|Home | About | Journals | Submit | Contact Us | Français|
Alcohol consumption has demonstrated substantial temporal trends, with some evidence suggesting strong birth cohort effects. The identification of at-risk birth cohorts can inform the interpretation of alcohol trends across age, time, and demographic characteristics such as gender. The present literature review has two objectives. First, we conduct a cross-national review of the literature on birth cohort differences in alcohol consumption, disorder, and mortality. Second, we determine the consistency of evidence for birth cohort effects on gender differences.
A search was conducted and key data on population characteristics, presence and direction of cohort effects, and interactions with gender compiled. Thirty-one articles were included.
Evidence suggests that younger birth cohorts in North America, especially those born after World War II, are more likely than older cohorts to engage in heavy episodic drinking and develop alcohol disorders, but this cohort effect is not found in Australia and Western Europe. Cross-nationally, substantial evidence indicates that women in younger cohorts are at especially high risk for heavy episodic drinking and alcohol disorders.
Younger birth cohorts in North America and Europe are engaging in more episodic and problem drinking. The gender gap in alcohol problems is narrowing in many countries, suggesting shifting social norms surrounding gender and alcohol consumption. These trends suggest that public health efforts to specifically target heavy drinking in women are necessary.
The World Health Organization has called for the development of global strategies to reduce alcohol consumption (Casswell and Thamarangsi 2009), noting the high social, medical, and economic costs of excessive alcohol consumption worldwide (Rehm, Mathers et al. 2009). A greater understanding of temporal and geographic variation in key indicators of alcohol consumption (e.g., incidence of heavy episodic drinking, development of alcohol disorders, and alcohol-related mortality) is critical to achieving such a reduction. The assessment of birth cohort effects in these indicators can serve three important purposes. First, the identification of birth cohorts at high risk for alcohol-related problems can inform research on etiology by suggesting exposures specific to particular birth cohorts that can be investigated. Second, cohort effects are important to identify for methodological reasons; if time trends are attributable to birth cohort effects, then alcohol trend data should be aggregated and presented by cohort rather than the traditional aggregation by time period. Finally, identification of increased alcohol consumption and related problems in more recently born birth cohorts is important for public health planning and resource allocation.
Concerning etiologic research, birth cohort effects suggest unique social, political and environmental factors to which members of each birth cohort are exposed as they progress through the life course impact health in potentially important ways (Keyes, Utz et al. 2010). Skog and others have written about the “collectivity” of drinking cultures that is both geographically and temporally defined (Skog 1985; Room and Makela 2000; Ahern, Galea et al. 2008), creating the emergence of group-level variation in alcohol use that is distinct from individual-level variation in risk factors for alcohol use. Individuals within certain birth cohorts come of age during specific socio-political moments, sharing experiences that impact health and well-being at critical periods of the life course. Economic fluctuation, political instability, policies and laws, and social norms are group-level exposures that can vary between time periods and countries, potentially impacting particular birth cohorts in ways that affect their risk for problematic alcohol consumption and disorder. Identifying group-level exposures that impact the alcohol outcomes of certain birth cohorts may thus have substantial public health relevance for future prevention efforts.
Cohort effects are also important to identify for methodological reasons, as the presence of these effects obscures the interpretation of age-related trends in alcohol and drug outcomes. For example, cross-sectional studies consistently find that the prevalence of current alcohol consumption and disorders is lower among older individuals compared to younger individuals (Grant 1997; Johnson, Gruenewald et al. 1998; Hasin, Stinson et al. 2007). Prospective studies, however, show mixed results; while some studies demonstrate drinking reductions with age (Adams, Garry et al. 1990; Fillmore, Hartka et al. 1991; Caetano and Kaskutas 1995; Eigenbrodt, Mosley et al. 2001), many others show no change or even increases (Gordon and Kannel 1983; Klatsky, Siegelaub et al. 1983; Glynn, Bouchard et al. 1985; Hubert, Eaker et al. 1987; Neve, Diederiks et al. 1993; Levenson, Aldwin et al. 1998; Wilsnack, Wilsnack et al. 2009). The discrepancy between cross-sectional and prospective findings could be a result of birth cohort effects. Individuals in cross-sectional studies represent many different birth cohorts; thus, age effects are difficult to separate from cohort effects. Conversely, individuals studied prospectively typically include only one or a few birth cohorts of individuals, making cohort effects difficult to detect because of a lack of adequate data across multiple birth cohorts. Examining findings from datasets with various sampling strategies and the estimation of formal age-period-cohort models may shed light on these seemingly discrepant results.
Differences in the effect of birth cohort on demographic risk factors may necessitate the replacement of old assumptions with new information. An important example concerns gender and alcohol. Greater risk in males is well-documented across countries and cultures (Brady and Randall 1999; Nolen-Hoeksema 2004; Nolen-Hoeksema and Hilt 2006; WHO 2011). However, recent evidence suggests a diminishing gender gap in the prevalence of alcohol use and disorders among more recently born cohorts (Grucza, Norberg et al. 2008; Keyes, Grant et al. 2008; Seedat, Scott et al. 2009; Wilsnack, Wilsnack et al. 2009). A cohort-specific increase in female drinking would signal the need for increased prevention and intervention efforts to address drinking problems in the birth cohorts identified at heightened risk. Further, if gender differences are truly diminishing, the paradigm of alcohol problems being a male-dominated health outcome will need to be shifted. Critical, careful review of the consistency of these findings across study designs and samples is therefore crucial, as the public health implications of such a major change are substantial.
To date, there have been no systematic reviews of information available regarding cohort effects on alcohol indicators across time, geographic location, and gender. This report aims to review and synthesize the existing population-based information on cohort effects for three alcohol consumption indicators: 1) lifetime prevalence of alcohol consumption; 2) current prevalence of alcohol consumption; and 3) lifetime prevalence of alcohol disorders, problems, and alcohol-related mortality. These outcomes were chosen to reflect key indicators of alcohol use across the life-course: initial use as represented by lifetime any alcohol consumption, regular patterns of alcohol consumption in terms of quantity and frequency, and alcohol-related problems such as alcohol use disorders and alcohol-attributable mortality. The review addresses three questions with respect to each of these outcomes: a) Is there evidence for a cohort effect, with risk higher in more recently born birth cohorts? b) Does the cohort effect vary cross-nationally? and c) Is there evidence for differences in the cohort effect across gender?
Materials for this review were primarily identified through searches of three electronic databases for peer-reviewed published papers: the MEDLINE/Pubmed, PsychInfo, and Scopus. Search terms were selected based on initial review of relevant subject headings across databases that were likely to yield relevant results. Using MEDLINE/Pubmed, initial search terms included: “age-period-cohort” or “birth cohorts” and “alcohol” (N=47), “liver” (N=47), “cirrhosis” (N=21), “substance use” (N=10). We then added a search term for “gender” (N=2) and “women” (N=18). Using PsychInfo, initial search terms included: “cohort analysis”, “time trends” and “alcohol drinking patterns” (N=27), “liver” (N=26), “cirrhosis” (N=24), and “substance abuse” (N=15), and a search terms for “gender” (N=27) and “women” (N=13). Using Scopus, initial search terms included: “age-period-cohort” and “alcohol” (N=20) “liver” (N=18) “cirrhosis” (N=6), “substance abuse” (N=1) and additional search terms for “gender” (N=1) and “women” (N=1). A search of related articles in these databases, references in relevant papers, and consultation with recognized experts yielded additional 8 papers for review.
Figure 1 details the process of article refinement for inclusion in the review. After duplicate articles were removed, an initial set of 151 papers was reviewed. Inclusion criteria were as follows: (1) published in the last 20 years (i.e., after 1988, in order to focus the review on the most recent information regarding birth cohort effects); (2) written in English; (3) peer-reviewed; (4) original research (i.e., no review articles); (5) population- or community-based sample; (5) directly assessed possible cohort effects by comparing at least two birth cohorts; (6) included alcohol use, alcohol use disorder, or alcohol-related mortality as an outcome.
The final number of papers included was 31. Four articles included more than one outcome that was relevant to the present review (Grant 1997; Levenson, Aldwin et al. 1998; Keyes, Grant et al. 2008; Zhang, Guo et al. 2008); these outcomes were considered separately, for a total of 31 papers and 35 outcomes assessed.
Lifetime prevalence of alcohol use was defined as either the lifetime proportion of alcohol use (percentage of population who have ever used alcohol) or as the incidence density of alcohol use (percentage of new alcohol users per unit of person-time at risk of alcohol).
Prevalence of current alcohol use was defined as the proportion of alcohol users in a specified time frame (e.g., past week, month, or year).
Alcohol disorders/problems included any prevalent DSM (substance abuse or dependence) or ICD (harmful substance use or substance dependence) defined alcohol disorder, chart review for alcohol problems, or scales capturing alcohol-related impairment in social, occupational, or physical domains.
Alcohol-attributable morbidity/mortality was included as an outcome for papers that examined morbidity and mortality that is solely caused by alcohol (e.g., ICD-10 codes F10 [mental and behavioral disorders due to use of alcohol], X45 [accidental poisoning by and exposure to alcohol], and Y15 [poisoning by alcohol, undetermined intent]), or for which alcohol is a primary contributing case (e.g., ICD-10 code K70 [alcoholic liver disease]).
We separated the studies into three quality levels of evidence, hierarchically organized by the rigor of the statistical modeling and the appropriateness of the dataset to answer age-period-cohort research questions. For convenience, we labeled these as Levels 1, 2 and 3.
Studies given a (1) provided the most rigorous evidence. These include cross-sectional studies repeated over time as well as panel designs (longitudinal follow-ups of sequential cohorts), and formal statistical modeling to decompose variance in trends over time into components that can be attributed to age, period, and cohort. We note that modeling of APC effects remain an ongoing statistical debate (Keyes, Utz et al. 2010), full elaboration of which is outside the scope of this review. All articles included in this review utilize common and well-documented methods for APC analysis.
Those studies given a (2) provided an intermediate level of evidence in terms of methodological rigor. These include longitudinal cohort studies with a limited number of birth cohorts (as opposed to panel designs with a greater number of birth cohorts), with and without statistical age-period-cohort models. While longitudinal studies offer more rigorous evidence for cohort effects than single time point cross-sectional studies, period and age effects are difficult to unconfound with cohort effects in the longitudinal design due to the often restricted variation in birth cohorts, and the fact that the cohorts age simultaneously with time so age and period effects cannot be disentangled.
Those studies given a (3) are those providing the least rigorous evidence, and for which the most caution should therefore be taken during synthesis and inference. Cross-sectional studies make up the majority of this category, since they offer the advantage of readily-available data. However, one-time cross-sectional studies rely heavily on retrospectively reported information, raising methodological issues in the validity of retrospective recall. Further, age and cohort effects cannot be rigorously disentangled in the cross-sectional design, as respondents’ current age at the time of the survey determines birth cohort. However, cross-sectional studies often control for a collection of important confounders, thus, we include these types of APC analyses in this review, but use caution in interpreting the results. We included only those cross-sectional studies that report on birth cohorts, rather than cross-sectional studies that report differences across age alone. While age can be used as a proxy for cohort in the cross-sectional design, we focused on studies that explicitly aimed to estimate parameters for birth cohorts.
Seven studies examined cohort effects in the lifetime prevalence of alcohol use (Grant 1997; Johnson and Gerstein 1998; Degenhardt, Lynskey et al. 2000; Johnson and Gerstein 2000; Degenhardt, Chiu et al. 2007; Keyes, Grant et al. 2008; van Heerden, Grimsrud et al. 2009). Only one study of the lifetime prevalence of alcohol consumption had a level (1) (i.e., most rigorous) methodology to examine evidence for cohort effects in the lifetime prevalence of alcohol use (Johnson and Gerstein 2000). This study did not find evidence for cohort effects in the incidence density of alcohol initiation among repeated assessments of young adults aged 10–24 in the U.S. Measures of incidence and population at risk were reconstructed through retrospective reports. Restricted age range of this study could account for the lack of a demonstrated cohort effect for alcohol initiation, although cohort effects were detected for other outcomes (i.e., marijuana) in the same sample.
The remaining six studies (four based in the U.S., one in Australia and one in South Africa) were level three (i.e., cross sectional) studies, using questions on whether the respondent has ever consumed alcohol as a measure of lifetime prevalence. Of these six, Grant (Grant 1997), Keyes et al. (Keyes, Grant et al. 2008), Johnson & Gerstein (Johnson and Gerstein 1998), van Heerden et al. (van Heerden, Grimsrud et al. 2009) and Degenhardt et al. 2007 (Degenhardt, Chiu et al. 2007) report that the cumulative proportion of lifetime alcohol use has increased in younger cohorts in the U.S., especially those born after World War II.
Five of the seven studies examining lifetime prevalence of alcohol use were conducted in the U.S., and of these, four reported a cohort effect with the highest risk in the latest born cohort (Grant 1997; Johnson and Gerstein 1998; Degenhardt, Chiu et al. 2007; Keyes, Grant et al. 2008). Van Heerden et al. (2009) reported a similar effect in South Africa. In contrast, Degenhardt et al. 2000 (Degenhardt, Lynskey et al. 2000) reports no similar cohort effect in Australia. Explanation for the discrepancy can be drawn from different drinking patterns between countries; 27.2% of individuals in the oldest cohort of a U.S. sample (born 1894–1937) were lifetime abstainers (Grant 1997), while less than 3% of individuals in approximately the same birth cohort (born 1904–1932) were lifetime abstainers in the Australian sample (Degenhardt, Lynskey et al. 2000).
The four studies that assessed lifetime alcohol use by gender all reported converging rates across successively more recent birth cohorts, including the level (1) methodology study (Johnson and Gerstein 2000); male to female ratios approached 1.0 in the most recent cohorts in Johnson & Gerstein (1998), 1.2 in Grant (Grant 1997), and 1.0 in Johnson & Gerstein (2000). A significant interaction between cohort and gender was reported in Keyes et al. (Keyes, Grant et al. 2008), with the gender ratio in lifetime initiation of alcohol decreasing from 1.94:1 to 1.54:1.
Table 1 shows selected information on the 15 studies that assessed the prevalence of alcohol consumption. Five studies were level (1) methodology, indicating the most rigorous evidence for cohort effects. All five reported evidence of a cohort effect, and three studies report reported that youngest cohort was at the highest risk for heavy episodic drinking (Kemm 2003; Kerr, Greenfield et al. 2009) or high household alcohol expenditure (Aristei, Perali et al. 2008). A fourth study reported that that youngest cohort consumed the highest quantity of alcohol, although this effect was restricted to women (Bjork, Thygesen et al. 2008). Kerr et al. (Kerr, Greenfield et al. 2004) is the only study to examine beverage specificity; this study reported birth cohort variation in type of beverage consumed, with spirit consumption highest in older born U.S. cohorts (those born before 1940). Beer and wine consumption was highest in cohorts born 1965–1976 (men) and pre-1940 (women), respectively.
Among the four studies at an intermediate level of rigor (level (2)), three studies reported that a cohort effect was observed, but the youngest cohort was not at the highest risk for the outcome (Menard and Huizinga 1989; Moore, Gould et al. 2005; Zhang, Guo et al. 2008); cohorts found to be at high risk were primarily those born from around 1900–1920. One study reported no cohort effect (Karlamangla, Zhou et al. 2006).
Finally, among those studies with the lowest level of rigor in assessing cohort effects (level (3)), five studies found evidence of a cohort effect (Black and Markides 1994; Johnstone, Leino et al. 1996; Levenson, Aldwin et al. 1998; Bachman, Freedman-Doan et al. 1999; Gilhooly 2005). Studies varied in the identified cohort with the highest alcohol consumption, but there was a general tendency toward younger cohorts consuming less mean alcohol than older cohorts per time period studied.
Nine of the fifteen studies examining current prevalence of alcohol use were conducted in the U.S., and of these, eight reported a cohort effect. Only two studies found that risk was highest in the latest-born cohort (Black and Markides 1994; Kerr, Greenfield et al. 2009), and one of these two only found highest risk in the latest-born cohort among Cuban Americans (Black and Markides 1994).
Studies in Great Britian (Kemm 2003), Scotland (Gilhooly 2005), Italy (Aristei, Perali et al. 2008), and Denmark (Bjork, Thygesen et al. 2008) have all documented cohort effects in alcohol consumption; the study in Italy found risk was highest in the latest born cohorts, and the studies in Denmark and Scotland found risk was highest in the latest born cohorts only in women. Two did not find evidence for a cohort effect; one in the Netherlands (Neve, Diederiks et al. 1993) and one cross-nationally (Johnstone, Leino et al. 1996).
Evidence for a narrowing of gender differences is mixed, although the strongest methodological studies all document convergence in gender differences in the U.S., Britain, and Denmark. Among level (1) studies, three studies document gender convergence (Kemm 2003; Bjork, Thygesen et al. 2008; Kerr, Greenfield et al. 2009); the remaining two studies did not assess gender differences. Among level (2) studies, two studies did not find evidence of a gender convergence (Moore, Gould et al. 2005; Karlamangla, Zhou et al. 2006), and the remaining two studies did not assess gender. Among level (3) studies, two studies found evidence of gender convergence (Neve, Diederiks et al. 1993; Gilhooly 2005), one did not (Johnstone, Leino et al. 1996), and the remainder did not assess gender differences.
Studies of alcohol disorder, problems, and mortality are reviewed in Table 2. Of the twelve studies identified with alcohol disorders/problems as an outcome, ten found evidence of a cohort effect (Burke, Burke et al. 1991; Grant 1997; Heath, Bucholz et al. 1997; Levenson, Aldwin et al. 1998; Holdcraft and Iacono 2002; Rice, Neuman et al. 2003; Grucza, Bucholz et al. 2008; Keyes, Grant et al. 2008; Zhang, Guo et al. 2008; Seedat, Scott et al. 2009), and nine found a step-wise increase in risk with each successively younger cohort (Burke, Burke et al. 1991; Grant 1997; Heath, Bucholz et al. 1997; Levenson, Aldwin et al. 1998; Holdcraft and Iacono 2002; Rice, Neuman et al. 2003; Grucza, Bucholz et al. 2008; Keyes, Grant et al. 2008; Zhang, Guo et al. 2008; Seedat, Scott et al. 2009). The studies that did not find cohort effects (Kendler, Prescott et al. 1997; Dube, Felitti et al. 2003) were level (2) and (3) studies.
Two level (1) studies examined age, period, and cohort effects in directly measured alcohol-associated mortality. Rosen & Haglund report that alcohol-attributable mortality is lower among more recently born cohorts in Sweden (Rosen and Haglund 2006). Corrao et al. (Corrao, Ferrari et al. 1997) reported substantial cross-region differences in the emergence of a cohort effect for liver cirrhosis mortality. Younger birth cohorts in Eastern and Northern European countries evidence higher risk of cirrhosis mortality, whereas younger birth cohorts in Southern and Western European countries are at slightly lower risk of cirrhosis mortality. Neither study found evidence for a converging gender gap.
Nine of the twelve studies assessing cohort effects in alcohol problems have been conducted in the U.S. All studies but one (Dube, Felitti et al. 2003) document a cohort effect, and all but two (Dube, Felitti et al. 2003; Keyes, Grant et al. 2008) document that risk is highest in the most recently born cohort. Cohort effects with the youngest cohort at risk are documented in an Australian sample (Heath, Bucholz et al. 1997) and across fifteen different countries as part of the WHO-WMH surveys (Seedat, Scott et al. 2009).
One level (2) study (Zhang, Guo et al. 2008) found no convergence in rates of gender differences in alcohol-related mortality. The remaining studies documenting gender convergence were level (3) studies (Grant 1997; Holdcraft and Iacono 2002; Rice, Neuman et al. 2003; Keyes, Grant et al. 2008; Seedat, Scott et al. 2009). We note that these samples were primarily drawn from the U.S. However, Seedat et al. (Seedat, Scott et al. 2009) reported data from 15 countries, documenting the greatest reduction in the gender gap among countries where women’s roles have evidenced the greatest shift from traditional to autonomous.
First, the most methodologically rigorous studies provide substantial (though not universal) evidence of a cohort effect in alcohol consumption, with younger birth cohorts consistently consuming more alcohol than older cohorts. Second, consistent evidence suggests increased risk of heavy episodic drinking and alcohol disorders in younger born cohorts compared to older cohorts. Third, studies are relatively consistent in demonstrating a narrowing of the gender gap for heavy drinking and alcohol disorders, whereby women in younger birth cohorts evidenced increasing prevalence of these outcomes over time while men did not. However, we note that most studies showing a converging gender gap are cross-sectional in design, indicating that caution should be taken when synthesizing this evidence due to the possibility of recall bias and the inability of these studies to separate cohort from age effects. Further, the predominance of evidence is from the U.S. and other English-speaking countries; these trends may not apply cross-nationally. Evidence is less consistent for gender gap narrowing in mean alcohol consumption. Fourth, substantial cross-cultural and demographic variation is evident in the emergence of cohort effects. Australia and Western European countries have little evidence of cohort effects, whereas Northern and Eastern European countries have increases in drinking, household expenditures on alcohol, and alcohol-related mortality. In summary, the picture that emerges is one in which younger birth cohorts, particularly in the U.S., Northern, and Eastern Europe, and increasingly women, consume consumer more alcohol, engage in more episodic drinking and evidence more alcohol problems.
Taken together, the present review indicates substantial evidence for cohort effects in various alcohol indicators including lifetime use, consumption, and alcohol problems, suggesting that alcohol use across the life-course aggregates within cohorts. However, we note that synthesis of this evidence is limited by a reliance on cross-sectional designs. Of those studies with the most rigorous methodology, the evidence for cohort effects in lifetime any alcohol use is weak (only one study rigorously assessed cohort effects and found none), the evidence for cohort effects in prevalent alcohol consumption across the life-course is strong (although variation in the cohort assessed, the measure used, and the geographic location of the sample make synthesis a challenge), and the evidence for cohort effects in alcohol problems/disorder is limited by the fact that no strong rigorous studies have been conducted (although less strong study designs predominately support evidence for a cohort effect across many countries).
While the identification of at-risk cohorts provides an initial characterization of drinking patterns over time, cohort analysis does not typically test the mechanisms through which birth cohorts collectively change in their drinking patterns. Neve et al. (Neve, Diederiks et al. 1993) divides potential mechanisms for changing alcohol consumption patterns into ‘endogenous’ and ‘exogenous’ mechanisms. ‘Endogenous’ mechanisms for changing drinking patterns are characterized by social norms and social transmission of alcohol consumption behavior at the population level. Using the example of the U.S., scholars have noted rapid social change among younger generations in the latter half of the 20th century (Traub and Dodder 1988) whereby the social norms of the older generation are not transmitted to the younger generation. Historical analyses suggest that cohorts following those with permissive norms regarding alcohol and drug use will have more restrictive norms (Musto 1999). These shifting social norms across generations could potentially account for the observed decrease in mean drinking but the suggestive increase in heavy episodic drinking.
‘Exogenous’ mechanisms, on the other hand, are characterized by policies, laws, and economic factors that affect the availability of and consequences from alcohol consumption. Substantial evidence documents an ecological association between alcohol policy and consumption (Leifman and Romelsjo 1997; Henderson, Liu et al. 2004; Norstrom and Ramstedt 2005; Andreasson, Holder et al. 2006); countries with stricter policies have less per capita consumption (Brand, Saisana et al. 2007). In the U.S. in the 1980’s, policies increased the minimum drinking age (O’Malley and Wagenaar 1991) and penalties for driving after drinking (Tippetts, Voas et al. 2004), with simultaneous decreases observed in traffic-associated alcohol-related mortality. However, the studies presented in this review suggest that problems are increasing among the more recent cohorts at the same time as increasingly strict policies. Thus, it is possible that policy changes could be an effect rather than a cause of change across cohorts, although more empirical data is necessary to formally disentangle these relationships. Data on the interplay between legal policies and age-specific drinking patterns would be helpful in understanding the underlying mechanisms driving these trends. Of course, endogenous and exogenous mechanisms are not independent; social norms drive the acceptability of policies and laws regarding alcohol use, and vice versa (Musto 1999).
Multiple lines of evidence suggest that gender differences in the prevalence of heavy episodic use and alcohol disorders are narrowing. Several studies reported here present a 1:1 ratio in the gender difference in initiation of alcohol among adolescents (Johnson and Gerstein 1998; Johnson and Gerstein 2000), and multiple lines of cross-national evidence suggest substantial increases in the prevalence of alcohol disorders among women in more recently born cohorts. Given that evidence suggests greater long-term health risks among women with chronic alcohol disorders compared to men (Klatsky, Armstrong et al. 1992; Deal and Cavaler 1994; Urbano-Marquez, Estruch et al. 1995; Key, Hodgson et al. 2006), these trends suggest that public health efforts to specifically target heavy drinking in women are necessary.
As with the emergence of cohort effects more generally, the mechanisms through which this gender by cohort interaction arises remains unclear. Taking the example of the U.S., the 20th century was a period of substantial change in the societal structuring of gender roles. Women in the latter part of the 20th century entered higher education in high proportions, entered the workforce in record numbers, and delayed childbearing. Inglehart (Inglehardt 1977; Inglehardt 1990; Inglehardt and Baker 2000) has noted that economic prosperity and increased educational possibilities in the U.S. after World War II and extending into the 1970s, especially for women, provided a political framework in which norms and values could be shifted to more permissive attitudes toward drinking. International evidence from Seedat et al. (Seedat, Scott et al. 2009) suggest that these factors may have contributed to the increased risk of problematic alcohol consumption, in that gender-by-cohort interactions were much stronger in countries where these changes in gender roles were operative. Further, recent evidence indicates that differences in average frequency of drinking in certain venues between men and women can be partially explained by variation in economic participation and opportunity of women, above and beyond the influence of overall country wealth (Bond, Roberts et al.). Though speculative, the results of this study suggest that perhaps improvements in economic conditions for women over time may contribute to potential convergences in gender differences. A full explanation for these patterns, however, is unlikely to be straightforward. For example, there is evidence of socio-economic interactions with the effect of workplace on drinking in women (Brady and Randall 1999). Working outside the home is a protective factor for excessive alcohol consumption among women of lower-SES, and a risk factor for women of higher-SES (Brady and Randall 1999), suggesting that the web of competing risks for alcohol problems operate in a complex way in the development of drinking problems (Brady and Randall 1999; Nolen-Hoeksema 2004; Nolen-Hoeksema and Hilt 2006). Data from Gender, Alcohol, and Culture: An International Study (GenACIS) suggest that, cross-nationally, women who are paid for labor are more likely to engage in risky drinking compared to women who are not paid for labor, but that engagement in more than one life role (e.g., paid for labor but also in a partnership with children) is protective against risky drinking in both women and men (Kuntsche, Knibbe et al. 2009). Taken together, the role of the workplace on female drinking patterns remains understudied, yet important as an explanation for the patterns observed here.
Cross-national comparisons are limited due to the predominance of evidence from the U.S. (19 of the 31 papers reviewed in this article are based on U.S. samples). While studies in additional countries are needed to provide a more complete global picture of time trends in alcohol use, available evidence indicates cross-national variation in some alcohol outcomes but not in others. Cross-national variation is most striking in the age-period-cohort analysis of cirrhosis mortality in Europe conducted by Corrao et al. (Corrao, Ferrari et al. 1997), whereby later born cohorts were at increased risk in Eastern and Northern European countries only. Also striking is the lack of cohort effects in the Australian data, due to the near-ubiquitous lifetime alcohol consumption across time (Degenhardt, Lynskey et al. 2000). There is also, however, some cross-national convergence in the emergence of cohort effects. For example, later-born cohorts are at higher risk for heavy drinking in the U.S. (Kerr, Greenfield et al. 2009), Britain (Kemm 2003), and Denmark (Bjork, Thygesen et al. 2008), with a narrowing of the gender gap documented across these countries as well. Cross-national variation for some alcohol indicators is unsurprising; countries differ substantially in the per capita alcohol consumption (WHO 2011) and in maturational norms with respect to drinking. Thus, the emergence of cohort effects in alcohol consumption and other alcohol outcomes is bounded by the level of drinking in each country, making cross-national comparisons difficult and convergence in the effect unlikely.
Population-level policies and laws that differ between countries are likely to contribute to cross-national variation in alcohol patterns (Casswell and Thamarangsi 2009). As a case study in the effects of these policies and laws on a particular country, consider the recent changes in population-level alcohol consumption in Finland. While per capita consumption is higher than the world average (Rehm, Mathers et al. 2009), Finland has relatively low consumption compared to other European countries, with strict government control of the production and distribution of alcohol (World Health Organization: alcohol control database). Finland’s entry into the European Union in 1995 required de-monopolization of many aspects of alcohol production and trade policy. In response to increasing international pressure, Finland substantially lowered alcohol taxes in 2004. In the following year, alcohol sales increased by 50% and per capita consumption rose by 10% (Anderson, Chisholm et al. 2009). This example highlights how local and international politics and history align to shape alcohol consumption patterns within and across geographic localities.
This review also highlights some methodological issues that should be considered in future research on alcohol consumption patterns over time. We will focus on four primary recommendations for future research.
First, the lack of any standardization of alcohol quantity and frequency measures creates ambiguity in the synthesis of research on trends over time. This is difficult for practical purposes, as standard drink sizes and drinking practices vary across countries. However, most studies included in the present review reported only a single measure of alcohol consumption, typically either drinking quantity or drinking frequency. While different measures of quantity and frequency each provide unique and important information about changing alcohol consumption patterns, more research that reports multiple quantity/frequency measures within a single study (where standard drink size would be controlled) would be beneficial for drawing conclusions. An example of the utility of cross-measure comparison is seen in Kerr et al. (Kerr, Greenfield et al. 2009), where in a single study it was reported that younger cohorts were at higher risk for heavy episodic drinking but were consuming fewer monthly drinks. Moreover, the quality of the data ranges across studies in potentially systematic ways; for example, there may be cross-cultural variation in the extent to which alcohol consumption is validly reported on surveys, limiting the validity of cross-national differences observed here.
Second, limited data are available from prospective studies on cohort effects in the incidence of alcohol-related phenomena. Studies identified in this review either used the lifetime proportion of alcohol users or retrospective age of onset reporting to estimate some measure of incidence. While these studies provide important information, the potential for selection bias, confounding of age and period effects, and recall bias (Rehm, Irving et al. 2008) limit the ability to draw conclusions. While incidence data is difficult to collect on a large scale due to practical difficulties of time and expense involved in prospective data collection, it is critical for the tracking of trends over time in order to design and implement public health programs when necessary.
Third, data on cohort effects generally and cohort effects on gender differences from a broader cultural framework are needed. As noted, most studies in this review are based on U.S. samples; the remaining countries included in these papers are also predominately high-income western countries, save for Seedat et al. (2009) and van Heerden et al., (2009). The data from Seedat et al. (2009) indicate substantial cross-cultural variation in the effect of birth cohorts on gender differences, suggesting that further data from a wider cross-section of countries could reveal important patterns that are obscured in the present published literature focused on Western countries. Emerging evidence suggests that young people, and especially girls, in many countries, and especially Eastern Europe, are increasing in the frequency of drunkenness faster than boys and girls in other countries, suggesting that data on cohort effects in these countries is urgently needed (Kuntsche, Kuntsche et al. 2011).
Finally, most studies included in this review are based on self-report of respondents via surveys. Those studies that did not collect data from self-report (Corrao, Ferrari et al. 1997; Kendler, Prescott et al. 1997; Rosen and Haglund 2006; Aristei, Perali et al. 2008; Zhang, Guo et al. 2008) did not consistently document cohort effects, raising questions about the validity of the self-report data. Survey data is an imperfect tool for gauging the total amount of alcohol consumed (Del Boca and Darkes 2003), and the validity of self-reported alcohol consumption is likely subject to some of the same social norms identified above as a potential mechanism driving cohort effects in alcohol consumption. Thus, to at least some extent, the cohort effects in alcohol consumption could arise from cohort effects in the reporting of alcohol consumption only. Age-period-cohort models using diverse data sources such as reported expenditures and alcohol-associated mortality would be beneficial to disaggregate cohort effects in consumption from those due to reporting.
The identification of at-risk cohorts provides crucial information on the dynamics of ever-changing societies. As each cohort progresses through life, they are collectively faced with different social, economic, and political barriers and resources that may impact health in significant ways. Birth cohort analyses can be used to predict future alcohol-associated harm and disability for public health planning and intervention by identifying the structural and societal-level forces that shape trends over time in alcohol phenomena. Collectively, the studies presented above suggest that drinking patterns aggregate in a cohort-specific manner, with variation in alcohol consumption over time that is best understood when incidence, consumption and disorder over time are modeled as a function of birth cohort. An especially important observation from this literature is that women in younger cohorts increasingly engage in risky drinking practices, challenging the paradigm of alcohol consumption as a male-dominated health issue, and suggesting a substantial role for population-level social norms in shaping consumption patterns.
This research was supported in part by funding from the National Institute of Drug Abuse (F31 DA026689, K. Keyes), the National Institute on Alcohol Abuse and Alcoholism (K05 AA014223, Hasin; R01AA09963, Li) and the New York State Psychiatric Institute. We thank Dr. William Kerr for feedback on earlier drafts of this manuscript.
The authors report no conflicts of interest.