Abstract
Introduction
Epidemiological research on alcohol-related harm has long given priority to studies of harm to the drinker and the risks associated with different levels and patterns of drinking by the drinker.1ȓ3 This focus is also reflected in international monitoring systems of alcohol-related harm, where measures of consumption, drinking patterns, and alcohol-related mortality often constitute core indicators. 4 In addition, health problems to the drinker are the basis when comparing alcohol-related harm across countries and regions within the global burden of disease project. 5 National comparisons based on this kind of data subsequently represent an important basis for determining how well countries measure up against each other with respect to alcohol-related harm and compose basic knowledge for policymakers.
A limitation with a focus on consumption and harm to the drinker is that the harm that drinking causes to people around the drinker is not sufficiently taken into account. Alcohol's harms to others have gained a renewed interest internationally in recent years, both among policymakers6,7 and researchers.8–10 A major focus within this perspective is the problems experienced by people living close to a heavy drinker, ie, by family members and friends. Most studies have considered smaller populations such as treatment samples 11 and children, 12 whereas fewer studies have addressed this problem in the whole population. A limited number of general population surveys suggest that up to roughly one-third of the adult population is negatively affected by the heavy drinking of family and friends 9 and that many of those who are affected have a reduced well-being and health status.13,14
Comparative studies of harm from drinking of family and friends are to our knowledge missing. Subsequently, national differences in the extent to which people who are close to heavy drinkers are adversely affected are not well documented and understood. For example, it is not known whether this type of harm is more common in countries with high consumption levels and alcohol-related mortality. In addition, little is known as to what extent the prevalence and severity of harm differ across subgroups in the population.
Based on the results from previous studies, we expect that women 15 – 17 are more affected than men and that younger age groups are more affected than older age groups, 9 ,15–17 whereas findings regarding other factors, such as family situation, place of residence, education, and own drinking habits, are less well documented.
The present article sheds light on these issues by comparing self-reported data on experiences of harm from heavy drinking by family and friends in five Nordic countries using data from recently conducted general population surveys. The Nordic countries are appropriate to compare in this context as they share important population, geographical, and welfare characteristics but have significant differences in drinking levels and alcohol-related mortality.18,19 In order to broaden the empirical base, we also include comparable data from Scotland, a country that resembles some of the Nordic countries with respect to population-level drinking indicators and alcohol-related mortality (refer footnote in Table 1). The same sample has been analyzed by Moan et al. 16 , with a focus on harms from the drinking of others in the wider population, rather than specifically on the harms of excessive drinking by family and friends, as here.
Comparative data on alcohol consumption, alcohol-related mortality, and estimated strictness of alcohol policies.
Liters of sold pure alcohol per inhabitants aged 15 years and older in 2013. Source: Henriksson et al, 2015.
Liters of sold pure alcohol per inhabitants aged 16 years and older in Scotland, 2012. Copyright: Nielsen/CGA, 2013.
Source: WHO, Country profiles, 2012.
Refers to the UK.
An indicator of the strength of a country's alcohol control policies. The higher score the more restrictive policy.
Scotland has stricter alcohol policies in place than the rest of the UK, but was not examined in the index or classification scale. For example, Scottish licensing laws have am explicit public health objective (protecting and improving public health), there are restrictions on alcohol availability (trains/ferries, ban on multi-buy promotions) and Scotland has introduced a minimum unit pricing policy but it introduction has been delayed due to a challenge in the European Courts. In fact, the Scottish Alcohol Policy is more in line with the Nordic alcohol control policy approach than the UK policy, see http://www.alcohol-focus-scotland.org.uk/campaigns/scottish-policy/”.
It is not obvious how the level of harm from heavy drinkers in the private life is expected to differ between countries, especially when the analysis is based on self-reported information that may consider both actual exposure and perceptions of harm. As to actual exposure to others’ drinking, we expect that the prevalence is high in countries with a high level of drinking and high rates of alcohol-related mortality as this tends to reflect a high prevalence of heavy drinkers in the country. 20 The importance of a high level of drinking is assumed to be especially strong in countries with a binge drinking-oriented drinking pattern. On the other hand, there is some evidence that tolerance toward alcohol tends to increase with the level of drinking21,22 and thus the threshold for perceiving others’ drinking as problematic is higher in countries with a high level of drinking. In addition, it is also possible that the reported levels of harm from others’ drinking will be higher in countries where alcohol is regarded as a major problem. This could be because such countries have a raised awareness of alcohol problems and also tend to attribute problems to drinking. Starting out from this assumption, it is likely that more harm is reported in countries with a restrictive alcohol policy than in countries with a more liberal policy.
To put the study countries in context, we present selected recent data on per capita consumption, prevalence of binge drinking, liver cirrhosis mortality (indicator of harm to the drinker), and measures of the strictness of alcohol policy (Table 1). With respect to per capita consumption and cirrhosis mortality, two groups emerge: Denmark, Finland, and Scotland with high levels of consumption and cirrhosis mortality and Sweden, Iceland, and Norway with lower levels. The prevalence of binge drinking follows a similar cross-national pattern for women, whereas among men, Sweden and Iceland have similar rates as Denmark and Scotland. Finland has the highest prevalence of binge drinking and Norway the lowest. As to the strictness of alcohol policy, a somewhat reverse pattern is revealed with a more restrictive alcohol policy in Norway, Iceland, and Sweden than in Finland and, in particular, Denmark.
No alcohol policy index is available for Scotland. However, the alcohol policy in Scotland is more restrictive than in the UK and is more comparable with the Nordic countries (refer footnote in Table 1).
Given this rough comparison of key indicators, our hypothesis is that levels of harm from the drinking of family and friends will be higher in Finland, Denmark, and Scotland than in Sweden, Norway, and Iceland. However, we acknowledge that this difference may be hampered by other aspects such as a higher tolerance toward drinking in general and toward alcohol-related problems in the high consumption countries as well as a lower tendency to attribute alcohol to problems due to their generally less restrictive views on alcohol.
The overall aim of this article is to increase knowledge and understanding of national differences in harm from heavy drinking in the private life. More specifically, we assess and compare the prevalence and severity of harm from heavy drinking of family and friends in six Northern European countries and discuss the findings in relation to national differences in population drinking, drinking patterns, alcohol-related mortality, and alcohol policy. In addition, the surveys included in the present article allow for a cross-country comparison of a wide range of correlates of experiencing harm from drinking by family and friends, some of which have received scant research attention to date. These include gender, age, place of residence, partner status, level of education, having children or not, and alcohol use.
Method
General population surveys were conducted in the Nordic countries (Denmark, Finland, Iceland, Norway, and Sweden) and in Scotland between 2008 and 2013. The surveys were not initially intended to be part of a directly comparative study implying that there are some methodological differences, eg, regarding mode of data collection and included questions (Table 2).
Study and sample characteristics according to country.
Nonsignificant (Pearson's chi-square;
Significant (Pearson's chi-square;
In Denmark, the sample was randomly divided into two halves. The participants in this survey include one of these halves, which used the questions of harm due to drinking of family and friends.
While the designs varied between face-to-face interviews, telephone interviews, and postal/web questionnaires, sampling was randomized in all countries except Scotland, where a quota sample was used. Response rates varied between 53% and 73%.
In order to obtain comparative samples with respect to age, only respondents aged 18–69 years were included in the analyses resulting in sample sizes ranging between 802 and 12,678 respondents. Each country calculated weights to reflect the demographic factors, and weight calculations were thus performed differently across countries.
Measures
Harm from heavy drinking of family and friends
Although the phrasing of the key questions differed somewhat between the countries, as did the coding of the items (Appendix), similar questions were used in Denmark, Iceland, Sweden, and Scotland. Respondents were first asked if they had someone in their life during the last 12 months whom they considered to be a heavy drinker or someone who drinks a lot sometimes. Second, to measure whether respondents had been
The Norwegian and Finish questionnaires differed in some important respects and had to be recoded for comparative purposes in the present study. In Finland, an overall question about being negatively affected by family or friends on a scale from 0 to 10 was used. Those reporting 0 were coded as not being negatively affected by a known drinker, whereas those reporting 1–10 were coded as being negatively affected. In Norway, a categorical question was asked with the phrasing
Demographic and socioeconomic variables
Demographic and socioeconomic variables included in the analyses were gender, age groups (18–29, 30–39, 40–49, 50–59, and 60–69 years), educational level (elementary school, high school, and university), living area (not urban and urban), cohabitation (living with partner and not living with a partner), and having children (no children and having children). The latter variable included underage children living at home.
The question about living in an urban area differed between the countries. For instance, in Norway, the definition was based on the centrality of the municipality, whereas in Sweden, the definition was based on whether they lived in a city or in a suburban area close to a large city.
Drinking habits
Drinking frequency during the past 12 months was categorized as follows: never, a few times, monthly, weekly, and four times a week or more. The last category was classified as five times a week or more for Scotland.
Binge drinking during the last 12 months was coded as never, a few times, monthly, and weekly. Nevertheless, binge drinking was assessed differently in some countries. While in Denmark and Iceland, respondents were asked how often they consumed five or more units of alcohol at one drinking occasion, Norwegian, Swedish, and Finnish respondents were asked how often they consumed about six units or more. Scottish respondents were asked how often they consumed eight units or more. The size of one standard unit varied between some of the countries, which is important to be aware of when interpreting the results. Five to six units in the Nordic countries contain about the same amount of pure alcohol as eight units in Scotland.
Analyses
First, prevalences were calculated for each country separately for (i) being negatively affected by the drinking of family and friends during a 12-month period in the total sample and (ii) among respondents who reported having a known heavy drinker in their life. Pearson's chi-square was conducted to estimate country differences in prevalence. Second, harm from a known drinker was estimated for different demographic factors and the respondents’ drinking habits in the different countries among the whole study sample and among those with a heavy drinker in their life. Third, to measure the severity of harm (on a scale from 1 to 10) mean values were calculated and country differences were assessed using one-way analysis of variance. Fourth, mean values were calculated among the different demographic factors and the respondents’ own drinking patterns in each country. Finally, to examine which demographic factors and drinking habits that were correlated with harm from drinking of family and friends and to compare these correlates between the countries, modified binary Poisson regression analyses 23 were used to calculate relative risks (RRs). Since both the dependent and independent variables were somewhat differently measured in the respective countries, we conducted separate analyses for each country and did not test statistically whether the correlates were significantly different between the six countries. Data analysis was carried out in the statistical program SPSS 22.0.
Results
Country differences in harm from heavy drinking of family and friends
The highest prevalence of self-reported harm from the heavy drinking of family and friends was found in Finland and Iceland where around 28% of respondents reported having experienced this during the last 12 months. Next followed Norway with 24%, Denmark with 19%, followed by lower figures in Sweden (17%) and Scotland (14%; Table 3). This cross-national pattern remained when restricting the analysis to only those in the population who reported to have a heavy drinker in their life with the highest proportions of around 60% found in Finland and Iceland and the lowest in Denmark (40%) and Scotland (35%). A similar proportion of respondents with a heavy drinker in their life reported experiences of problems in Norway and Sweden (47% and 49%, respectively).
Respondents that have been negatively affected by known drinkers during the last 12 months (%), according to sociodemographics and own drinking habits.
Significant country differences (Pearson's chi-square;
Significant country differences (Pearson's chi-square;
The self-reported severity of the harm from others’ drinking assessed with a scale from 1 (low severity) to 10 (high severity) showed an almost opposite pattern than the prevalence of harm. The Scottish respondents reported the highest average severity (5.5) followed by Iceland and Sweden (4.0), Norway (3.8), and Finland with the lowest estimate of 3.5 (severity not estimated in Denmark; Table 4).
Respondents that have been negatively affected by known drinkers the last 12 months, according to sociodemographics and own drinking habits. Mean severity of harm on a 1–10 scale.
Question not asked in Denmark.
Significant country differences (one-way analysis of variance;
Differences in subgroups of the population
Women were found to be more often negatively affected by a heavy drinker in their private life than men in all countries (Table 3). The prevalence declined gradually by increasing age in Sweden, Finland, and Norway but not in the other countries. There was no clear pattern for area of residence, education, and family situation. The respondents’ own drinking frequency was not related to experiencing harm from drinking of family and friends, but in Sweden, Finland, and Norway, binge drinking frequency was related to higher estimates.
Women consistently rated harm more severely than men, as did those not living with a partner, those with lower education, and those living in urban settings. Severity of harm by age showed no consistent pattern across the study countries, with it peaking in the younger age groups in some countries and in the older age groups in others.
Estimation of correlates in multivariate models
Multivariate Poisson regression analyses estimating RR verified most descriptive findings (Table 5). Women had a significantly higher risk than men in all countries except in Scotland. In comparison with the oldest age group, younger age groups were at a higher risk in all countries with the majority of estimated RR being statistically significant. A clear age gradient was found in Sweden, Finland, and Norway, where the RR decreased with increasing age. Education, area of residence, family situation, and drinking frequency were not significantly related to the risk of being negatively affected heavy drinkers in the private life. However, binge drinking frequency was associated with an elevated risk in Finland and most significantly in Sweden and Norway.
Relative risk of being negatively affected by known drinkers.
Discussion
Comparisons of alcohol-related harm across countries are usually based on indicators of drinking and harms to the drinkers without considering harms to people around the drinkers. Therefore, the aim of this article was to compare the prevalence of harm from heavy drinkers in the private life in six Northern European countries differing, for example, with respect to population drinking and alcohol-related mortality. To our knowledge, this is the first comparative study of harm from the drinking of family and friends as reported in general population surveys.
There were significant cross-country differences in prevalence of harm from drinking of family and friends with estimates ranging from 14% to 28%. The prevalence was highest in Finland and Iceland, followed by Norway, whereas the estimates for Denmark Sweden, and Scotland were lower. These differences were not those expected based on known variations in population drinking, binge drinking frequency or liver cirrhosis mortality rates in these countries. Thus, all countries with high levels of drinking and high cirrhosis mortality rates did not report more harm from heavy drinking of family and friends. From a comparative perspective, these findings thus suggest that including this form of harm would modify comparisons based on consumption and liver cirrhosis mortality in these countries. For instance, estimates of alcohol-related harm will be relatively lower in Denmark and Scotland and relatively higher in Finland.
On the other hand, the self-reported severity of the harm caused by heavy drinkers was higher in Scotland and relatively low in Finland suggesting a somewhat different pattern when reported severity of harm was taken into account. This shows that it is important to take the severity of the harm into account in comparative studies. Overall, the results suggest that comparative studies on alcohol-related harm should consider including the measurement of this form of harm from others’ drinking as these problems do not necessarily follow differences obtained with the commonly applied indicators.
As to the correlates of being negatively affected by the heavy drinking of family and friends, in all countries women and younger respondents were more likely to report these problems than men and older respondents. The findings on gender differences are consistent with previous research addressing harm from others’ drinking.15–17 Moreover, previous research has consistently shown that younger individuals are more affected by others’ drinking than older age groups.9,15–17 However, it should be noted that we do not know whether this is related to age and gender differences in the exposure or variations in tolerance and threshold for experiencing others’ drinking as problematic. As to women, the difference is at least partly likely to be due to the fact that men consume about twice as much alcohol as women, which implies that men account for a larger share of alcohol problems affecting their female counterparts’ lives. Moreover, previous research has shown that women are significantly more likely to worry about others’ alcohol use than men are. 24 Finally, a recent study found that more women than men reported that their parents had alcohol problems. 15 This suggests that women could have a lower threshold for reporting problems related to family and friends.
Own drinking habits were related to experiences of harm from others’ drinking in the private life in half of our study countries where frequent binge drinking was significantly related in Sweden, Norway, and Finland. We have no explanation as to why a higher binge drinking frequency is important in these countries and not in the other countries. However, this relationship has been found in another research on third party harm and suggests that those who experience harm are also heavy drinkers themselves and most likely circulate in heavy drinking environments. 25 These findings may thus reflect that heavy drinkers are more strongly clustered in these countries, although we have no data to support this interpretation.
The differences between the national surveys with respect to data collection mode, sample sizes, response rates and also the wording of the questions in some cases are limitations that warrant caution in the interpretation of the observed differences between countries.
Furthermore, the questions used to measure experiences of harm from heavy drinkers in the private life were relatively crude and it was not possible to establish what kind of harm the respondents experienced and whether the patterns of such harm vary across countries.
Future research should aim at establishing a comparative design where data in the different countries are collected in a more similar manner and preferably include other countries that differ with regard to alcohol policies and drinking culture. It would also be valuable to include more detailed questions of the type of harm experienced from heavy drinkers in the private life and what relationship there is between the persons involved. Studies addressing differences in tolerance toward heavy drinking in general and among family and friends in particular, would also be important, both across countries and across subgroups of the population.
Conclusions
The findings from this study suggest substantial national differences in the proportion who experience negative effects of the heavy drinking by family and friends (14%–28%), and that these differences do not follow variations in indicators of drinking and other alcohol-related harm at the country level. Further, across all the countries included in this study, women carry a significantly larger burden of harm from heavy drinking in the private sphere. The findings suggest that in order to obtain a more complete picture of national and gender differences in alcohol-related harm, it is important to include harm from others’ drinking.
Author Contributions
Contributed to study design: MR, ES, ISM, EES, IOL, KB, AH, CT, SK. Analyzed the data: ES. Wrote the first draft of the manuscript: MR, ES. Contributed to the writing of the manuscript MR, ES, ISM, EES, IOL, KB, AH, CT, SK. Agree with manuscript results and conclusions: MR, ES, ISM, EES, IOL, KB, AH, CT, SK. Jointly developed the structure and arguments for the paper: MR, ES, ISM, EES, IOL, KB, AH, CT, SK. Made critical revisions and approved final version: MR, ES, ISM, EES, IOL, KB, AH, CT, SK. All authors reviewed and approved of the final manuscript.