Abstract

Background: This study aims to examine whether the Swedish-speaking minority in Finland differ from the Finnish-speaking majority in respect to alcohol consumption and, whether such differences could be explained by aspects of social capital measured by both individual and area level variables. Methods: This cross-sectional dataset consisted of 17 352 Finnish speakers (baseline response rate 40%) and 2018 Swedish speakers (baseline response rate 37%), aged 25–59 years. Multilevel logistic regression models were used to analyse the differences in alcohol consumption between the language groups, and to adjust for several potential individual and area level confounders. Results: Finnish-speaking men and women reported more frequent drunkenness, suffered more frequent hangovers, and had alcohol-induced pass-outs significantly more often than men and women in the Swedish-speaking population. Demographic, social, or environmental factors did not explain the observed differences in drinking patterns between these groups. Active social participation, social engagement, and trust in others were significantly related to drinking patterns only among Finnish speakers, but not among Swedish speakers. Conclusions: Drinking patterns are likely to have a direct impact on the health differences between the two populations, especially in relation to alcohol-related acute harm. It seems unlikely that the effect of social capital on the health differences between the two populations would be mediated through drinking patterns.

Introduction

Swedish-speaking Finns represent about 6% of the population of Finland. Typically ethnic minorities are in many ways disadvantaged compared with the majority of the population.1–3 The Swedish-speaking Finns are exceptional in that they have lower unemployment rates,4 a higher level of education,5 and greater marital stability6 than the Finnish speakers. The Swedish-speaking Finns also have a significant health advantage over the Finnish-speaking majority as measured by both total mortality7 and other health outcomes,8–10 and which cannot be explained easily by differences in demographic,11 environmental,12 genetic,13 or behavioural factors.11,14,15

There is some evidence that Swedish-speaking Finns would have more moderate drinking habits than Finnish speakers.11,16,17 However, these studies have either been very small in size,16 or alcohol consumption was not examined as an a priori measure.11,17 As such the evidence for the differences in alcohol consumption between the two language groups remains tentative. Since the adverse health effects of heavy alcohol consumption are well documented,18 possible differences in harmful alcohol consumption should be established in order to fully understand the health disparity between the two language groups.

It has been suggested that higher levels of social capital in the Swedish-speaking population might explain their health advantage over the Finnish speakers.17,19 Although Swedish speakers report more trust in others and more active social participation in some community engagements than the Finnish speakers,17,19 there is no empirical support for this hypothesis so far. One mechanism through which the beneficial effects of social capital could be mediated is alcohol consumption. Studies in Sweden and the United States have shown a negative correlation between measures of social participation and trust (the suggested two key dimensions of social capital) and heavy drinking patterns. Among Swedish men, a high alcohol intake (>250 g/week) has been associated with lower social participation and fewer contacts with friends and relatives.20 Another Swedish study among men found that high social participation combined with low trust was positively associated with a high alcohol intake (≥168 g/week).21 In the United States, high volunteerism at school among college students was associated with a lower probability of alcohol abuse, frequent drunkenness, and alcohol-related harm.22,23

The main aim of this paper is to examine whether, and to what extent, alcohol consumption differs between the Swedish speakers and the Finnish speakers. The Swedish-speaking Finns are expected to have higher levels of social participation and trust,17,19 potentially indicating higher levels of social capital. Thus, the second aim is to determine whether differences in alcohol consumption can be explained by measures of social capital. To provide information on possible mechanisms of alcohol-related health differences between populations and the relative contribution of social factors in that process.24 In addition, our data enable us to control for several individual and area level confounders. Because the Swedish-speaking population is geographically concentrated to the western and southern coastal regions of Finland, we will include municipality-level information on the area of residence in the analysis. Some of these variables can be considered to measure aspects of social capital (i.e. voting turnout and proportion of single parents).10 In order to take into account the possible area effects on alcohol consumption, we will utilize a multilevel method where municipality of residence is used as an area unit.

Methods

This study is part of the national Health and Social Support study of Finnish men and women of working-age: 20–24, 30–34, 40–44 and 50–54 years at baseline. The baseline (T1) postal survey questionnaire in 1998 yielded an overall response rate of 40% (N = 25 902). For specific research purposes, the data collection included two oversamples (N = 3629); one from a specific geographic area, and the other from the minority group of Swedish-speaking Finns (baseline response rate 37%). The oversamples were collected by using the same protocol as was used for the main sample, but only with a higher representation. The same questionnaire was used in a follow-up survey in 2003 (T2), but with an additional question of mistrust. During follow-up 216 participants died, 234 emigrated and 969 were unreachable due to an unknown postal address. After excluding those lost to follow up, the overall response rate of the T2 survey was 80% (N = 19 629), representing ∼31% of the original sample. According to baseline non-response analysis (without the oversamples), the T1 study population is representative of the Finnish general population, and there is no significant selection due to health related factors.25 After excluding those with missing information on measures of drinking pattern (N = 259), the final dataset consisted of 17 352 Finnish speakers and 2018 Swedish speakers. All data in this analysis are derived from the T2 questionnaire. The ethical review board of Turku University Hospital has approved the study protocol.

Individual level variables

Based on the official language status, the respondent was classified as a Finnish or Swedish speaker. The official language status is based on information derived from the national population register, and is linked using a personal identification number. Because Swedish is an official language in Finland, Swedish-speaking Finns have certain constitutional rights, if they register their mother tongue as Swedish. Individual level variables thought to be potential confounders that were controlled for in the analysis included: gender, age, educational level (high/low), whether respondent was living alone (yes/no), and unemployment (yes/no).26

Individual level social engagement was assessed as at least monthly engagement in the following leisure time social activities: organizational participation, cultural participation, participation in religious events, and visiting relatives and friends. The frequency of visiting restaurants/pubs was also measured. This can be considered as a form of social engagement, but also as a measure of drinking pattern because drinking at restaurants and pubs is associated with higher intake and more frequent drunkenness than drinking, e.g. at home. The choice of cutoff was based on the distributions of the responders across the various frequency levels with the aim of capturing aspects of social participation that are more frequent than just occasional, thus indicating a potentially stronger impact on health behaviour. Expressed mistrust towards others was based on the Cook-Medley Cynical Hostility Scale.27 We used one of the several statements to measure mistrust (‘It is better not to trust anyone’), where the response options on a five-point scale ranged from ‘I totally disagree’ to ‘I totally agree’. A dichotomous measure was used to indicate mistrust for those responders who chose either ‘I totally agree’ or ‘I somewhat agree’.

Other individual level measures of social characteristics were social support and negative childhood experiences. Social support was measured by the Brief Social Support Questionnaire.28 Lowest decile (score under 6) of the sum score (range 0–36) was used as a cutoff point for low social support.29 We used information on parental divorce30 and the presence of an alcohol problem in a childhood family member as an indicator of a childhood social environment. Only consistent ‘yes’ responses on both T1 and T2 were used. The presence of an alcohol problem in a family member was considered an indication of a familial predisposition to heavy drinking.31 The inclusion of childhood social environment provides a life-course perspective32 into the development of values and attitudes towards social engagement in later life.

Measures of alcohol consumption included overall annual frequency of drinking, beverage specific quantity of intake, and pattern of drinking. For the estimation of total quantity of alcohol intake, the respondent was asked to estimate their average consumption, by type of alcoholic beverage (see Appendix 1). The beverage specific consumption was converted to grams of absolute alcohol and summed up as total weekly consumption. Abstainers were classified as participants who had not consumed any alcohol in their lifetime or those who had not used any alcohol during the preceding year of the survey. Pattern of drinking was measured by annual frequency of drunkenness, and number of hangovers and alcohol-induced pass-outs (in Finnish and Swedish, there is a specific expression for this loss of consciousness due to drinking too much alcohol). The term used for alcohol-induced pass-out refers to loss of consciousness without reference to loss of memory (blackout). For the purposes of this study, we dichotomized the measures of alcohol consumption using a cutoff point of ‘At least twice weekly’ for frequency of drinking, indicating possibly heavy drinking, but also frequent social drinking such as drinking with meals. Heavy weekly intake was defined as >260 g/week for men, and >170 g/week for women.33 The cutoff point for categorizing the frequency of drunkenness and frequency of hangovers was set at ‘At least monthly’ and the cutoff point for pass-outs was set at ‘At least once a year’.34

Area level variables

Municipality of residence of the respondent was selected as the area unit in the study. A finer classification of postal code areas could have been used here, but because of the sample size, we chose to use municipalities as the area unit. In Finland, there were 446 municipalities in 2003. Our data consist of 435 municipalities with a population size varying from 233 to 559 330 (with a mean size of 12 083). Some of the smallest municipalities were not represented in the data because of no respondents.

Area level information was obtained through public databases of the Statistic Finland and the National Research and Development Centre for Welfare and Health (SOTKA database). We used the following measures to indicate the area level social environment of the respondents: Unemployment rate (range 1–30%), Educational index (range 152–531), which indicates the average length (in years) of the highest educational qualification after compulsory basic education per person in a population aged over 19 (e.g. the value 531 shows that the average length of education is 5.31years per person after basic education), voting turnout (range 51–90%) in most recent municipality elections (year 2004), proportion of Swedish speakers in the population (range <1–96%), proportion of 16–64 years old (range 54–71%), proportion of single parents (range 4–30%), population density (range 1–3010; persons living per square kilometre) and level of urbanization indicating whether the respondent lived in a city or in a rural area. All measures, except the level of urbanization and the proportion of Swedish-speakers, were categorized into quartiles. For the proportion of Swedish speakers, a finer categorization was used, because 96% of the Finnish speakers in our data were living in areas where the proportion of Swedish speakers was <11%. All area variables, with the exception of voting turnout, were describing the situation in 2003.

Statistical analysis

We used SAS 9.1 procedure add-on PROC GLIMMIX to fit multilevel models to the data. Two-level logistic regression models with random intercepts were fitted to take into account the hierarchical structure of the data. The results are presented as odds ratios (OR) and their 95% confidence intervals (95% CI). When adjusting for confounders, new variables were added into the models in a stepwise manner, keeping all previous variables in the models. Effect modification by language group was tested by the Breslow–Day test.35

Results

The data consisted of 17 352 Finnish-speaking men (38%) and women, and 2018 Swedish-speaking men (41%) and women. Table 1 shows the distributions of selected background variables according to gender and language group. The Swedish speakers were socially more active in respects of organizational participation and visiting relatives and friends than the Finnish speakers. The Swedish-speaking men also participated more often in cultural events than the Finnish-speaking men. There were no differences between language groups in reporting mistrust towards others. The Swedish speakers differed from Finnish speakers in relation to measures of drinking pattern. In particular, the Finnish-speaking men reported considerably more drunkenness, hangovers, and pass-outs than the Swedish-speaking men. However, the average intake was similar in both language groups. For area-level variables, the Swedish speakers lived in areas characterized by a low unemployment rate (P < 0.001), a high-voting turnout (P < 0.001) and a low proportion of single parents (P < 0.001).

Table 1

Background characteristics of the study population by gender and language group

 Men
 
P-value Women
 
P-value 
 Language group
 
 Language group
 
 
 Finnish Swedish  Finnish Swedish  
No. of subjects (N6651 829  10 701 1189  
Age group (%)   0.890   0.865 
    25–29 21 21  28 27  
    35–39 22 22  23 24  
    45–49 25 25  25 24  
    55–59 32 33  25 25  
Living alone 20 16 0.007 18 16 0.310 
Higher education 48 58 <0.001 61 69 <0.001 
Unemployed <0.001 <0.001 
Social support (low) 15 12 0.046 0.273 
Negative childhood experiences (yes)       
    Parental divorce 15 11 0.005 18 12 <0.001 
    Family history of alcohol problems 23 11 <0.001 28 15 <0.001 
Social participation (at least monthly)       
    Organisational participation 20 33 <0.001 18 29 <0.001 
    Visiting relatives and friends 68 78 <0.001 81 84 <0.001 
    Cultural participation 21 27 <0.001 28 31 0.039 
    Religious participation 0.003 13 <0.001 
Mistrust (yes) 22 24 0.157 18 18 0.942 
Abstainer 0.336 0.285 
Age at first drink (median) 15 15  16 15  
Visits restaurants/pubs (at least monthly) 31 29 0.173 26 23 0.009 
Frequent drinker (2+ times weekly) 32 36 0.011 17 18 0.465 
Amount of consumption (grams/week), median 77 78  30 30  
Heavy drinkera <0.001 0.080 
Drunkenness (at least once a month) 37 28 <0.001 16 11 <0.001 
Hangover (at least once a month) 22 15 <0.001 <0.001 
Pass-out (at least once a year) 19 12 <0.001 <0.001 
 Men
 
P-value Women
 
P-value 
 Language group
 
 Language group
 
 
 Finnish Swedish  Finnish Swedish  
No. of subjects (N6651 829  10 701 1189  
Age group (%)   0.890   0.865 
    25–29 21 21  28 27  
    35–39 22 22  23 24  
    45–49 25 25  25 24  
    55–59 32 33  25 25  
Living alone 20 16 0.007 18 16 0.310 
Higher education 48 58 <0.001 61 69 <0.001 
Unemployed <0.001 <0.001 
Social support (low) 15 12 0.046 0.273 
Negative childhood experiences (yes)       
    Parental divorce 15 11 0.005 18 12 <0.001 
    Family history of alcohol problems 23 11 <0.001 28 15 <0.001 
Social participation (at least monthly)       
    Organisational participation 20 33 <0.001 18 29 <0.001 
    Visiting relatives and friends 68 78 <0.001 81 84 <0.001 
    Cultural participation 21 27 <0.001 28 31 0.039 
    Religious participation 0.003 13 <0.001 
Mistrust (yes) 22 24 0.157 18 18 0.942 
Abstainer 0.336 0.285 
Age at first drink (median) 15 15  16 15  
Visits restaurants/pubs (at least monthly) 31 29 0.173 26 23 0.009 
Frequent drinker (2+ times weekly) 32 36 0.011 17 18 0.465 
Amount of consumption (grams/week), median 77 78  30 30  
Heavy drinkera <0.001 0.080 
Drunkenness (at least once a month) 37 28 <0.001 16 11 <0.001 
Hangover (at least once a month) 22 15 <0.001 <0.001 
Pass-out (at least once a year) 19 12 <0.001 <0.001 

P-value for the difference between language groups (Chi-square test, two-tailed). Health and Social Support study

a: Intake over 260 g/week for men and for women intake over 170 g/week

Table 2 shows logistic regression models for the Finnish speakers to get drunk at least monthly, suffer a hangover at least monthly and having a pass-out at least once a year compared with the Swedish speakers. Adjusting for all available variables attenuated the estimates only slightly. Largest attenuation of point estimates was in relation to pass-outs. Table 3 shows additional models where available area level variables are adjusted for. When all available individual level and area level variables were adjusted for, the Finnish speakers were still significantly more likely to report drunkenness, hangovers, and pass-outs compared with the Swedish speakers.

Table 2

The ORs with 95% CI for Finnish speakers to get drunk at least monthly or suffer a hangover at least monthly or have a pass-out at least once a year, compared to Swedish speakers

 Drunkenness
 
Hangover
 
Pass-out
 
 OR 95% CI OR 95% CI OR 95% CI 
Model 1 (base model) 1.58 1.37–1.82 1.62 1.36–1.94 1.70 1.42–2.04 
Model 2 (childhood) 1.51 1.31–1.75 1.52 1.27–1.82 1.57 1.31–1.88 
Model 3 1.48 1.29–1.71 1.48 1.24–1.77 1.49 1.24–1.79 
Model 4 (social capital) 1.55 1.35–1.79 1.58 1.32–1.89 1.64 1.37–1.97 
Model 5 1.46 1.27–1.69 1.45 1.21–1.74 1.46 1.21–1.75 
Model 6 (full model) 1.46 1.27–1.69 1.42 1.18–1.70 1.40 1.17–1.69 
 Drunkenness
 
Hangover
 
Pass-out
 
 OR 95% CI OR 95% CI OR 95% CI 
Model 1 (base model) 1.58 1.37–1.82 1.62 1.36–1.94 1.70 1.42–2.04 
Model 2 (childhood) 1.51 1.31–1.75 1.52 1.27–1.82 1.57 1.31–1.88 
Model 3 1.48 1.29–1.71 1.48 1.24–1.77 1.49 1.24–1.79 
Model 4 (social capital) 1.55 1.35–1.79 1.58 1.32–1.89 1.64 1.37–1.97 
Model 5 1.46 1.27–1.69 1.45 1.21–1.74 1.46 1.21–1.75 
Model 6 (full model) 1.46 1.27–1.69 1.42 1.18–1.70 1.40 1.17–1.69 

Abstainers excluded. Health and Social Support study Model 1 = gender, age; Model 2 = + parental divorce, family history of alcohol problems; Model 3 = + education, living alone, unemployment; Model 4 = gender, age, mistrust, organisational participation, cultural participation, visiting relatives and friends, religious participation; Model 5 = Model 4 + parental divorce, family history of alcohol problems, education, living alone, unemployment; Model 6 = Model 5 + visits restaurants/pubs

Table 3

The ORs with 95% CI for Finnish speakers to get drunk at least monthly or suffer a hangover at least monthly or having a pass-out at least once a year, compared to Swedish speakers

 Drunkenness
 
Hangover
 
Pass-out
 
 OR 95% CI OR 95% CI OR 95% CI 
Model 1 (base model) 1.58 1.37–1.82 1.62 1.36–1.94 1.70 1.42–2.04 
Model 2 1.53 1.33–1.89 1.58 1.33–1.89 1.68 1.40–2.02 
Model 3 1.55 1.35–1.79 1.59 1.33–1.91 1.65 1.37–1.98 
Model 4 1.48 1.29–1.70 1.51 1.26–1.81 1.56 1.29–1.88 
Model 5 1.49 1.30–1.72 1.53 1.28–1.83 1.59 1.31–1.92 
Model 6 1.51 1.31–1.73 1.54 1.28–1.85 1.62 1.33–1.96 
Model 7 1.41 1.20–1.67 1.48 1.20–1.82 1.54 1.23–1.93 
Model 8 (full model) 1.36 1.14–1.63 1.37 1.10–1.71 1.35 1.07–1.71 
 Drunkenness
 
Hangover
 
Pass-out
 
 OR 95% CI OR 95% CI OR 95% CI 
Model 1 (base model) 1.58 1.37–1.82 1.62 1.36–1.94 1.70 1.42–2.04 
Model 2 1.53 1.33–1.89 1.58 1.33–1.89 1.68 1.40–2.02 
Model 3 1.55 1.35–1.79 1.59 1.33–1.91 1.65 1.37–1.98 
Model 4 1.48 1.29–1.70 1.51 1.26–1.81 1.56 1.29–1.88 
Model 5 1.49 1.30–1.72 1.53 1.28–1.83 1.59 1.31–1.92 
Model 6 1.51 1.31–1.73 1.54 1.28–1.85 1.62 1.33–1.96 
Model 7 1.41 1.20–1.67 1.48 1.20–1.82 1.54 1.23–1.93 
Model 8 (full model) 1.36 1.14–1.63 1.37 1.10–1.71 1.35 1.07–1.71 

Controlling for area level variables. Abstainers excluded. Health and social support study Model 1 = gender, age; Model 2 = model 1 + proportion of 16–64 years old; Model 3 = model 2 + educational index; Model 4 = model 3 + level of urbanisation, population density; Model 5 = model 4 + unemployment rate; Model 6 = model 5 + proportion of single parents; Model 7 = model 6 + voting turnout Model 8 = model 7 + proportion of Swedish speakers; Model 9 = all area level variables + all individual level variables shown in table 2

Table 4 shows the association between social characteristics and drinking pattern by language group. Among the Finnish speakers, all measures of social characteristics, except social support and cultural participation, were associated with drunkenness, hangovers, and alcohol-induced pass-outs. Among the Swedish speakers, only measures of living alone, visiting restaurants and pubs, and religious participation were associated with drinking patterns, and which was significantly different to that of the Finnish speakers.

Table 4

The ORsa with 95% CI for selected background factors associated with drunkenness at least monthly, suffering a hangover at least monthly and having a pass-out at least once a year by language group

 Finnish speakers (n = 17 352)
 
Swedish speakers (n = 2018)
 
 Drunkenness (n = 1655)
 
Hangover (n = 1881)
 
Pass-out (n = 1853)
 
Drunkenness (n = 353)
 
Hangover (n = 188)
 
Pass-out (n = 153)
 
 OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI 
Living alone (yes/no) 1.83* 1.67–2.00 2.06* 1.86–2.29 1.64 1.47–1.83 2.35* 1.74–3.19 3.35* 2.35–4.79 1.81 1.22–2.69 
Higher education (yes/no) 0.77 0.71–0.83 0.71 0.64–0.78 0.61 0.55–0.67 0.90 0.69–1.18 0.87 0.62–1.23 0.61 0.44–0.86 
Unemployed (yes/no) 1.52 1.33–1.74 1.93 1.65–2.25 1.91 1.64–2.22 1.21 0.63–2.34 1.86 0.86–4.03 2.04 1.01–4.13 
Parental divorce (yes/no) 1.32 1.20–1.44 1.40 1.26–1.56 1.37 1.22–1.53 1.17 0.83–1.65 1.78 1.19–2.65 1.59 1.04–2.43 
Family history of alcohol problems (yes/no) 1.43 1.32–1.55 1.65 1.50–1.82 1.78 1.62–1.97 1.27 0.90–1.79 1.56 1.03–2.38 2.14 1.43–3.21 
Visiting restaurants/pubs (monthly/less) 5.74 5.28–6.23 6.35 5.74–7.02 2.82 2.56–3.11 5.30 4.01–7.00 6.01 4.16–8.68 3.13 2.18–4.48 
Religious participation (at least monthly) 0.32 0.28–0.38 0.34 0.28–0.42 0.49 0.40–0.59 0.43 0.28–0.69 0.50 0.28–0.90 0.43 0.22–0.83 
Cultural participation (monthly/less) 1.21 1.11–1.32 1.16 1.04–1.28 0.92* 0.82–1.02 1.22 0.93–1.61 1.23 0.87–1.74 1.32* 0.92–1.88 
Visiting relatives and friends (monthly/less) 0.89 0.81–0.97 0.83 0.74–0.93 0.85* 0.76–0.95 1.04 0.72–1.51 0.85 0.53–1.35 1.56* 0.91–2.68 
Organizational participation (monthly/less) 0.86 0.78–0.95 0.84 0.75–0.95 0.86 0.77–0.97 0.93 0.71–1.23 0.90 0.63–1.28 0.83 0.57–1.19 
Mistrust (yes/no) 1.41 1.28–1.55 1.55 1.39–1.73 1.59 1.42–1.77 1.22 0.89–1.68 1.25 0.83–1.88 1.45 0.98–2.15 
 Finnish speakers (n = 17 352)
 
Swedish speakers (n = 2018)
 
 Drunkenness (n = 1655)
 
Hangover (n = 1881)
 
Pass-out (n = 1853)
 
Drunkenness (n = 353)
 
Hangover (n = 188)
 
Pass-out (n = 153)
 
 OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI OR 95% CI 
Living alone (yes/no) 1.83* 1.67–2.00 2.06* 1.86–2.29 1.64 1.47–1.83 2.35* 1.74–3.19 3.35* 2.35–4.79 1.81 1.22–2.69 
Higher education (yes/no) 0.77 0.71–0.83 0.71 0.64–0.78 0.61 0.55–0.67 0.90 0.69–1.18 0.87 0.62–1.23 0.61 0.44–0.86 
Unemployed (yes/no) 1.52 1.33–1.74 1.93 1.65–2.25 1.91 1.64–2.22 1.21 0.63–2.34 1.86 0.86–4.03 2.04 1.01–4.13 
Parental divorce (yes/no) 1.32 1.20–1.44 1.40 1.26–1.56 1.37 1.22–1.53 1.17 0.83–1.65 1.78 1.19–2.65 1.59 1.04–2.43 
Family history of alcohol problems (yes/no) 1.43 1.32–1.55 1.65 1.50–1.82 1.78 1.62–1.97 1.27 0.90–1.79 1.56 1.03–2.38 2.14 1.43–3.21 
Visiting restaurants/pubs (monthly/less) 5.74 5.28–6.23 6.35 5.74–7.02 2.82 2.56–3.11 5.30 4.01–7.00 6.01 4.16–8.68 3.13 2.18–4.48 
Religious participation (at least monthly) 0.32 0.28–0.38 0.34 0.28–0.42 0.49 0.40–0.59 0.43 0.28–0.69 0.50 0.28–0.90 0.43 0.22–0.83 
Cultural participation (monthly/less) 1.21 1.11–1.32 1.16 1.04–1.28 0.92* 0.82–1.02 1.22 0.93–1.61 1.23 0.87–1.74 1.32* 0.92–1.88 
Visiting relatives and friends (monthly/less) 0.89 0.81–0.97 0.83 0.74–0.93 0.85* 0.76–0.95 1.04 0.72–1.51 0.85 0.53–1.35 1.56* 0.91–2.68 
Organizational participation (monthly/less) 0.86 0.78–0.95 0.84 0.75–0.95 0.86 0.77–0.97 0.93 0.71–1.23 0.90 0.63–1.28 0.83 0.57–1.19 
Mistrust (yes/no) 1.41 1.28–1.55 1.55 1.39–1.73 1.59 1.42–1.77 1.22 0.89–1.68 1.25 0.83–1.88 1.45 0.98–2.15 

Abstainers excluded. Statistically significant estimates are given in bold type. Health and Social Support study

a: Models adjusted for gender, age and each variable separately

*: P < 0.05 for Breslow-Day test of homogeneity of OR

Discussion

This study shows that in Finland, the Finnish-speaking majority has significantly more harmful drinking patterns than the Swedish-speaking minority. Finnish-speaking men and women report more frequent drunkenness, suffer more frequent hangovers, and have more alcohol-induced pass-outs (loss of consciousness) than the Swedish-speaking population. Our results also show that the differences in demographic, social, or environmental factors do not explain the differences in drinking patterns between the two populations.

Methodological considerations

The strength of our study is the large number of Swedish speakers in our data compared with earlier studies.11,16,17 In addition, we had detailed measurement of drinking patterns, and both individual and area level social factors, enabling us to control for several potential confounders. Undeniably, a possible limitation of our data is the low baseline response rate. The non-response analysis showed that the data without the oversamples is representative of the Finnish general population according to several demographic and health-related factors.25 However, it should be noted that there still might be selection due to some other factors that were not included in the non-response analysis. Although the original non-response analysis25 could not be analysed by language group, our comparisons did not suggest bias by differences in relative representativeness, i.e. if there is bias due to selection, the selection would have been largely similar in both language groups. Therefore, we can conclude that the findings would apply to those populations represented in our data. All individual level variables were based on self-reports, so there is a possibility of reporting bias. Because Swedish-speaking Finns are well integrated into the Finnish society6, it does not seem plausible that reporting alcohol consumption would be less socially acceptable among the Swedish speakers than among the Finnish speakers. It is unlikely that the observed differences in drinking patterns would be explained by the translations or wordings of the questions because the words describing drunkenness, hangover, and alcohol-induced pass-out have specific and comparable meanings in both languages. All models were replicated without the oversamples, and these models showed that the results were not biased by these inclusions.

For practical reasons, we used the municipality of residence as a proxy for social environment. It can be argued that some other level of measurement of social context, such as peer groups may have been more relevant for studying alcohol consumption. However, the relationships between individual behaviour and social contexts are complex,36 which makes it challenging to capture all the relevant aspects of these factors in epidemiological studies. Individuals may be influenced by various social environments,37 and their effects on behaviour may not necessarily be in the same direction. The relationships between individuals and social contexts may also vary over time,37 making it difficult to establish which one affects which, or whether the relationships are reciprocal. In relation to social behaviour, an individual's own perception of the environment may be more important than the objective characteristics of the environment.36

Differences in alcohol consumption

Despite of potential problems introduced by the low baseline response proportion, our results corroborate the earlier findings of more moderate drinking patterns among the Swedish-speaking population compared with the Finnish-speaking majority.11,16,17 In addition, we were able to extend the understanding about the differences in drinking patterns, and it seems plausible that the differences in the patterns of alcohol drinking make a direct contribution to the health differences between the Finnish and Swedish speakers. This is more likely to be related alcohol-related acute conditions or other conditions related to binge drinking than for long-term conditions such as alcoholic cirrhosis, because the average amount of consumption was similar in both groups.

Although Swedish speakers have a more favourable social profile, these factors do not explain the observed differences in drinking patterns between the Swedish and Finnish speakers. Gradual adjustment of the variables into the models showed that none of the variables had a strong impact on the associations. Our results were in agreement with earlier studies showing a protective effect of social participation, social engagement, and trust in others on harmful drinking patterns,21,23 but these associations were significant only among the Finnish-speaking majority. There are some potential explanations for this. Although we had a large number of Swedish speakers in our data, the groups with harmful drinking patterns were relatively small and there may not have been enough statistical power to detect an association among the Swedish speakers. However, among the Swedish speakers, the largest number of significant estimates was among those having pass-outs, i.e. the group of drinkers with least statistical power due to its size. Theoretically, this is not surprising because pass-outs can be considered as an indication of a severe form of alcohol abuse and might be more strongly related to social disadvantage. In contrast, frequent drunkenness and number of hangovers may not be so closely related to social exclusion38 among the Swedish speakers than among the Finnish speakers. This may result from more tolerable social norms or that there are qualitative differences in similar drinking patterns across language groups that cannot be captured by measuring just frequencies of specific drinking behaviour.39

Our analysis was based on cross-sectional information which means that causality and direction of the effect cannot be established here. Further research is needed to find out which social factors are more strongly predictive of harmful drinking patterns and the mechanisms through which these factors operate. Furthermore, research is needed to determine whether there are qualitative differences in drinking patterns between the two language groups. This information could help us to understand the complex reciprocal process between social factors and alcohol consumption.40,41

In conclusion, the Swedish-speaking population in Finland has significantly less harmful drinking patterns than the Finnish-speaking majority. Drinking patterns are likely to have a direct impact on the health differences between the two populations, especially in relation to alcohol-related acute harm. It seems, however, unlikely that the effects of social capital on the health differences between the two populations is mediated through drinking patterns.

Funding

A grant from the Academy of Finland was awarded for this study (#118551).

Conflicts of interest: None declared.

Acknowledgements

We are in debt of gratitude to MSc Hans Helenius for his invaluable assistance in conducting the multilevel analysis, and to Dr Naomi Allen for her help with the write-up of the manuscript.

Key points

  • In earlier studies individual and environmental level social factors are found to be related with health and health behaviour.

  • Although several demographic and social factors seem to favour the Swedish speakers compared with the Finnish speakers, these factors cannot fully explain the observed differences in drinking patterns.

  • Our results underline the importance of taking into account ethnicity when studying the relationships between social factors and health behaviour.

  • From health promotion point of view, our results suggest that promoting community engagement and social participation does not necessarily have the same impact on health behaviours across all population groups.

References

1
Balarajan
R
Ethnic differences in mortality from ischaemic heart disease and cerebrovascular disease in England and Wales
Br Med J
 , 
1991
, vol. 
9
 (pg. 
560
-
4
)
2
Davey Smith
G
Neaton
JD
Wentworth
D
, et al.  . 
Mortality differences between black and white men in the USA: contribution of income and other risk factors among men screened for the MRFIT. MRFIT Research Group. Multiple risk factor intervention trial
Lancet
 , 
1998
, vol. 
28
 (pg. 
934
-
9
)
3
Bos
V
Kunst
AE
Keij-Deerenberg
IM
, et al.  . 
Ethnic inequalities in age- and cause-specific mortality in the Netherlands
Int J Epidemiol
 , 
2004
, vol. 
33
 (pg. 
1112
-
19
)
4
Saarela
J
Finnäs
F
Unemployment and native language: the Finnish case
J Socio Econ
 , 
2003
, vol. 
32
 (pg. 
59
-
80
)
5
Saarela
J
Finnäs
F
Social background and education of Swedish and Finnish speakers in Finland
Eur J Educ
 , 
2003
, vol. 
38
 (pg. 
445
-
56
)
6
Finnäs
F
Social integration, heterogeneity, and divorce: the case of the Swedish-speaking population in Finland
Acta Sociol
 , 
1997
, vol. 
40
 (pg. 
263
-
77
)
7
Näyhä
S
Geographic variations in cardiovascular mortality in Finland 1961–1985
Scand J Soc Med
 , 
1989
, vol. 
40
 
Suppl
(pg. 
1
-
48
)
8
Salminen
S
Johansson
A
Occupational accidents of Finnish- and Swedish-speaking workers in Finland: a mental model view
Int J Occup Saf Ergon
 , 
2000
, vol. 
6
 (pg. 
293
-
302
)
9
Suominen
SB
Välimaa
RS
Tynjälä
JA
, et al.  . 
Minority status and perceived health: a comparative study of Finnish- and Swedish-speaking schoolchildren in Finland
Scand J Public Health
 , 
2000
, vol. 
28
 (pg. 
179
-
87
)
10
Blomgren
J
Martikainen
P
Mäkelä
P
, et al.  . 
The effects of regional characteristics on alcohol-related mortality—a register-based multilevel analysis of 1.1 million men
Soc Sci Med
 , 
2004
, vol. 
58
 (pg. 
2523
-
35
)
11
Hyyppä
M
Mäki
J
Why do Swdish-speaking Finns have longer active life? An area for social capital research
Health Promot Int
 , 
2001
, vol. 
16
 (pg. 
55
-
64
)
12
Saarela
J
Finnäs
F
Geographical extraction and the Finnsh-Swedish health differential in Finland
Yearbook Popul Res Finland
 , 
2005
, vol. 
41
 (pg. 
61
-
73
)
13
Virtaranta-Knowles
K
Sistonen
P
Nevanlinna
HR
A Population genetic study in Finland: comparison of the Finnish- and Swedish-speaking populations
Hum Hered
 , 
1989
, vol. 
41
 (pg. 
248
-
64
)
14
Koskinen
S
Martelin
T
Why is mortality low among the Swedish-speaking minority in Finland?
Yearbook Popul Res Finland
 , 
2003
, vol. 
39
 (pg. 
15
-
31
)
15
Saarela
J
Finnäs
F
Mortality inequality in two native population groups
Popul Stud
 , 
2005
, vol. 
59
 (pg. 
313
-
20
)
16
Simpura
J
Hur dricker finlandssvensken? [How do Swedish-speaking population in Finland drink?]
Finska Läkaresällskapets Handlingar
 , 
1990
, vol. 
150
 (pg. 
168
-
70
)
17
Hyyppä
M
Mäki
J
Social participation and health in a community rich in stock of social capital
Health Educ Res
 , 
2003
, vol. 
18
 (pg. 
770
-
9
)
18
Room
R
Babor
T
Rehm
J
Alcohol and Public Health
Lancet
 , 
2005
, vol. 
365
 (pg. 
519
-
30
)
19
Hyyppä
M
Mäki
J
Individual-level relationships between social capital and self-rated health in a bilingual community
Prev Med
 , 
2001
, vol. 
32
 (pg. 
148
-
55
)
20
Hanson
BS
Social network, social support and heavy drinking in elderly men—a population study of men born in 1914, Malmö, Sweden
Addiction
 , 
1994
, vol. 
89
 (pg. 
725
-
32
)
21
Lindström
M
Social capital, the minituarization of community and high alcohol consumption: a population-based study
Alcohol Alcohol
 , 
2005
, vol. 
40
 (pg. 
556
-
62
)
22
Weitzman
E
Kawachi
I
Giving means receiving: the protective effect of social capital on binge drinking on college campuses
Am J Public Health
 , 
2000
, vol. 
90
 (pg. 
1936
-
39
)
23
Weitzman
E
Chen
Y
Risk modifying effect of social capital on measures of heavy alcohol consumption, alcohol abuse, harms, and secondhand effects: national survey findings
J Epidemiol Community Health
 , 
2005
, vol. 
59
 (pg. 
303
-
9
)
24
Rehm
J
Fischer
B
Graham
K
, et al.  . 
The importance of environmental modifiers of the relationship between substance use and harm
Addiction
 , 
2004
, vol. 
99
 (pg. 
663
-
66
)
25
Korkeila
K
Suominen
S
Ahvenainen
J
, et al.  . 
Non-response and related factors in a nation-wide health survey
Eur J Epidemiol
 , 
2001
, vol. 
17
 (pg. 
991
-
99
)
26
Poikolainen
K
Preedy
VR
Watson
RR
Predictors of alcohol consumption
Comprehensive handbook of alcohol related pathology, vol 1.
 , 
2005
Amsterdam
Elsevier Ltd
(pg. 
49
-
58
)
27
Cook
W
Medley
D
Proposed hostility and pharisaic-virtue scales for the MMPI
J Appl Psychol
 , 
1954
, vol. 
38
 (pg. 
414
-
8
)
28
Sarason
IG
Sarason
BR
Shearin
EN
, et al.  . 
A brief measure of social support: practical and theoretical implications
J Soc Pers Relat
 , 
1987
, vol. 
4
 (pg. 
497
-
510
)
29
Korkeila
K
Korkeila
J
Vahtera
J
, et al.  . 
Childhood adversities, adult risk factors and depressiveness: a population study
Soc Psychiatry Psychiatr Epidemiol
 , 
2005
, vol. 
40
 (pg. 
700
-
6
)
30
Hope
S
Power
C
Rodgers
B
The relationship between parental separation in childhood and problem drinking in adulthood
Addiction
 , 
1998
, vol. 
93
 (pg. 
505
-
14
)
31
Anda
RF
Whitfield
CL
Felitti
VJ
, et al.  . 
Adverse childhood experiences, alcoholic parents, and later risk of alcoholism and depression
Psychiatr Serv
 , 
2002
, vol. 
53
 (pg. 
1001
-
9
)
32
Ben-Shlomo
Y
Kuh
D
A life-course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives
Int J Epidemiol
 , 
2002
, vol. 
31
 (pg. 
285
-
93
)
33
Seppä
K
Lepistö
J
Sillanaukee
P
Five-Shot Questionnaire on heavy drinking
Alcohol Clin Exp Res
 , 
1998
, vol. 
22
 (pg. 
1788
-
91
)
34
Kauhanen
J
Kaplan
GA
Goldberg
DD
, et al.  . 
Frequent hangovers and cardiovascular mortality in middle-aged men
Epidemiology
 , 
1997
, vol. 
8
 (pg. 
310
-
4
)
35
Breslow
NE
Day
NE
Statistical methods in cancer research, volume I: the analysis of case-control studies, IARC Scientific Publications, No. 32.
 , 
1980
Lyon, France
International Agency for Research on Cancer
36
Diex-Roux
A
Investigating neighbourhood and area effects on health
Am J Public Health
 , 
2001
, vol. 
91
 (pg. 
1783
-
89
)
37
Cummins
S
Curtis
S
Diez-Roux
A
, et al.  . 
Understanding and representing ‘place’ in health research: a relational approach
Soc Sci Med
 , 
2007
, vol. 
65
 (pg. 
1825
-
38
)
38
Baum
F
Bush
R
Modra
C
, et al.  . 
Epidemiology of participation: an Australian community study
J Epidemiol Community Health
 , 
2000
, vol. 
54
 (pg. 
414
-
23
)
39
Demers
A
When at risk? Drinking contexts and heavy drinking in the Montreal adult population
Contemp Drug Prob
 , 
1997
, vol. 
24
 (pg. 
449
-
71
)
40
Room
R
Intoxication and bad behavior: understanding cultural differences in the link
Soc Sci Med
 , 
2001
, vol. 
53
 (pg. 
189
-
98
)
41
Galea
S
Nandi
A
Vlahov
D
The social epidemiology of substance use
Epidemiol Rev
 , 
2004
, vol. 
26
 (pg. 
36
-
52
)

Appendix 1 The original format and order of the selected survey questions in the Health and Social Support survey

Drinking frequency was asked as ‘How often do you drink alcohol nowadays? Which one of the following alternatives best describes your consumption of beer, wine and spirits?’

  1. I don't use alcohol

  2. Once a year or less

  3. Three to four times a year

  4. Once in 2 months

  5. One to two times in a month

  6. Once a week

  7. Couple of times a week

  8. Daily or almost daily

Frequency of intoxications and hangovers was asked as ‘How often during the past 12 months have you been intoxicated/suffered a hangover?’

  1. Never

  2. Once

  3. Two to three times

  4. Four to five times

  5. Once in 2 months

  6. Once a month

  7. Two to three times a month

  8. Once a week

  9. Couple of times a week or more

For the estimation of total quantity of alcohol consumption, the respondent was asked to estimate one's beverage specific average consumption.

‘How much do you drink the following alcoholic beverages on average?’

Beer in a week

  1. Not at all

  2. Less than a bottle (0.33l)

  3. 1–4 bottles

  4. 5–12 bottles

  5. 13–24 bottles

  6. 25–47 bottles

  7. Over 48 bottles

Wine or other equivalent mild alcoholic beverages in a week

  1. Not at all

  2. Less than a glass

  3. 1–4 glasses

  4. 1–2.5 bottles

  5. 3–4.5 bottles

  6. 5–9 bottles

  7. over 10 bottles

Spirits in a month

  1. Not at all

  2. Less than half a bottle (0.5l)

  3. 0.5–1.5 bottles

  4. 2–3.5 bottles

  5. 4–9 bottles

  6. 10–19 bottles

  7. Over 20 bottles

Frequency of alcohol-induced pass outs was asked as ‘Have you “passed out” while drinking alcohol during the past year?’

  1. Never

  2. Once

  3. Two to three times

  4. Four to six times

  5. Seven times or more

Social capital

Here we give some leisure-time activities. Indicate how often during the past years have you spent your free-time as given in the list.

Visiting relatives, friends, or acquaintances.

Participating spiritual, religious events.

Organizational, voluntary, civil, or other societal participation.

Visiting cultural or entertainment events such as movies, concerts, theatre, art exhibitions, or similar.

Visiting restaurants, discos, or pubs.

The frequency options for all of the above were:

  1. Weekly or more often

  2. One to three times a month

  3. Once or couple of times a year

  4. Never

Childhood

When you think about your childhood:

Did your parents separate (divorce or similar)?

Did someone of your family members have problems because of alcohol?

The response options for these questions were:

  1. No

  2. Yes

  3. I don't know/I can't say

Comments

0 Comments