This study suggests that alcohol drinking patterns among adults in Norway have changed in several ways over the past 2–3 decades. Abstaining has become rarer while consumption, recent drinking and lifetime problematic drinking have increased. Most changes were in the same direction for men and women, but women have moved towards men’s drinking patterns in abstaining, recent drinking, problematic drinking and in mean volume of consumption (liters a year). Among recent drinkers, intoxication has decreased in both genders, but more in men than in women. The declines we noted in gender differences, however, did not occur for all age x gender groups nor for all alcoholic beverages: they were age-specific and varied depending on which drinking behavior and which beverage was observed. For example, in the youngest age group, intoxication became less frequent in men and more frequent in women, indicating gender convergence. Some gender convergence in alcohol consumption was due to the greater decline in men’s than in women’s mean consumption of liquor.
Changing drinking patterns
The overall changes in drinking patterns in this study are very much in accordance with previous Norwegian studies [
3,
4,
38]. A range of secular changes in society may help explain the changing drinking patterns, and some of the most important conditions may be the improved family economy and the ongoing globalisation [
26,
39]. Norway has historically kept a strict alcohol policy with high prices and restricted availability [
40]. Although prices have not decreased in real terms, the increase in general spending power has made alcohol “less expensive” for most Norwegians. Simultaneously the access to “cheaper” alcoholic beverages through cross-border and international tax-free shopping has expanded dramatically [
41]. The number of alcohol outlets have increased and the time that pubs and other places that serve alcoholic beverages are open have expanded rather than declined [
42].
Extended international travelling has not only made alcohol more available, but has apparently also influenced the drinking culture. In this study most of the consumption of liquors seems to have been “replaced” by wine drinking that increased from 25 % to almost half of all consumption (42 %) within the same period. Correspondingly, national sales statistics show that wine consumption increased 18-fold from 1960 to 2005 and made up 37% of all alcohol intake in 2012, reflecting the change in people’s beverage preferences [
41]. Despite the fact that alcohol consumption has increased, all changes in this study should not be considered disadvantageous. Norway has previously been given a high harm score associated with alcohol drinking [
2] and any changes towards more moderate use should be considered beneficial. The observed decline in drinking to intoxication may at least indicate a trend toward less heavy episodic drinking.
Changing gender differences in alcohol drinking
Our results suggest that gender differences have decreased for many alcohol measures, but that the size of the decrease varied depending on age group. This is in accordance with most previous studies of gender differences, but Norwegian men and women may constitute an exception in some aspects of alcohol drinking. Results from the cross-cultural and multinational GENACIS study showed that Norwegian gender ratios differed from most of the other countries included, although some differences were restricted to the youngest age group (18–34 years) [
10]. The GENACIS analyses found that Norwegian women were as likely as men to be current drinkers (in all age groups) and in the youngest age group there were no gender differences in high-frequency drinking.
In studies of changing drinking patterns, there is an ongoing debate whether decreasing gender differences are due to age, period or cohort effects. Although these factors are overlapping and time effects always can be interpreted as a combination of cohort and age effects, birth cohort effects may reflect historical influences and experiences that have imprinted the life-course attitudes and behaviors of individuals born at particular times. We have not investigated possible birth cohort effects in this study, but the idea that gender convergence is most evident in younger cohorts [
6], is partly consistent with our finding a decline in gender differences and a narrowing of the gender gap in the youngest age group (20–29 years). On the other hand, our findings also suggest a more consistent decline in most gender differences in most age groups of adults and for most alcohol measures during the past 2–3 decades. A somewhat similar consistency among adults was reported in a study from New Zealand between 1995 and 2000, but in that study all changes in gender differences (in adults 20+) were explained by increases in women’s drinking [
22].
Some research has hypothesized that gender differences in drinking may become smaller as women’s rights and social status improve relative to men’s. For example, Rahav et al., [
43] found that in countries with lower levels of gender inequality and greater gender empowerment, gender differences in current drinking were smaller. However, these data were cross-sectional and so did not measure how changes in gender inequality might be related to changes in gender ratios of drinking. Gender equality is not synonymous with increased alcohol-related harm among women [
44], but associations between women’s social status and alcohol drinking may be complicated and may be very different in higher-income versus lower-income countries [
45]. They may also differ between women at high and low economic levels within a country [
7]. Bergmark (2004) suggested on basis of Swedish data that social background factors may play a lesser role for gender differences in alcohol use today than during the 1970s [
16]. This may be true for Sweden and other Nordic countries, given the combination of high levels of gender equality and social welfare, but not necessarily for other countries.
Only a few studies of gender convergence have focused on men’s drinking, but there is some evidence of changes towards more moderate drinking in men. For example, Neve et al. reported gender convergence that was explained by a decline in more highly educated men’s consumption [
23]. Saelan et al. followed a Danish birth cohort and found decreasing consumption in men and the reverse in women [
46]. In a study by Bloomfield et al., there was a tendency of men in both Germany and Switzerland (not in Finland) to have decreased more than women in rates of hazardous drinking between 1984 and 1992 [
17]. These findings towards more moderate drinking in men, have to our knowledge not attracted much attention. This may reflect that, at least in part, more attention has focused on changes in women’s alcohol use than in men’s. The one-sided and negative focus on increases in women’s drinking has been criticized as a social construct that serves to blame women [
47]. More important perhaps, a hypothesis based solely on changes in women’s drinking pays no attention to ways that gender convergence in drinking may occur because of changes in men’s drinking, or changes in both genders.
Although most changes in our study were in the same direction for men and women there were some desirable changes in men’s drinking that should be emphasized. The halving of intoxication in men (among recent drinkers) should be noted, since male intoxication has consistently been associated with substantial alcohol-related harm [
9,
10]. Men’s changes in beverage preferences from liquors to beer and wine consumption may also be considered a positive change, in terms of shifting to beverages with lower alcohol concentration.
Another “favorable finding” was observed in women aged 30–39 years, which is the period most (Norwegian) women give birth to and bring up children. Between HUNT2 (1995–97) and HUNT3 (2006–08) the average consumption in this age group was stable, abstaining increased, while drinking to intoxication became more rare. Problematic drinking increased, but seemingly to a lesser extent than in other age groups. Future research should investigate to what extent the culture of childbearing and childrearing in Norway may have changed to discourage alcohol consumption among women at this stage of life.
Despite the possible trend towards more moderate use of alcohol (decline in intoxication) in this Norwegian population, rates of lifetime problematic drinking increased in both genders, slightly more in women than in men. While the incidence of developing alcohol problems or dependence during one’s lifetime in men seems to be stable in many countries, a recent longitudinal Swedish study suggested that the incidence among women increased between 1972 and 1997 [
48]. Another recent Icelandic study reported gender convergence in discharge diagnosis of alcohol use disorders among psychiatric patients [
24]. These changes may be of great concern and opposite to what the World Health Organization has appealed for: a reduction in the social, medical and economic costs of excessive alcohol drinking [
49].
Strengths and limitations
One advantage of the HUNT surveys is that the county surveyed, situated in the middle part of Norway has a very stable and homogenous population, with little migration and few cultural disparities related to religion and ethnicity. The repeated cross-sectional surveys of the general population located within the same geographic area, strengthen the likelihood of reliable estimates of change and comparisons across age groups and genders. The demographic stability of the county also lends strength to the idea that changes observed in drinking behaviour were cultural, and not the result of demographically different persons moving into and out of the county. The Nord-Trøndelag population has been shown to be very similar to the general population of Norway in many health aspects [
31,
33]
https://www.khs.fhi.no/en/health-in-the-municipalities/. Since this county lacks major cities, estimates of change in alcohol consumption may not correspond to the changes in the most urban parts of the country, but they are very much in accordance with changes reported in previous studies of adults [
4,
38,
41], based on national representative samples between 1973–2004.
As in general population surveys elsewhere in the world [
50] the participation rate in HUNT decreased considerably since the 1980s, especially in younger (20–39 years) and older age groups (80+) and more in men than in women. In HUNT3, women constituted 55 % of the study population. Participation rates among middle-aged men and women (50–79 years) have been stable and of less concern. In HUNT3 68 % of eligible men and 74 % of eligible women aged 60–69 years participated in the study [
33].
Of more concern in studies of change, perhaps, is the increasing gender imbalance in attrition and whether e.g. male responders in HUNT have become less representative of the general population. According to a previous non-response study of HUNT1 and HUNT2, attrition was moderately associated with both abstaining and heavy drinking, but was not considered a major cause of nonresponse after taking other characteristics into account [
51]. Although the gender imbalance in response rates may explain some of the observed decrease in gender differences, it is important to note that gender ratios of intoxication and mean annual consumption were based on recent drinkers and not total samples. Changing gender ratios of these two measures are therefore not a consequence of more women drinking or fewer women abstaining.
Due to the substantial decline in participation rate in HUNT3, a thorough nonparticipation study was conducted [
33], including questionnaire data from 6922 HUNT non-responders, National Registries data and data from General Practitioners in the county. Non- responders were (among other factors) characterized by lower socioeconomic status, higher mortality and higher prevalence of chronic disorders. The study gives no direct answer to the question of why more women than men responded in HUNT3, but among those younger than 40 years of age, the main reasons for non-response in both genders were “had no time/ inconvenient” and “got no invitation”. Of more importance however, there were no differences in alcohol drinking (i.e., drinking 2–3 times a week or more often) or in daily smoking between responding and non-responding men and women aged 20–39 years. Lifestyle factors accounted only for a small fraction of the observed underestimation of a range of outcomes (e.g., morbidity and mortality) after taking social status (SES) into account. In this study, educational level was used as proxy of SES and possible confounding variable.
On the other hand, if gender imbalance in attrition has to do with gender specific changes in educational level, the survey samples of men and women may have become less representative for their source populations, and samples less comparable over time. According to available national statistics of men and women aged 30–39 years
http://www.norgeshelsa.no), higher education (high school or university) was slightly more common among female responders in HUNT responders, than in this region (3 counties) and in the country, while the corresponding proportion in HUNT male responders was lower than in the country, but similar to that in the region. More important perhaps, these differences have not changed over time and thus not suggested to represent a selection bias. Although the increasing attrition, in younger men in particular, may have reduced the precision of estimates in this study, attrition is not considered seriously to limit the findings in this study.
There are other limitations to acknowledge. First, some drinking behaviours were not assessed at all survey time points, which limits the opportunity to draw conclusions about historical trends of change. Second, variation in how questions were asked in the three surveys also makes it necessary to be cautious when interpreting the results.
The validity of the CAGE as a measure of problem drinking has been debated, but our findings on problematic drinking (or alcohol misuse) in HUNT are consistent with previous findings of alcohol abuse based on other diagnostic tools (DSM-IIIR and CIDI) [
52]. Skogen et al. (2011) investigated the concurrent validity and psychometric properties of the CAGE by using data from HUNT1 and HUNT2 and concluded that the internal reliability of the CAGE was adequate [
53]. Findings also suggested a better concurrent validity in women than in men, which stands in contrast to Dhalla and Kopec (2007), who found that the CAGE did not perform well among white women in a primary care setting [
37].