Janne Härkönen

Age, period and cohort analysis of light and binge drinking in Finland, 1968-2008

AIMS To analyse the effects of age, period and cohort (APC) on light and binge drinking in the general population of Finland over the past 40 years. METHODS All analyses were based on six Drinking Habits Surveys between 1968 and 2008 of representative samples of the Finnish population aged between 15 and 69 (n = 16,400). The number of drinking occasions per year involving 1-2 drinks (light) and 4+ or 6+ drinks (binges) was used as a dependent variable in APC modelling. Descriptive cohort profiles and negative binomial models were used to assess the effects of APC. RESULTS Descriptive cohort profiles differed for light and binge drinking. No substantial differences were found across cohort profiles for light drinking, while APC modelling predicted declining cohort and increasing period effects. Differences between cohorts were found for binge drinking, with predictions of slightly declining or increasing period and increasing cohort effects. CONCLUSIONS Light drinking has increased over time for each cohort, with no substantial differences between cohort profiles. Binge drinking has increased with more recent cohorts and there are distinct differences between cohort profiles, especially among women.

Changes in the norms guiding alcohol use in the Finnish general population over the past 40 years

Attitudinal climate can be seen as a central factor influencing drinking habits; the level of consumption has been claimed to depend on how many adverse consequences of drinking a society tolerates. In the past few decades, however, only a few studies have looked at normative aspects of drinking in the general population. This article investigates changes in attitudes and norms around drinking in Finland by reviewing existing research and presenting new results. We ask whether men and womens views on alcohol differ and are there differences between age groups. The focus is on identifying long-term changes in the attitudinal climate of alcohol use. The main data is based on six national drinking habits surveys of 15–69-year olds conducted in 1968, 1976, 1984, 1992, 2000 and 2008. Data from the Scandinavian drinking survey in 1979 and the Finnish alcohol panel survey in 2003 are also used. Since the 1960s, attitudes towards ones own use of alcohol have become progressively more permissive and gender differences have converged. Results on situational norms showed that the acceptance of drinking increased in a leisure context, while there were no substantial changes in everyday and work situations. However, attitudes around alcohol policy varied through the studied time frame. In the 1970s, new liberalizations in alcohol policy induced great concerns, which settled down through the 1980s until the early 1990s. Since the end of 1990s, greater calls have been made for stricter alcohol policies, even though views on personal use of alcohol have become more permissive than ever.

Changes in mortality due to major alcohol‐related diseases in four Nordic countries, France and Germany between 1980 and 2009: a comparative age–period–cohort analysis

AIMS To investigate age, period and cohort effects on time trends of alcohol-related mortality in countries with different drinking habits and alcohol policies. DESIGN AND SETTING Age-period-cohort (APC) analyses on alcohol-related mortality were conducted in Denmark, Finland, Norway, Sweden, France and Germany. PARTICIPANTS Cases included alcohol-related deaths in the age range 20-84 years between 1980 and 2009. MEASUREMENTS Mortality data were taken from national causes of death registries and covered the ICD codes alcoholic psychosis, alcohol use disorders, alcoholic liver disease and toxic effect of alcohol. FINDINGS In all countries changes across age, period and cohort were found to be significant for both genders [effect value with confidence interval (CI) shown in Supporting information, Table S1]. Period effects pointed to an increase in alcohol-related mortality in Denmark, Finland and Germany and a slightly decreasing trend in Sweden, while in Norway an inverse U-shaped curve and in France a U-shaped curve was found. Compared with the cohorts born before 1960, the risk of alcohol-related mortality declined substantially in cohorts born in the 1960s and later. Pairwise between-country comparisons revealed more statistically significant differences for period (P < 0.001 for all 15 comparisons by gender) than for age [P < 0.001 in seven (men) and four (women) of 15 comparisons] or cohort [P < 0.01 in two (men) and three (women) of 15 comparisons]. CONCLUSIONS Strong period effects suggest that temporal changes in alcohol-related mortality in Denmark, Finland, Norway, Sweden, France and Germany between 1980 and 2009 were related to secular differences affecting the whole population and that these effects differed across countries.

Changes in mortality due to major alcohol-related diseases in four Nordic countries, France and Germany between 1980 and 2009: a comparative age-period-cohort analysis

AIMS To investigate age, period and cohort effects on time trends of alcohol-related mortality in countries with different drinking habits and alcohol policies. DESIGN AND SETTING Age-period-cohort (APC) analyses on alcohol-related mortality were conducted in Denmark, Finland, Norway, Sweden, France and Germany. PARTICIPANTS Cases included alcohol-related deaths in the age range 20-84 years between 1980 and 2009. MEASUREMENTS Mortality data were taken from national causes of death registries and covered the ICD codes alcoholic psychosis, alcohol use disorders, alcoholic liver disease and toxic effect of alcohol. FINDINGS In all countries changes across age, period and cohort were found to be significant for both genders [effect value with confidence interval (CI) shown in Supporting information, Table S1]. Period effects pointed to an increase in alcohol-related mortality in Denmark, Finland and Germany and a slightly decreasing trend in Sweden, while in Norway an inverse U-shaped curve and in France a U-shaped curve was found. Compared with the cohorts born before 1960, the risk of alcohol-related mortality declined substantially in cohorts born in the 1960s and later. Pairwise between-country comparisons revealed more statistically significant differences for period (P < 0.001 for all 15 comparisons by gender) than for age [P < 0.001 in seven (men) and four (women) of 15 comparisons] or cohort [P < 0.01 in two (men) and three (women) of 15 comparisons]. CONCLUSIONS Strong period effects suggest that temporal changes in alcohol-related mortality in Denmark, Finland, Norway, Sweden, France and Germany between 1980 and 2009 were related to secular differences affecting the whole population and that these effects differed across countries.

Changes in Finnish drinking occasions between 1976 and 2008 – The waxing and waning of drinking contexts

A deeper understanding of drinking practices of a population requires a study of the situations in which people drink, i.e. the context of drinking. This study explores the changes and continuities in the prevalence and nature of drinking in terms of place, company, time and the social drinking context, during a period when overall alcohol consumption in Finland grew by half and drinking culture at large was transformed radically. Analyses were based on two national drinking habits surveys with a representative sample of the Finnish population aged 15–69 years in 1976 (N = 2835) and 2008 (N = 2725). In addition, original semi-qualitative data in 2008 were utilized, using a 15-category social drinking context typology, with the results contrasted to previously reported results from 1976. Overall, the number of drinking occasions increased between 1976 and 2008. A major part of the increase comprised drinking in home settings and with ones partner. The weekly rhythm of drinking was also concentrated on weekends even more than before. Changes in the nature of drinking showed that the typical degree of intoxication decreased for men and increased for women. The proportion of heavy drinking occasions decreased for men especially in home settings and for women, remained the same across drinking contexts except for single gender contexts. Changes in social drinking contexts reflected an overall shifting of drinking into the private sphere, with a major increase in the proportion of evenings at home and sauna drinking.

Decreasing adolescent drinking: Is there evidence of a continuation into future adult cohorts? APC analysis of adolescent drinking in Finland, 1983–2013:

Aims: Unlike adults, abstaining has increased and regular use of alcohol has decreased among 12–16-year-olds over the past two decades. The paper studies whether these developments will be continued as the adolescent cohorts come of age. Methods: The Adolescent Health and Lifestyle Survey is a nationally representative monitoring system of the health habits of 12-, 14-, 16-, and 18-year-old Finns, conducted biannually between 1981 and 2013. The prevalence of alcohol use and drunkenness were measured for each 5-year cohort born in 1967–1995. Age-by-cohort trajectories and hierarchical age–period–cohort (APC) modeling were used to assess effects of age, period, and birth cohort. Results: Cohorts differentiate for underage drinking, but not at the age of 18. The younger cohorts postpone their drinking debut compared with older cohorts and thus age profiles are steeper than before. The most recent cohorts born in the 1990s, and the oldest cohorts born in 1967–71, have the highest prevalence in abstinence but drinking has been more prevalent for cohorts born in 1973–1989. APC modeling confirms significant cohort effects, but no significant decrease in drinking or drunkenness at the age of 18 years. Some of the changes can also be attributed to period effects. Conclusions: Despite the decrease in underage drinking in Finland, 18-year-olds continue to drink similarly from cohort to another. Postponing the onset of drinking has a preventive effect on alcohol-related harms, but a reduction in drinking among adult cohorts is not evident in the future.

Predictors of Persistence of Risky Drinking in Adults: An 11-Year Follow-Up Study

Aim: To examine the long-term predictors of persistence of risky drinking in a baseline group of risky drinkers in whom alcohol use disorder had not been diagnosed. Methods: The data was derived from a representative sample of the Finnish adult population aged 30 years or more, surveyed at 2 time points in the years 2000 (n = 5,726) and 2011 (n = 3,848, 67.2% of the baseline sample). Risky drinking was defined using BSQF-measurement (for men, 21 standard UK drinks or more per week; for women 14+ drinks) and not having alcohol abuse or alcohol dependence. The sample of risky drinkers in baseline comprised 642 persons, of whom 380 (59.2%) people provided follow-up data. Multivariable logistic regression models were estimated to identify determinants of persistence of risky drinking. Results: The rate for persistence of risky drinking was 48.7%. Persistence was predicted by daily smoking, low physical activity, and male gender, whereas higher age and later onset of drinking predicted cessation of risky drinking. Daily smoking remained an independent predictor after adjusting for other risk factors. Conclusions: Health behaviour predicts the persistence of risky drinking in a study population of adults aged 30 and over. These factors should be taken into account when assessing the long-term prognosis on risky drinking.

Lessons learned from Sweden

Acknowledgement This work has received support from the Academy of Finland (project no. 137685). Researchers and policymakers have been intrigued and puzzled by the developments and trends in alcohol consumption and related harms in Sweden. Researchers have tried to discern the underlying “true” picture, striving to pinpoint mechanisms and explanations of the seeming “mismatch” between consumption and harm trends, but this has not been easy given the multitude of estimates that contain varying types and degrees of uncertainty and error in addition to real differences in the different dimensions of drinking (such as per capita consumption vs. binge drinking) and related harm. This has been further complicated by differences between regions and population subgroups. Kalle Tryggvesson (2013) has taken up the challenge of contributing to this undertaking by looking at fresh high-quality survey data and by trying to illuminate the development in drinking patterns and the distribution of drinking in the puzzle. In our comments, we do not seek to be judges of the “true answer” to the developments in Sweden, but will instead concentrate on two issues. First, we will look at Tryggvesson’s findings as general results about the development of drinking patterns in Sweden and compare the developments to those that have been observed for Finland – quite apart from the “mismatch question”. Second, we will discuss some challenges in answering the questions that Tryggvesson has posed and will consider whether the Swedish case has changed the way we think about how developments in total consumption predict alcohol-related harm.