Article Text
PDF
Abstract
Background In adults, it has consistently been observed that with equal alcohol consumption levels, alcohol harms individuals with low socioeconomic position (SEP) more than individuals with high SEP. It is unknown if this so-called alcohol harm paradox is also present in adolescents. We aim to test the hypothesis that low SEP is associated with more alcohol-related harm as compared with higher SEP in Danish adolescents.
Methods We used survey data from the Danish National Youth Study 2014 including 70 566 students from 119 high schools. Alcohol-related harm was measured by self-report as having been in a fight, involved in an accident, had problems with parents or friends, had sex that was regretted afterwards and done drugs that was regretted afterwards, because of alcohol. Further, a combined measure was constructed defined as having experienced any harm more than once within the last year. SEP was measured as parents’ educational level, family income, parents’ employment status and experiencing financial strain in the family. Analyses were adjusted by age, sex, perceived ethnicity, class type and binge drinking.
Results Lower SEP was associated with higher odds of experiencing alcohol-related harm more than once as compared with highest SEP, and lower SEP was associated with higher odds of having been in a fight, problems with parents or friends, sex and regretting afterwards or drugs and regretting afterwards, as compared with highest SEP. These associations were statistically significant. Adjustment for the frequency of binge drinking did not change results.
Conclusion The alcohol harm paradox exists in Danish adolescents, and this should be addressed in future prevention strategies to reduce the risk of alcohol-related harm among adolescents.
- adolescents
- alcohol
- alcohol-related harm
- socioeconomic differences
Statistics from Altmetric.com
Background
Social inequalities in harm from alcohol consumption have consistently been observed with people from low socioeconomic position (SEP) experiencing relatively more harm than people from high SEP, also when the level of alcohol consumption is taken into account.1–6 For example, it was found that those in the lowest income quintile had an almost five times higher risk of alcohol-attributable death compared with those in the highest income quintile.3 This phenomenon, that alcohol-related harm is greater among those with low SEP compared with those with high SEP who have the same alcohol consumption level, is known as the alcohol harm paradox. A proposed explanation is socioeconomic differences in alcohol-drinking patterns so that lower SEP may be associated with a lower alcohol consumption than higher SEP, but if the lower overall consumption is concentrated in fewer bingeing occasions, such an alcohol-drinking pattern could be relatively more harmful. Another proposed explanation is accumulation of other risk factors for disease that increase the sensitivity towards alcohol among those with low SEP. Examples of such risk factors with a higher prevalence among people with low SEP compared with high SEP could be smoking and physical inactivity.7 It has also been suggested that the alcohol harm paradox could be explained by socioeconomic differences in treatment quality or in quality of social support.5 Finally, it has been suggested that differential misclassification could explain the paradox if low SEP is associated with greater underestimation of alcohol consumption than high SEP is.1
Among 15–24 year-olds, alcohol use is the leading global risk factor for years of life lost due to disability or premature mortality (DALYs).8 In Europe, 13% of adolescents have been intoxicated within a 30-day period and Denmark is well above the European average with 32%.9 Also, 73% of Danish adolescents have consumed alcohol within a 30-day period as opposed to the European average of 48%.9 Danish alcohol laws allow purchasing of wine and beer in shops from age 16 years and purchasing of stronger alcohol (>16.5%) and any alcohol in bars and restaurants from age 18 years. However, age-limits do not apply for parties held at high schools, as they are considered private parties and thus, it is legal to sell and serve alcohol irrespective of age. Considering the high alcohol intake, it is relevant to explore if the alcohol harm paradox is seen already in adolescence, defined as age 15–25 years. Since the incubation time for alcohol-related diseases, such as liver cirrhosis, pancreatitis and certain cancers, is long, such diseases rarely occur until midlife and are, thus, not possible to study in adolescents. However, in this age group, alcohol causes acute harm, such as violence, accidents, problems with parents or friends, unprotected sexual intercourse and illicit drug use.10–12
We aim at testing the hypothesis that SEP is associated with experiencing alcohol-related harm, and that those with low SEP experience more harm as compared with those with high SEP, even after taking the level of alcohol consumption into account. That is, we aim at testing if the alcohol harm paradox is manifested already in adolescence. To be able to describe multiple dimensions of SEP, we use four different measures of this, that is, parents’ educational level, family income, parents’ employment status and financial strain.
Methods
The Danish National Youth Study 2014
The study is based on data from the Danish National Youth Study 2014. A national survey of 75 853 high school and vocational school students conducted with the aim of investigating health, health behaviour and mental health among young people in secondary education in Denmark. As students on vocational schools mostly drink alcohol at parties outside of the school context, whereas high school students often drink alcohol at school parties, the drinking contexts are considered to be very different between vocational schools and high schools. Therefore, we decided to include only high school students in this study, because the differences in drinking context could cause spurious associations between our exposure and outcome. Data were collected from January to November 2014. All 137 high schools in Denmark were invited to participate, and 119 (87%) of these did so. Of the invited classes, 96% participated, and of the invited students, 85% participated in the study (n=70 674). In high schools, students are generally aged 15–20 years, but students aged 15–25 years were encouraged to participate. However, a few (n=171) were younger or older than this (12–64 years). In participating classes, teachers gave students a code for accessing the electronic survey, and students answered the electronic questionnaire, which consisted of 380 questions, in class. The Danish National Youth Study is described in more detail elsewhere.13 For this study, students with missing data on whether they drink alcohol (n=108) were excluded from analyses. The final study population, thus, consisted of 70 566 high school students.
Socioeconomic measures
SEP was measured in four different ways: parents’ educational level, family income, parents’ employment status and financial strains in the family. Information on the former three was obtained through linkage with national Danish registers using a unique personal identification number available for all residents in Denmark, and hence all students. This identification number also permits the identification of students’ parents. Information on parents’ highest educational level attained was obtained through linkage with the Population’s Education Register.14 Educational level was categorised into four groups: elementary school, short education (high school and vocational education), medium education (short and medium tertiary education) and long education (long tertiary education). Information on family income was based on household disposable income obtained through linking with the Income Statistics Register, which contains information on the income of everyone residing in Denmark, who are liable to pay Danish taxes.15 Incomes were categorised into five groups, based on income quintiles. Information on parents’ highest employment status was obtained through linking the cohort with the Register-based Labor Force Statistics.16 We categorised this variable into two groups (working and not working). Financial strain was based on a question in the Danish National Youth Study 2014 on the students’ experience on financial strain in the family within the last year (parents had difficulties paying the bills: yes/no).
Alcohol-related harm
Students who answered that they had consumed alcohol were asked if they had experienced any of five individual alcohol-related harm within the last year (0 times, once, twice or more). The question was phrased: “Within the previous year: How often have you experienced the following in connexion to your alcohol consumption?” They were asked if they had been in a fight, been involved in an accident, had problems with parents or friends, had sex that they regretted afterwards or done drugs that they regretted afterwards. These were treated as five distinct outcomes, and a combined measure was also constructed with the outcome being an experience of any alcohol-related harm twice or more, or experience of more than one individual alcohol-related harm at least once.
Other covariates
To capture more dimensions of different alcohol-drinking patterns among Danish high school students, we included two measures of this: (1) weekly alcohol consumption and (2) frequency of binge drinking.
Weekly alcohol consumption was based on students’ answers to a question on how many alcoholic drinks (12 g of pure alcohol is defined as a standard alcoholic drink in Denmark) they normally drank each day in a typical week. A weekly alcohol consumption score was calculated using the sum of alcoholic drinks consumed on each of the weekdays and categorised into the following groups: <7 drinks, 7–13 drinks, 14–20 drinks, 21–27 drinks and >27 drinks.
Frequency of binge drinking was derived from students’ answer to a question on how many times within the last 30 days they had consumed five or more alcoholic drinks on one occasion (0 times, 1–3 times, 4–6 times and >6 times).
Other covariates included sex (boys/girls), age (continuous), perceived ethnicity (Danish/Danish and other/other ethnicities than Danish) and class type (upper secondary school leaving examination/higher preparatory examination).
Statistical analysis
Associations between SEP and alcohol-related harm were assessed with multilevel logistic regression nesting students within schools (n=119) to account for dependency among students within the same school. Separate multilevel logistic regression models were performed for each of the four SEP exposures (parents’ educational level, family income, parents’ employment status and financial strain) and for each of the six outcomes (been in a fight, been involved in an accident, had problems with parents or friends, had sex that they regretted afterwards, done drugs that they regretted afterwards and experienced an alcohol-related harm more than once). Analyses were adjusted for identified confounders, that is, age, sex and perceived ethnicity, and for school type. To asses if associations between SEP and alcohol-related harm could be explained by differences in alcohol consumption, analyses were additionally adjusted for the frequency of binge drinking. To explore effect modification, we also tested if the association between frequency of binge drinking and experiencing alcohol-related harm more than once differed by SEP. This was done by testing if the interaction between SEP and frequency of binge drinking in the association with an alcohol-related harm was statistically significant (p<0.05). Only alcohol drinkers answered the questions on alcohol-related harm, so never-drinkers were excluded from all analyses on alcohol-related harm (n=7316). Never-drinkers were defined as participants who answered that they never drank alcohol on weekdays or weekends.
To account for missing values, we used multiple imputations by chained equations.17 For each outcome, imputation models were run separately. Models included the outcome and the SEP variable in question, and age, sex, ethnicity, alcohol binge drinking frequency and school id. Information on age, sex and school id was complete. The number of missing values for questionnaire variables (alcohol-related harm, ethnicity and alcohol consumption) was maximally 1.6%. Register-based information on SEP was missing for 6.2% of the sample. Following imputation, estimation was performed on each imputation separately and then combined using Rubin’s rules.18
All analyses were performed using Stata V.14.2.
Results
The mean age of the students was 17.9 years, and the majority perceived their ethnicity as Danish (90.0 %). Most of the students had parents with short (34.6 %) or medium (38.9 %) education, and 4.9% had parents that were not working (table 1). Additionally, 15.9% of the students indicated having experienced that their parents had difficulties paying the bills during the previous year (financial strain in the family).
Characteristics of the Danish National Youth Cohort (n=70 566), Denmark 2014
Frequency of binge drinking was associated with alcohol-related harm in a dose-dependent relation for both boys and girls (figure 1). For instance, OR of experiencing any harm was 11.71 (95% CI 10.29 to 13.32) among girls who binged more than six times within the last 30 days compared with girls who did not binge within the last 30 days. Similar results were observed for weekly alcohol consumption (online supplementary table S1).
Supplemental material
ORs of alcohol-related harm in alcohol drinkers by frequency of binge drinking, Denmark 2014.
There was no clear association between SEP and alcohol consumption (table 2). It seemed that longer parental education was associated with lower weekly alcohol consumption but with more binge drinking occasions among alcohol drinkers. The same trend was seen for parents’ employment status. Family income did not seem to be associated with weekly alcohol consumption, but higher income was associated with more binge drinking occasions. Having experienced financial strain was associated with both higher weekly alcohol consumption and more bingeing occasions. Non-drinkers constituted 10.4% of the population.
Socioeconomic position and alcohol consumption, Denmark 2014
Among alcohol drinkers, having parents with shorter education, lower family income, parents not working and experiencing financial strain was associated with higher odds of having experienced alcohol-related harm more than once as compared with higher SEP. This association could be found both before and after adjustment for the frequency of binge drinking (table 3). For example, students having parents with elementary school as the highest education had the highest odds of experiencing alcohol-related harm more than once (OR=1.26, 95% CI 1.12 to 1.42). For SEP measured by parents’ educational level and family income, a socioeconomic gradient in odds of alcohol-related harm was observed.
ORs of alcohol drinkers having experienced alcohol-related harm more than once by socioeconomic position, Denmark 2014
Associations between SEP and individual measures of alcohol-related harm in alcohol drinkers are presented in table 4. Having parents with shorter education, lower family income, parents not working and experiencing financial strain was associated with higher odds of being in a fight, having problems with parents or friends, having sex and regretting afterwards or doing drugs and regretting afterwards. For example, having the lowest family income was associated with higher odds of being in a fight (OR=1.49, 95% CI 1.35 to 1.64), having problems with parents or friends (OR=1.13, 95% CI 1.06 to 1.20), having sex and regretting afterwards (OR=1.13, 95% CI 1.06 to 1.22) and doing drugs and regretting afterwards (OR=1.23, 95% CI 1.10 to 1.36) compared with those with the highest family income. Being involved in an accident was, however, only associated with financial strain (OR=1.47, 95% CI 1.37 to 1.58). When results were stratified on sex, they showed similar results (online supplementary tables S2 and table S3).
ORs of alcohol-related harm in alcohol drinkers by socioeconomic position, Denmark 2014
At all levels of binge drinking, those with lower SEP as measured by low family income or having financial strain had higher odds of experiencing alcohol-related harm than those with higher SEP. There was no statistically significant interaction between frequency of binge drinking and SEP (p=0.2).
All results with complete case analyses were similar to those described.
Discussion
Odds of alcohol-related harm were higher among high school students who had parents with shorter education, lower family income, parents not working or experienced financial strain, also after adjustment for the frequency of binge drinking, indicating that the alcohol harm paradox also exists in alcohol-drinking adolescents. Results were similar in boys and girls. Interaction between SEP and frequency of binge drinking was either non-significant or negligible indicating that the influence of frequency of binge drinking on odds of alcohol-related harm did not differ between those with lower and higher SEP.
Other studies have found that lower SEP is associated with greater risk of developing alcohol-related and alcohol-attributable diseases among adults, regardless of alcohol consumption.1–5 Our results show that social inequalities in alcohol-related harm also exist in relation to adolescent alcohol consumption. The results also show that in addition to previously reported associations between SEP and alcohol-related morbidity and mortality, social inequalities also exist with regard to more acute and interpersonal alcohol-related harm in adolescence. A previous study has found an association between SEP and alcohol-related harm measured by the Alcohol Use Disorders Identification Test, and they also found this association in the age group of 16–24 years.4 These findings are important as adolescence is a period in which adult health behaviour is shaped and may be influenced by social determinants of health.19 20 Also, a Danish study has found that early risk behaviour, including early drunkenness debut, is associated with higher probability of later negative life events, such as crime conviction, not being in education, employment or training, and hospitalisation, especially among those with lower SEP,21 indicating that adolescent alcohol consumption has consequences beyond acute alcohol-related harm.
We chose to include four independent measures of SEP, because we wanted to evaluate possible differences between the measures, and these could not be identified if a combined measure of SEP was applied. Educational level, income and employment status are all correlated, but they also represent distinct properties of the social stratification conceptualised by SEP.22 Educational level is interpreted as an indicator of knowledge among the parents, while income represents economic resources in the family and employment status is an indicator of social status and resources.22 Financial strain is a subjective measure and differs from the other three measures in that it identifies those who have subjectively experienced lack of resources in the family. All measures were associated with odds of experiencing alcohol-related harm, but there was an indication that financial strain was generally strongly associated with both alcohol consumption and alcohol-related harm. SEP was associated with all alcohol-related harm except for being involved in an accident. This finding may not be surprising as this outcome stands out from the other outcomes. If the experienced accidents are mostly tripping and falling, cuts and traffic accidents, the relationship between alcohol consumption and being involved in an accident is more likely to be mediated through motor function difficulties or lower risk aversion, which is probably not influenced by SEP. However, the other four outcomes are primarily mediated through undesirable behaviour, which could be influenced by factors associated with SEP.
Possible explanations for the alcohol harm paradox as observed among adults could also apply to adolescents. One of these explanations is the socioeconomic difference in alcohol consumption pattern, but findings on this among adolescents have been inconsistent.23–25 We adjusted our analyses for the frequency of binge drinking, but it is possible that the amount of drinks at each bingeing occasion differs by SEP. However, analyses adjusted by weekly alcohol consumption instead of the frequency of binge drinking did not substantially change our results.
Another suggested explanation is the accumulation of risk factors in disadvantaged socioeconomic groups, and it is possible that risk factors for acute alcohol-related harm besides alcohol consumption accumulate among adolescents with lower SEP. Risk factors for the explored measures of alcohol-related harm not related to alcohol consumption could be mental distress or psychological symptoms, such as nervousness, irritability, helplessness and loneliness. Studies have found these risk factors to be more prevalent among adolescents with lower SEP26 27 possibly adding to an increased risk of alcohol-related harm among these regardless of alcohol consumption. Drinking context may also be a risk factor for alcohol-related harm that could differ between those with lower and higher SEP, though the influence from drinking context may be limited by the inclusion of only students enrolled in high schools.
It has also been suggested that the alcohol harm paradox could be explained by the association between alcohol intake and alcohol-related harm being modified by SEP.3 However, our analyses did not indicate that this was the case for Danish students. Instead, it seemed that the socioeconomic inequalities in alcohol-related harm were evident at low frequency of binge drinking and that these inequalities remained at higher frequency of binge drinking. This indicates that those with lower SEP are more prone to experience alcohol-related harm regardless of alcohol intake, and it may suggest that their risk of being involved in a fight, having problems with parents or friends, having sex that they regret afterwards or doing drugs that they regret afterwards is higher even with no alcohol intake. Unfortunately, we had no information on whether students had experienced these outcomes when they did not consume alcohol.
Strengths of the study include that information on three measures of SEP was obtained by national registers, and thus independently of measurements of information in alcohol-related harm. Another strength is the application of four separate measures of SEP, parents’ education, family income, parents’ employment status and financial strain, to explore possible differences between the measures that could not be identified if a combined measure of SEP was applied. Finally, the large survey sample should reduce the risk of random error and the design of the data collection should minimise the risk of self-selection, as all students in the class were invited to participate, and 85% of invited students did so.
Although the study was conducted with great care, the study also had some methodological limitations. First, the self-reported information is based on a cross-sectional design impeding the possibilities to make inferences on temporality. This limitation is mostly relevant to the temporality between alcohol consumption and alcohol-related harm, as it is possible that experiencing alcohol-related harm within the previous year may have influenced alcohol consumption. A further limitation is a possibility that any misclassification of alcohol consumption may be influenced by SEP allowing for differential misclassification of alcohol consumption. This has previously been suggested as an explanation to the alcohol harm paradox, as under-reporting may be more common among those with lower SEP.1
The findings of the study indicate that the alcohol harm paradox also exists in alcohol-drinking adolescents. Adolescents who had parents with shorter education, lower family income, parents not working and experienced financial strain were identified as having the highest odds of experiencing alcohol-related harm regardless of the level of alcohol consumption, among both boys and girls. The implications of the findings are that prevention strategies are needed to reduce the risk of alcohol-related harm among adolescents, and that these should focus especially on effective interventions targeting alcohol consumption among adolescents with low SEP.
In conclusion, the study found that odds of alcohol-related harm were higher among those who had parents with shorter education, lower family income, parents not working or experienced financial strain as compared with those with high SEP in Danish high school students. Results persisted after adjusting for the level of alcohol consumption, which indicates that the alcohol harm paradox exists among Danish adolescents.
What is already known on this subject
Studies have shown that among adults, people from low socioeconomic position (SEP) experience relatively more alcohol-related harm than people from high SEP, also when the level of alcohol consumption is taken into account. This phenomenon had been named the alcohol harm paradox.
What this study adds
Among Danish adolescents, odds of alcohol-related harm were higher among those who had parents with shorter education, lower family income, parents not working or experienced financial strain as compared with those with high socioeconomic position. Adjustment for the level of alcohol consumption did not change the results, which indicates that the alcohol harm paradox exists among Danish adolescents.
References
Footnotes
Contributors SPM and JT planned and designed the study. SPM was responsible for data management and analyses, and drafted the manuscript. All the authors contributed to the interpretation of study results, critical revision of the paper and approval of the final version, and agree to be accountable for all aspects of this article.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval Approval of the study was obtained from the Danish Data Protection Agency (J. No. 2013-54-0526).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data may be obtained from a third party and are not publicly available.