Trends in alcohol intoxication among native and immigrant youth in Sweden, 1999-2017: A comparison across family structure and parental employment status

Background: Developing a better understanding of drinking patterns across immigrant generations and how these change over time is important for the development of eﬀective alcohol polices. This study investigates the direction and rate of change in youth alcohol intoxication over time, based on immigrant status, and by family structure and parental employment status. Method: The study is based on eight nationally representative school surveys conducted by the Swedish National Council for Crime Prevention between 1999 and 2017, with a combined sample of 50,657 adolescents. Group by time interactions were examined to compare rates of change of alcohol intoxication over time across immigrant generations. Results: The results show a decreasing trend in alcohol intoxication among both ﬁrst and second generation immigrant youth, and also among immigrant youth across diﬀerent family structures and parental employment statuses. The results also show that the decrease in alcohol intoxication over time is greater for youths born abroad and for youths with two immigrant parents than for native Swedes, and that the decrease over time is greater for youths from intact families than for native Swedish youths from non-intact families and youths with one immigrant parent. Conclusion: Native and ﬁrst-and second-generation immigrant youth may diﬀer substantially from one another in many ways, and may therefore manifest diﬀerent patterns of drinking behaviours. From a policy and prevention perspective, the data in this study imply that native youths and youths with one immigrant parent should be a central target group for alcohol prevention policy in Sweden

Sweden experienced a steep increase in per capita alcohol consumption during the late 1990s and early 2000s making alcohol consumption, and the potential harms arising from it, a major concern among Swedish policy makers ( Norström & Ramstedt, 2006 ).Alcohol use among Swedish youth resembled per capita consumption and adult consumption during the 1980s and 1990s, but developed an independent trajectory just before 2000 ( Norström & Raninen, 2015 ).Since the turn of the millennium, most indicators of alcohol consumption among youth in Sweden have been declining and reached historically low levels in 2017.According to school surveys conducted by CAN (The Swedish Council for Information on Alcohol and Other Drugs), both the frequency of drinking and heavy episodic drinking declined sharply among 15-year-old boys and girls between 2000 and 2017, while abstention https://doi.org/10.1016/j.drugpo.2021.1033970955-3959/© 2021 The Author(s).Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ) rates increased from 32 to 57 percent among girls and from 33 to 63 percent among boys ( Thor, 2017 ).
Although the vast majority of Nordic studies suggest that immigrant youth are underrepresented in alcohol consumption by comparison with the native-born population (see for example Svensson & Hagquist, 2010 ), few studies have yet examined whether this gap is increasing or decreasing and whether the direction and rate of change differs across first and second generation immigrant youth.Similar trends in alcohol consumption across immigrant and native youth in Sweden and Norway have been reported in one Swedish and one Norwegian study ( Kim, Evans, & Hagquist, 2019 ;Pedersen & von Soest, 2015 ), whereas another Norwegian study found that although heavy episodic drinking had changed over time among adolescents from immigrant backgrounds, the decrease in drinking was more pronounced among native majority adolescents in Oslo ( Rogne, Pedersen, & Bakken, 2019 ).However, these studies did not differentiate between the first and second generation, or only reported on differences between first generation migrants (born abroad) and the native population.As a result these studies cannot demonstrate whether the direction and rate of change in alcohol trends differs across immigrant generations.
Length of residence in the receiving country plays a significant role in influencing immigrants' patterns of alcohol use, and there is an overall anticipation that the alcohol use of immigrants, and particularly their children, will eventually converge toward that of the native population ( Johnson & Svensson, 2020 ;Amundsen, Rossow, & Skurtveit, 2005 ;Sordo et al, 2015 ).Changes in alcohol consumption across immigrant generations are often understood in terms of differences in acculturation, with first generation migrants from non-Western countries being expected to adopt Western behavioural practices to a lesser degree than second generation migrants as a result of their stronger contact and identification with the culture in their country of origin ( Johnson & Svensson, 2020 ).Furthermore, while numerous studies have investigated the relationship between trends in alcohol use among youth and structural variables such as family structure and parental employment status (see for example, Pedersen & von Soest, 2015 ), little research has examined whether such variables may impact trends in youth drinking differently across immigrant generations.For example, the link between increased drinking and living in a single family household may be particularly salient among second generation immigrant families due to greater discrepancies in levels of acculturation between parents and children, resulting in even greater difficulties for the parental monitoring of children's activities ( Pantin, Schwartz, Sullivan, Coatsworth, & Szapocznik, 2003 ).Although the association between parental employment status and adolescent drinking is unclear, previous studies have indicated that the stress produced in families by employment instability may be more intense among ethnic minorities due to their already marginalized position, resulting in higher vulnerability to risk behaviours among children ( Mendoza, Dmitrieva, Perreira, Hurwich-Reiss, & Watamura, 2017 ;Mistry, Lowe, Benner, & Chien, 2008 ;Bui, 2015 ).
Developing a better understanding of drinking patterns across immigrant generations and how these change over time is important for the development of effective alcohol polices and prevention and treatment programs.If immigration to a new society is associated with the adoption of harmful alcohol behaviours, this may have important implications for the possible success of public health policy.Although research indicates that problem behaviours among immigrant youth are declining ( Vasiljevic, Svensson, & Shannon, 2020 ;Bäckman, Estrada, Nilsson, & Sivertsson, 2021 ) and that only a small number of first-and second-generation immigrants exhibit problem behaviours, those who do suffer are usually at a disadvantage as a result of cumulative problems ( Berry, Phinney, Sam, & Vedder, 2006 ;Oppedal, Røysamb, & Heyerdahl, 2005 ).
On the basis of nationally representative school survey data from 1999-2017, the aim of this study is to investigate whether and how selfreported episodes of alcohol intoxication have changed over time among youths of different immigrant status (i.e.first-or second generation), and with differing family structure (i.e.non-intact/intact family) and parental employment status (i.e.employed/not in employment).

Participants
The study is based on eight waves of a nationally representative school survey of year-nine youth, (aged 15), conducted by the Swedish National Council for Crime Prevention between 1999 and 2017. 2 The survey was conducted every second year between 1999 and 2005, and thereafter every third or fourth year.The surveys are based on random samples of schools with year-nine classes.The data have primarily been collected in December, and the questionnaires were completed during lesson time.
For the surveys conducted between 1999 and 2017, the response frequency varied between 87 and 69 percent.At the school level, response rates ranged from 98 to 56 percent over the sample period.To compensate for non-response and to make the data as representative as possible, calibration weights were used for the surveys conducted in 2015 and 2017.Statistics Sweden provided the calibration weights based on available auxiliary variables from national registries, and their association with selected survey variables and the probability of participating in the survey.
In each survey, between 4200 and 7400 students have anonymously completed a questionnaire containing questions about their social situation, family, school, peer group and leisure activities.The eight subsamples combined produce a total sample of 50,657 adolescents at 1065 schools.
Since the youths have provided informed consent and completed the questionnaire anonymously, we see no general ethical problems linked to this study. 3We would nonetheless like to emphasise that studies such as this, which are based on differentiating between different groups of youths, are always associated with a risk of stigmatising certain groups.
We are aware of this risk, but feel that it must be balanced against the benefits that may accrue from improved knowledge in this area.The present study was approved by the Regional Ethics Committee at Lund University .

Measures
Immigrant background.The relevant questions included in the survey are: In which country were you born?In which country was your mother born?In which country was your father born?The response alternatives are: Sweden/Another country.If the respondent answers that he/she or one/both parents were born in another country, the respondent is asked which country.Different definitions of immigrant background have been employed in different research contexts, based on whether individuals themselves and one or both of their parents were born in Sweden or abroad.We will be using the following division: (1) Born in Sweden to two Swedish-born parents; (2) Born in Sweden to one Swedish-born and one foreign-born parent; (3) Born in Sweden to two foreign-born parents; (4) Born abroad (i.e.first generation immigrant youth).In the total sample, 74.5 percent of the respondents were born in Sweden to two Swedish-born parents, 9.2 percent were born in Sweden to one Swedish-born and one foreign-born parent, 6.5 percent were born in Sweden to two foreign-born parents and 6.2 percent of the respondents were born abroad. 4The proportions of youths in selected years 1999-2017 by immigrant background are presented in Table 1 .
Information on the respondents' place of birth was only collected for the surveys conducted between 1999 and 2011 and is not included in the analysis.Because many immigrant groups are small and not suitable for separate statistical analysis, reporting changes in alcohol rates over time by countries of origin is not possible.Data on the 10 most common low-consumption countries of origin among first generation immigrant youths included in the sample for the years 1999-2011, as well as population data from Statistics Sweden for the years 2000-2017, are presented in the appendix. 5amily structure is defined in the following way: Intact families are those where the parents live together, and the youth lives with them.Non-intact homes include single-parent households and households with step-parents or other guardians.In the total sample, the proportion of youths from intact families was 69.3 percent.
Parental employment status is a measure of whether the mother and the father are in work.The variable is coded as 0 if both parents are employed and 1 if either the mother or father, or both, are not in employment.The definition of not in employment includes cases where the mother, father, or both are seeking work, studying, on a disability pension or early retirement benefits, or are involved in some other labour market measure.While we acknowledge this is a rather rough measure, we have chosen to dichotomize this measure into two categories because of low response rates in some of the categories.In total, 20.8 percent of the respondents have at least one parent not in employment.
Alcohol intoxication is measured by the following item: How many times during the past 12 months have you drunk to the point of intoxication?' Response options: 0 -never ; 1 -one to two times; 2 -three to five times; 3 -six to ten times; 4 -eleven or more times.Since 2015, the response options have been subject to some minor adjustments.Because there were only a few responses in the upper part of the scale and the majority of youths reported having been intoxicated 1-2 times a year, a dichotomous version of this variable (0 -never, 1 -one or more times) is employed in the statistical analysis.A similar dichotomous measure of alcohol intoxication has previously been used by Rogne, Pedersen, and Bakken (2019) .In the total sample, the proportion of youths who had drunk alcohol to the point of intoxication one or more times during the last 12 months was 50.8 percent.

Statistical analysis
For this analysis, a Poisson regression model with robust standard errors has been employed to examine whether there are any differences in the direction and rate of change over time in alcohol intoxication.When the outcome variable is dichotomous and the outcome prevalence is high ( > 10%), a modified Poisson regression with robust standard errors is thought to be a more suitable alternative than logistic regression, if the effect estimate of interest is the risk ratio ( Knol, Le Cessie, Algra, Vandenbroucke, & Groenwold, 2012 ).Logistic regression will overestimate the risk ratio if the outcome prevalence is high, and this overestimation becomes larger with the increasing prevalence of the outcome ( Zhang & Yu, 1998 ).
A time trend variable was created by transforming the years of the survey into an ordered set of numbers with uniform intervals, t = (1, 2, 3, 4 …8), which measures the time span between observations.This means that even if there is some irregularity in the sampling years, we treat the values as regularly sampled, assuming that the responses would be valid if sampled on a regular 2.5 yearly basis.
The terms for the group variables (i.e., immigrant status and family structure) multiplied by the time variable were added as covariates to investigate whether the rates of change in prevalence over time differed across groups.A statistically significant interaction between time and a given group indicates that the rate of change in prevalence differs over time by comparison with the rate of change in the reference group.
The equation that defines the Poisson regression model employed in this study can be written, where Y is the estimated number of cases, i.e. youths reporting alcohol intoxication; e is the base of the natural logarithm; b 0 is the intercept; b 1 is the rate of change over time in the log of the estimated counts per time point (analogous to the slope of a line); X 1 is the time period (i.e., The beta coefficients derived from the time trend regression models were converted into the percentage change in risk using the formula, 100×( exp ( b ) −1 ) Time trends are reported as the percentage change in the risk of alcohol intoxication for each 2.5 year increase in time, based on the assumption that the change was constant from assessment to assessment.Even if this assumption is not true, the estimates may be viewed as an approximate measure of the average change or trend ( Mcnamee, Carder, Chen, & Agius, 2008 ).
All analyses were conducted with weights included.Prior to the analysis, weights were normalized using the formula, where wN is the normalized weight, w i is the raw weight for the ith observation, n is the sample size, and N is the sum of the raw weight variable.
In order to take into account the large number of statistical tests, effects were considered significant if p < 0.01.

Descriptive statistics
The proportions of youths self-reporting alcohol intoxication during the last 12 months over the observation period (1999-2017), by immigrant status, family structure, and parental employment status are pre-sented in Figs. 1 -3 .In general, the proportion of youths who reported that they had felt intoxicated once or more during the last 12 months appears to have decreased since 1999 in all groups.

Differences in the rate of change in alcohol intoxication between native and immigrant youth
The analysis focused on immigrant status and change in alcohol intoxication over time began by analysing the relationship between the time trend and alcohol intoxication for each immigrant category, and for boys and girls separately.The results are presented in the first section of Table 2 and show that alcohol intoxication decreased over time in all groups of youth.The average percentage decrease in alcohol intoxication for each 2.5 year increase in time ranged from 7.6 % (CI: -8.1.,-7.0) and 7.6 % (CI: -9.1, -6.1), for girls with Swedish parents and girls with one immigrant parent respectively, to 13.3 % (CI: -16.6, -11.0), for boys born abroad.

Differences in the rate of change in alcohol intoxication between native and immigrant youth, by family structure and parental employment status
Time trends by family structure and parental employment status for each immigrant category are presented in the first section of Table 3 .The results show that alcohol intoxication has decreased over time in all groups of youth.The average percentage decrease in alcohol intoxication for each 2.5 year increase in time ranges from 6.1 % (CI: -8.0, -4.0) for youths with one immigrant parent and a non-intact family, to 15.0 % (CI: -19.5, -11.4) for boys born abroad with one or both parents not in employment.
Differences in the rate of change by family structure for each immigrant group were also investigated by regressing the dependent variable -alcohol intoxication -on the time trend, the family structure variable, and an interaction term between family structure and the time trend.
These results are presented in the second section of Table 3 and show that in the sample as a whole, the decrease in alcohol intoxication over time is greater for youths from intact families than for youths from nonintact families (-1.6 %, CI:-2.4,-0.8),but also among native Swedes (-1.8 %, CI:-2.7,-0.8),and among youths with one immigrant parent (-3.5 %, CI: -5.8, -1.0).A corresponding analysis of boys and girls separately shows that in the total sample and among native Swedes, a significantly greater decrease in alcohol intoxication was found only for girls; girls from intact families showed a 2.0 % greater decrease (CI -3.1, -0.9 for the total sample, and CI -3.2, -0.8 for native Swedes) by comparison with girls from non-intact families.
Comparing time trends by parental employment status produced no significant difference in any of the immigrants groups, in the sample as a whole or in separate analyses of boys and girls respectively.

Robustness and sensitivity analysis
As a robustness test for the main analysis, several different analyses have been performed.Firstly, the data were re-analysed using an alternative dichotomous version of the alcohol intoxication variable (0 -"never " or "one or two times ", 1 -"three or more times ").The analysis using the "three or more times " version of the outcome variable indicated a decreasing trend in all groups.However, the only significant group difference was found among native girls from intact families, for whom there was a larger decrease in alcohol intoxication over time by comparison with native girls from non-intact families.
Secondly, the data were re-analysed using year as a categorical predictor (with 1999 as the reference category).The results from this analysis indicated a significant decreasing trend in alcohol intoxication for almost all years in all groups. 6No group comparison was performed using the year as a categorical predictor, however, since using the year as a categorical predictor would result in a large number of interaction terms, which increases the risk of overfitting the model.
Finally, sensitivity analyses were performed comparing the outputs of regression models with weighted and unweighted data.The unweighted analyses were consistent with the main analyses, and the statistical significance of the results did not vary in any of the cases.

Discussion
This study investigated the direction and rate of change in alcohol intoxication over time, based on immigrant status, and by family structure and parental employment status.Using data from a large, repeated crosssectional survey from 1999-2017, we have demonstrated a decreasing trend in alcohol intoxication among native Swedes, and both first and second generation immigrants, and across family structure and parental employment status.This is in line with previous research indicating that the declining trends in alcohol use among Swedish adolescents may be occurring in all subgroups of youth, even with regard to ethnic, parental and socioeconomic factors ( Kim, Evans, & Hagquist, 2019 ).
The results in this study also show that the decrease in alcohol intoxication over time is greater for youth born abroad and for youth with two immigrant parents than for native Swedes, and that the greater decrease among youth with two immigrant parents is largely driven by girls.Although previous Scandinavian studies on similar age groups have indicated that alcohol intoxication has decreased over time among adolescents from immigrant backgrounds, in contrast with the results of the current study, the decrease in drinking was found to be similar or more pronounced among native majority adolescents ( Rogne, Pedersen, & Bakken, 2019 ;Kim, Evans, & Hagquist, 2019 ;Pedersen & von Soest, 2015 ).In those studies, however, immigrants were treated as a single group, which means we have little knowledge about potential changes in alcohol use over time among different groups of immigrant youth.
One possible explanation for the greater decrease in alcohol intoxication found in the current study among youth born abroad and among youth with two immigrant parents, and particularly immigrant girls, may be due to differences in acculturation.Youth born abroad and youth with two immigrant parents may have stronger bonds with lowconsumption cultures in their countries of origin by comparison with youth with one immigrant parent, resulting in lower levels of alcohol consumption in general and a slower decline in consumption over time.
How acculturation occurs in relation to the alcohol use of immigrant adolescents has not been well researched.However, the influence of parental and peer approval seems to play a central role ( Johnsson & Svensson, 2020 ).Although it has been well established that the influence of friends and peers are important factors in adolescent alcohol consumption ( Kuntsche & Jordan, 2006 ), previous studies have suggested that alcohol use patterns among immigrant groups from some restrictive alcohol cultures may be more strongly influenced by family and by the alcohol behaviour in their parents' country of origin, and less by peers in the receiving country ( Svensson, 2010 ;Barsties et al, 2017 ).
Gender differences among immigrant boys and girls have been noted in several studies ( Amundsen, Rossow, & Skurtveit 2005 ;Svensson & Hagquist, 2010 ).However, few of these have discussed gender differences in patterns of acculturation.Social norms and values concerning alcohol use vary considerably across cultures, and in many lowconsumption cultures, girls are less likely than boys to participate in drinking activities ( Amundsen, Rossow, & Skurtveit 2005 ).High levels of non-drinkers have previously been found among second generation Muslim girls ( Degni, Suominen, Kauhanen, & Ansari, 2010 ;Amundsen, Rossow, & Skurtveit, 2005 ), which suggests that having a Muslim background may be an important obstacle to girls initiating drinking behaviour ( Amundsen, Rossow, & Skurtveit, 2005 ).
Over the past decades, Sweden, in line with many other Nordic countries, has evolved from an ethnically homogeneous society into a multicultural society, with a large number of immigrants from Muslim countries. 7Alcohol consumption is usually less accepted in Muslim communities, especially among girls ( ( Albar, 1998 ).If religious norms or attitudes towards a behaviour are strongly rooted in family tradition, it may be the case that behavioural convergence towards that of the native population occurs more slowly ( Hosper, Nierkens, Nicolaou, & Stronks, 2007 ).This tendency may be more strongly pronounced among second generation immigrant girls, as a result of their experiencing greater restrictions on their autonomy and freedom of movement than boys ( Dion & Dion, 2001 ;Stevens et al, 2015 ;Nilsson, 2017 ), and less pronounced among youth with one immigrant parent, due to at least one parent being likely to support mainstream norms on alcohol use ( van Tubergen & Poortman, 2010 ).Several studies have shown that first generation immigrant parents may exert a powerful influence on the behaviour of their native children, especially girls, in order to prevent threats to cultural continuity ( Gilani 2005 ;Killian & Johnson 2006 ;Drouhot & Nee, 2019 ).
The comparisons across family structure revealed that the decrease over time in alcohol intoxication has been greater for youths from intact families than for youths from non-intact families, although this was the case only among native Swedes and among youth with one immigrant parent.Subgroup analysis revealed that the associations seemed strongest for girls.These findings suggest that the disadvantage associated with growing up in a non-intact family appears to be of greatest consequence for native girls and for girls with one immigrant parent, resulting in a slower decline in alcohol consumption over time.These results are partially consistent with earlier studies demonstrating that instability in family structure may alter parents' ability to supervise their children ( Cookston, 1999 ) and that girls who are unsupervised during out-of-school hours display more deviant behaviours ( Svensson, 2003 ) and have a higher risk of alcohol intoxication than boys ( Jovicis Buric et al, 2021 ).
In contrast to expectations, no difference between intact and nonintact families in alcohol intoxication over time was found among youth born abroad or youth with two immigrant parents.However, previous studies have suggested that among some immigrant groups, kin networks are more strongly developed and that these networks may offer support to children in cases where the nuclear family breaks down, arguably providing protection against risky behaviours such as alcohol intoxication ( Kalmijn, 2017 ).
Comparing time trends by parental employment status produced no difference in any of the immigrant groups, suggesting that the decline in alcohol intoxication may be stable across different socioeconomic groups.Previous studies have suggested that the decline in drinking to the point of intoxication may be similar across all socioeconomic groups ( Kim, Evans, & Hagquist, 2019 ;Richter et al, 2013 ;Livingston, 2014 ).However, other studies have indicated that alcohol use has increased among marginalized adolescent groups, when marginalization is measured by several different indicators such as unemployment, income, low educational level, and community deprivation ( Hallgren et al, 2012 ;Healey et al, 2014 ).Parental employment status, as used in this study, is generally considered to be a rough measure of socioeconomic status, which may have affected the study's results.

Methodological considerations
Self-report data on alcohol use have been criticized on several grounds, including the underreporting and overreporting of alcohol use ( Lundborg, 2002 ).However, even if levels of alcohol intoxication were over-or underestimated in this study, this ought not to affect the validity of the findings with regard to the comparison over time, given that it is reasonable to assume that the pattern of over-and underreporting should be similar across the eight waves of the survey conducted since 1999.However, possible variations in underreporting and/or biases in reporting between students of different background need to be studied further.
The study's limitations also include both the absence of objective measures and quantity/frequency-based measures of alcohol use.Future designs could be improved by using validated scales, or by performing complementary assessments using objective measures.
One critical question in all studies is the representativeness of the sample studied.Until 2011, the survey employed in this study was characterized by low levels of non-response in comparison with many other studies, and the representativeness of the study sample may thus be viewed as good.For the surveys conducted in 2015 and 2017, the data were weighted to account for the non-response and the high level of agreement between the unweighted and weighted analyses provides some reassurance that the imbalances between the survey sample and the population may not be that large.
A further limitation is that the study did not include information on country of origin and religious affiliation, which would have been preferable as a means of further increasing our knowledge on the association between changes in alcohol intoxication over time and immigrant generational status.The large influx of immigrants during the later survey years from countries such as Syria, Afghanistan, Somalia and Iraq may have influenced the study's findings.This should be taken into consideration when interpreting the results.
Finally, no information was available on the length of residence in Sweden, making it difficult to determine the degree of socialization among first-generation immigrant adolescents within the receiving country.

Conclusion
In summary, this study has found that changes over time in alcohol consumption differ among children of immigrants, compared to native Swedes, and that there were large differences in alcohol consumption between children with two immigrant parents and those with one immigrant parent.This study also highlights the importance of studying how different risk factors and explanations, such as instability in family structure, may impact trends in youth drinking differently across immigrant groups and generations, which should also be examined further in future research.From a policy and prevention perspective, the data in this study imply that native swedes and youths with one immigrant parent should be a central target group for alcohol prevention policy in Sweden.

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

Fig. 1 .
Fig.1.Proportion reporting having drunk so much that they felt intoxicated during the past 12 months, by immigrant status.

Fig. 2 .Fig. 3 .
Fig. 2. Proportion reporting having drunk so much that they felt intoxicated during the past 12 months, by family structure.

Table 1
Proportion of youths in selected years 1999-2017, by immigrant background.