa (excluding North Africa): Ethiopia, Somalia, Eritrea, Ghana, Kenya, Congo, Nigeria, Gambia, Uganda, South Africa, and others.

South America: Argentina, Chile, Colombia, Brazil, Uruguay, and Peru

Southern Europe: France, Greece, Italy, Spain, and Portugal.

Baltic countries: Estonia, Latvia, Lithuania, and Belarus

North America: U.S.A., Canada

Table 2 shows cumulative alcohol and drug use disorder rates by region of origin. For first-generation, alcohol use disorder rates were highest among individuals from other Nordic countries, South America, and Central Europe, and lowest among those from the Middle East/North Africa. Drug use disorder rates were highest among immigrants from Africa (excluding North Africa) and South America. Among second-generation, individuals from other Nordic countries had the highest alcohol use disorder rate, followed by those from Baltic countries and Russia. Second-generation immigrants from Africa, the Middle East/North Africa, and South America had the highest drug use disorder rates. Second-generation immigrants had higher alcohol and drug use disorder rates than first-generation, although alcohol use disorder rate for first-generation immigrants from Central Europe was higher than that for second-generation immigrants. Greater absolute differences in alcohol use disorder between second- and first-generation immigrants were shown for immigrants from Russia (1.56%) and the Baltics (1.38%). In drug use disorder, greater differences were shown for immigrants from the Middle East/North Africa (7.95%), other parts of Africa (7.10%), and South America (6.90%). Second-generation immigrants had higher alcohol and drug use disorder rates than first-generation, although alcohol use disorder rate for first-generation immigrants from Central Europe was higher than that for second-generation. Excepting Russia and Baltic countries, the risk differences between second- and first-generation were larger for drug use disorder than for alcohol use disorder.

Table 2.

Cumulative rate (%) of alcohol use disorder (AUD) and drug use disorder (DUD) in immigrants in Sweden, ages 15+, by region of origin

	First generation (born outside Sweden), N=815,778		Second generation (born in Sweden), N=674,757		Absolute difference between second and first generation
	AUD	DA	AUD	DA	Risk difference in AUD	95% CI		Risk difference in DUD	95% CI
Sweden	2.22	1.18	2.89	1.88	0.67	0.65	0.69	0.70	0.68	0.71
Region of origin
Nordic countries	3.54	1.29	4.64	3.79	1.10	0.99	1.21	2.50	2.41	2.58
Middle East/North Africa	0.97	2.07	1.94	10.02	0.97	0.83	1.11	7.95	7.66	8.24
Asia (excluding Middle East)	1.10	1.20	1.72	5.96	0.62	0.47	0.77	4.76	4.50	5.02
Central Europe	3.24	1.76	2.92	4.58	−0.32	−0.55	−0.09	2.82	2.58	3.06
Western Europe	1.71	0.84	2.88	2.55	1.18	1.01	1.34	1.71	1.57	1.85
Africa (excluding North Africa)	1.60	4.19	2.46	11.29	0.86	0.54	1.18	7.10	6.46	7.73
South America	2.50	3.13	2.64	10.03	0.14	−0.14	0.43	6.90	6.41	7.38
Southern Europe	1.44	1.32	2.52	4.98	1.08	0.85	1.31	3.66	3.38	3.94
Russia	2.05	1.38	3.61	2.72	1.56	1.01	2.12	1.34	0.86	1.82
Baltic countries	2.39	0.64	3.77	1.34	1.38	0.96	1.79	0.70	0.47	0.94
Northern America	2.21	1.53	3.06	2.72	0.85	0.47	1.23	1.19	0.85	1.53
Others	2.30	1.84	2.12	6.47	−0.18	−0.97	0.62	4.63	3.44	5.82

Note. List of countries included in each region here.

Nordic countries: Denmark, Finland, Norway and Iceland.

Middle East/North Africa: Afghanistan, Algeria, Egypt, Indonesia, Iraq, Iran, Jordan, Lebanon, Morocco, Pakistan, Palestine, Saudi Arabia, Syria, Tunis, and Uzbekistan.

Eastern Europe: Bosnia, Yugoslavia, Croatia, Poland, Rumania, Bulgaria, and other Eastern countries.

Asia (excluding Middle East): Bangladesh, Philippines, India, Japan, China, South Korea, Turkey, Sri Lanka, Vietnam, and other Asia countries.

Central Europe: Poland, Czechoslovakia, Hungary, and Moldovia

Western Europe: Netherland, England and Ireland, Germany, Austria, and other European countries.

Africa (excluding North Africa): Ethiopia, Somalia, Eritrea, Ghana, Kenya, Congo, Nigeria, Gambia, Uganda, South Africa, and others.

South America: Argentina, Chile, Colombia, Brazil, Uruguay, and Peru

Southern Europe: France, Greece, Italy, Spain, and Portugal.

Baltic countries: Estonia, Latvia, Lithuania, and Belarus

North America: U.S.A., Canada

3.2. Adjusted Regression Models

The adjusted Cox regression models showed second-generation immigrants had higher risks for alcohol and drug use disorder incidences than first-generation (Table 3). Country-of-origin alcohol consumption level was positively associated with both alcohol and drug use disorders. For the most part, SES was inversely associated with alcohol and drug use disorder risks, but the socioeconomic gradient associated with drug use disorder was clearer than that for alcohol use disorder. Increased risk for drug use disorder was consistently associated with lower SES across all three indicators (education, family income, and neighborhood deprivation). Similarly, alcohol use disorder was associated with greater neighborhood deprivation. While immigrants with college education had lower alcohol use disorder risk than those with lowest education (<10 years), those with 10–12 years of education had somewhat higher risk for alcohol use disorder than those with lower education. Similarly, only those with higher family incomes, and not middle-low incomes, had lower alcohol use disorder risk than those with the lowest incomes.

Table 3.

Hazard ratios for alcohol use disorder (AUD) and drug use disorder (DUD) in immigrants to Sweden

	AUD			DUD
Variables in adjusted models	HR	95% CI		HR	95% CI
Second generation immigrants (ref. First-generation)	1.60	1.56	1.64	1.42	1.38	1.46
Age	1.01	1.01	1.01	0.96	0.96	0.97
Male (ref. Female)	2.81	2.75	2.88	3.65	3.57	3.73
Educational attainment (ref. <10 years)
10– 12 years	1.05	1.03	1.08	0.83	0.81	0.86
> 12 years	0.53	0.52	0.55	0.38	0.37	0.39
Family income (ref. Low)
Middle-low	1.02	0.99	1.05	0.83	0.81	0.85
Middle-high	0.80	0.77	0.82	0.55	0.53	0.56
High	0.64	0.62	0.66	0.27	0.26	0.28
Neighborhood deprivation (ref. High)
Low	0.62	0.59	0.64	0.53	0.51	0.55
Middle	0.80	0.77	0.82	0.70	0.68	0.72
Unknown	0.86	0.83	0.88	0.73	0.71	0.75
Neighborhood alcohol outlets (ref. None)	0.94	0.90	0.98	0.99	0.94	1.04
Region of residence (ref. Large cities)	0.84	0.83	0.86	0.78	0.76	0.80
Per capita consumption in country of origin	1.10	1.09	1.10	1.03	1.03	1.03
Interaction Model 1: *
Second generation immigrants (ref. First-generation)	1.78	1.66	1.91	1.77	1.69	1.86
Per capita consumption in country of origin	1.11	1.10	1.12	1.07	1.06	1.08
Second generation X Per capita consumption	0.99	0.98	1.00	0.98	0.97	0.98
Interaction Model 2: *
Second generation immigrants (ref. First-generation)	1.04	0.95	1.13	0.89	0.77	1.03
Socioeconomic status summary score (ref. Score 6: highest)**
Score 1 (Lowest)	2.42	2.20	2.66	9.98	8.72	11.42
Score 2	2.01	1.84	2.19	6.06	5.29	6.93
Score 3	1.61	1.48	1.75	3.90	3.41	4.46
Score 4	1.50	1.38	1.63	2.54	2.21	2.92
Score 5	1.28	1.18	1.40	1.73	1.49	2.01
Second generation X Socioeconomic status
Second generation X SES summary score 1	1.74	1.55	1.95	1.15	0.98	1.34
Second generation X SES summary score 2	1.69	1.52	1.88	1.20	1.02	1.41
Second generation X SES summary score 3	1.53	1.38	1.70	1.18	1.01	1.38
Second generation X SES summary score 4	1.22	1.10	1.35	1.14	0.97	1.35
Second generation X SES summary score 5	1.07	0.96	1.20	0.99	0.83	1.18
Interaction model 3: *
Second generation immigrants (ref. First-generation)	0.85	0.75	0.97	0.83	0.71	0.97
Socioeconomic status summary score (ref. Score 6: highest)**
Score 1 (Lowest)	2.39	2.17	2.63	9.87	8.62	11.30
Score 2	1.99	1.83	2.18	6.02	5.26	6.89
Score 3	1.60	1.47	1.74	3.88	3.39	4.45
Score 4	1.50	1.38	1.63	2.53	2.20	2.91
Score 5	1.28	1.17	1.40	1.73	1.49	2.00
Second generation X Socioeconomic status
Second generation X SES summary score 1	1.78	1.58	2.00	1.17	0.99	1.37
Second generation X SES summary score 2	1.71	1.54	1.91	1.21	1.03	1.42
Second generation X SES summary score 3	1.54	1.39	1.71	1.19	1.01	1.39
Second generation X SES summary score 4	1.22	1.11	1.36	1.14	0.97	1.35
Second generation X SES summary score 5	1.08	0.96	1.20	0.99	0.83	1.18
Per capita consumption in country of origin	1.08	1.08	1.09	1.01	1.00	1.02
Second generation X per capita consumption	1.02	1.01	1.03	1.01	1.00	1.02

^*Interaction models 2 and 3 adjust for age, gender, neighborhood alcohol outlets, and region of residence, but exclude SES variables included in the summary score (income, education, and neighborhood deprivation).

^**Excludes cases missing on neighborhood deprivation.

HR: Hazard ratio; CI: Confidence interval

Similar patterns were found in gender-specific analyses (Supplemental tables A and B), but the effect sizes for country-of-origin alcohol consumption were larger for females than males. Second-generation immigrants had higher risks for alcohol and drug use disorders than first-generation in gender-specific analyses as well, with effect sizes somewhat larger for females than males.

In the moderation model including all of the interaction terms (Table 3, Model 3), the interaction between generational status and country-of-origin drinking level was significant for alcohol use disorder (with risk increasing with higher country-of-origin drinking level) but not for drug use disorder. The relationship between country-of-origin drinking level and drug use disorder varied by generational status; first-generation consistently had lower raw rates of alcohol and drug use disorders than second-generation, with alcohol use disorder rates being higher for immigrants from heavy-drinking countries but drug use disorder rates varying little by country-of-origin drinking level. However, for second-generation immigrants, raw rates of drug use disorder were highest for those originating from countries with the lowest level of country-of-origin alcohol consumption (Figure 1). In models stratified by country-of-origin drinking level, the differences in adjusted hazard ratios between first- and second-generation were larger at the lowest levels of country-of-origin alcohol consumption for both alcohol and drug use disorders, suggesting generational differences were less pronounced for immigrants from high-consumption countries (Supplemental Figure 1; Supplemental Tables C & D).

Figure 1:

Unadjusted rate (%) of alcohol use disorder (top panel) and drug use disorder (bottom panel) for first- and second-generation immigrants across different levels of country-of-origin per capita alcohol consumption.

Note: Rates are raw percentages.

The interaction between generational status and the summary SES score was mostly significant for alcohol use disorder, but mixed for drug use disorder. As Figure 2 shows, both first- and second-generation immigrants with higher-SES had lower raw rates of alcohol use disorder (top panel) and drug use disorder (bottom panel), with the rates highest for second-generation immigrants with low SES. In models stratified by SES, the differences in adjusted hazard ratios between first- and second-generation were larger at the lowest levels of SES for both alcohol and drug use disorders, suggesting generational differences were less pronounced for immigrants with higher SES (Supplemental Figure 2; Supplemental Tables E & F).

Figure 2:

Unadjusted rate (%) of alcohol use disorder (top panel) and drug use disorder (bottom panel) for first- and second-generation immigrants across different levels of socioeconomic status.

Note: Rates are raw percentages.

In gender-stratified models including all the interaction terms (Supplemental Tables A and B, Model 3), for men, low-SES groups mostly had higher alcohol use disorder risk compared to the highest-SES group (though with no clear social gradient), but the model for drug use disorder showed no significant interactions. For women, there were significant interactions between generational status and SES for both alcohol use disorder (all but the highest level of SES) and drug use disorder (all but the two highest SES levels).

4. Discussion

As hypothesized, immigrants originating from countries with high levels of alcohol consumption had higher risks for alcohol and drug use disorders. Also as expected, immigrants with high individual- and neighborhood-level SES had lower alcohol and drug use disorder risks, with socioeconomic gradients observed more clearly for drug use disorder than for alcohol use disorder. As in prior studies, second-generation immigrants had higher alcohol and drug use disorder risks than first-generation, with the effect sizes greater for females than males.

Consistent with past research (Chen and Chien, 2018; Cook and Caetano, 2014; Cook et al., 2015; Shmulewitz et al., 2012), our findings suggest higher risks for alcohol and drug use disorders in immigrants originating from countries with high levels of alcohol consumption. Notably, this relationship held for both first- and second-generation immigrants, suggesting the enduring influence of country-of-origin cultures on children of immigrants born in the host society, as suggested by transnationalism theories (Portes et al., 1999; Schiller et al., 1995). Furthermore, among individuals from heavy-drinking countries, second-generation immigrants had higher alcohol use disorder risk compared to first-generation. While country-of-origin drinking cultures may be a determinant of alcohol use disorder in first-generation immigrants, second-generation immigrants are likely to be influenced by drinking patterns in the majority population as well (Hjern and Allebeck, 2004). In a heavy-drinking country like Sweden, the confluence of these two forces may pose a particularly high risk for alcohol use disorder among second-generation immigrants.

Prior studies of alcohol use disorder in immigrants have mostly focused on discrete categories representing specific countries or regions of origin. For example, Puerto Ricans and Mexican Americans in the US had higher alcohol use disorder risk than Cuban and South/Central Americans (Caetano et al., 2009; Caetano et al., 2014; Ramisetty-Mikler et al., 2010); Finnish immigrants had higher alcohol use disorder risk, and immigrants from southern Europe and non-European countries had lower risk, than native Swedes (Hjern and Allebeck, 2004). However, the specific country characteristics that may explain such differences have not been elucidated previously. To address this gap, we operationalized one such characteristic—drinking prevalence, with country-of-origin alcohol consumption level as a proxy—using a continuous measure that allowed us to conduct robust analyses to test its influence on alcohol and drug use disorder risk for immigrants.

Our findings on the increased risks for alcohol and drug use disorders for low-SES immigrants are consistent with studies from the United States that link low SES and other indicators of disadvantage such as financial hardship and discrimination to problem drinking (Mossakowski, 2008; Mulia et al., 2008; Zemore et al., 2011) and an Italian study showing harmful drinking was associated with unemployment among African immigrants (Di Napoli et al., 2020). Prior studies also reported associations between neighborhood deprivation and drug use, likely mediated through increased stress and distress (Boardman et al., 2001) and greater exposure to drugs (Crum et al., 1996; Storr et al., 2004). Our findings are also consistent with our prior studies that show higher alcohol use disorder risks associated with low SES in the general Swedish population (Karriker-Jaffe et al., 2018a; Karriker-Jaffe et al., 2018b; Kendler et al., 2014), suggesting such risks are not specific to immigrants.

The lower risks for alcohol and drug use disorders we found for first-generation immigrants are largely consistent with the immigrant paradox, whereby foreign nativity is protective against psychiatric disorders and premature mortality despite the stressful experiences and poverty often associated with immigration (Alegria et al., 2008; Burnam et al., 1987; Rubalcava et al., 2008). A mechanism that may account for this paradox is healthier lifestyles (including light/moderate drinking) in first-generation immigrants to the West, which tend to become less healthy with increasing acculturation of immigrants and in their native-born offspring (Abraido-Lanza et al., 2005; Abraido-Lanza et al., 1999). Our findings lend support to this mechanism. At the same time, our findings show differential risks associated with SES even among immigrants, providing a more nuanced picture of immigrant substance use, beyond the typically low risks observed for immigrants as highlighted by the immigrant paradox (Alegría et al., 2006; Rubalcava et al., 2008; Salas-Wright et al., 2014). In this respect, our findings are well-aligned with the alcohol harm paradox thesis that refers to the greater burden of alcohol-related morbidity and mortality on lower-SES individuals (Bellis et al., 2016; Lewer et al., 2016), demonstrating the applicability of this thesis to immigrants for the first time. Future work might explore whether low SES is differentially associated with substance use outcomes for immigrants compared to their native-born counterparts.

Though not the central focus of this study, gender differences are worth noting. We found a greater effect of country-of-origin alcohol consumption level on alcohol and drug use disorder risks for females than for males. Gender norms powerfully influence female drinking (Kuntsche et al., 2011), and country-of-origin drinking cultures may shape such norms in immigrant communities. The erosion of the influence of country-of-origin cultural norms on immigrant women with increasing acculturation may increase alcohol or drug use disorder risk (Zemore, 2007), especially if combined with greater availability of drugs (Borges et al., 2012). This may explain our findings of higher alcohol and drug use disorder risks for female second-generation immigrants than for their first-generation counterparts. Notably, we found far higher alcohol and drug use disorder risks for female second-generation immigrants with low SES, compared with female first-generation immigrants with high SES. Substance-using women often experience other comorbid medical conditions and psychosocial stressors (Meyer et al., 2019), and our findings call attention to one such vulnerable group, second-generation immigrants with low SES.

4.1. Strengths and Limitations

We acknowledge several limitations of this study. First, we used country-of-origin alcohol consumption level in 2010 as a proxy for drinking culture (or prevalence, to be more specific), but as immigrants moved to Sweden over the years, this variable may or may not have adequately captured country-of-origin drinking level prior to an individual’s move to Sweden. Second, according to our analysis of the World Mental Health Survey, which collected alcohol and drug use data in 25 countries (Degenhardt et al., 2019; Glantz et al., 2020)—the only data source, to our knowledge, that has country-level alcohol and drug use data—our assumption that drug use is more likely to be tolerated in heavy-drinking countries was upheld for low- and middle-income countries (as demonstrated by high correlation between alcohol use disorder and drug use disorder) but not for high-income countries. This may explain the low risk for drug use disorder associated with country-of-origin drinking level. Third, given the use of hospital and crime registries, we capture alcohol and drug use disorder incidence only among a subset (i.e. the individuals who used health care or came into contact with law enforcement) of all cases. Potential selection bias of this nature may partly explain low alcohol use disorder rates for immigrants from some countries known for their high rates of alcohol use disorder (e.g., Russia). Still, given universal healthcare coverage in Sweden, systematic barriers to healthcare access may not be present (hence precluding significant underrepresentation of low-SES immigrants in our data); past research is largely mixed on the relationship between SES and healthcare use in Sweden (Di Thiene et al., 2018; Jöud et al., 2014; Kristenson et al., 2011; Mosquera et al., 2017). Additionally, a failure to detect alcohol or drug use disorder in some individuals is likely to entail non-differential misclassification errors that result in conservative estimates, with significant results remaining valid (Dosemeci et al., 1990; Rothman et al., 2008). Fourth, to the extent first-generation immigrants receive alcohol use disorder treatment less than the native-born, as found in our recent study of alcohol use disorder pharmacotherapy (Karriker-Jaffe et al., 2017), lower alcohol and drug use disorder rates among first-generation immigrants may be partly attributed to lower access to treatment. Fifth, we were unable to further explore socioeconomic diversity among immigrants—e.g,, professional/economic migrants versus refugees who differ in circumstances of migration and resources they command (Brandt et al., 2019; Duggal et al., 2019; Lindert et al., 2009)—in relation to alcohol and drug use disorder risk. Information about the reasons for immigration that may shed light on these circumstances was lacking for over 60% of immigrants in our data. As most immigrants in our data (particularly second-generation) originated from other parts of Europe, our findings may reflect their experiences more so than immigrants from other countries.

The current study has important strengths that balance these limitations. The population data offer a diverse pool of immigrants from 64 countries covering all the major regions of the world, which allowed us to explore cultural and socioeconomic diversity among immigrants in relation to alcohol use disorder and drug use disorder. These registry data provide credible information on more severe cases of alcohol and drug use disorders, which is a clear advantage over self-report data. Finally, the longitudinal design allowed us to track alcohol and drug use disorder incidences, helping to generate credible findings that significantly contribute to the literature.

4.2. Conclusions

Among immigrants in Sweden, second-generation immigrants are at increased risk of developing alcohol and drug use disorder, particularly if they come from heavy-drinking countries or have lower SES. Policy and community attention to these high-risk subgroups in immigrant communities is warranted.