Can a Label Help me Drink in Moderation? A Review of the Evidence on Standard Drink Labelling

ABSTRACT

Introduction: Understanding the concept of a standard drink (SD) is foundational knowledge to many public health policies aimed at reducing alcohol-related harms. These policies include adhering to low-risk drinking guidelines, screening brief intervention and referral activities, and counter alcohol-impaired driving initiatives. A lack of awareness of SDs might preclude the effectiveness of these interventions. A systematic review was conducted to review the evidence about how effective alcohol labels are in communicating SD information to the consumer. Methods: A systematic review was conducted to identify peer-reviewed articles and grey literature from relevant indexes from January 1990 to January 2016. Additionally, policy makers and researchers in countries where standard drink labels (SDLs) have been implemented were consulted to help identify relevant literature. The search strategy was focused on the impact of SDLs relative to a range of outcomes, including awareness of SDs, pouring behaviors, and consumption patterns. Results: Eleven records were eligible for inclusion. The evidence suggests that knowledge of the definition of an SD is low. However, SDLs can help individuals more accurately identify and pour an SD. SDLs need to be supported by educational initiatives to help the consumer understand the SD information provided on the beverage container. To date, there has been no comprehensive evaluation of the impact of SDLs. Conclusions: SDLs have the potential to increase awareness of SDs and facilitate the monitoring of personal alcohol consumption in the context of a comprehensive alcohol strategy. However, their impact on drinking behaviors requires further exploration, especially among high-risk populations.

KEYWORDS:

• Alcohol
• alcohol by volume
• alcohol unit
• labelling
• review
• standard drink

Introduction

Alcohol is a leading contributor to disease and disability, ranking fifth internationally and higher in some countries (Lim et al., 2012). International research indicates that population rates of chronic disease and acute harms rise as overall alcohol consumption rates increase (National Alcohol Strategy Working Group, 2007). Globally, 40% of adults drink; consuming an average of 17.1 L each per year (Shield et al., 2013).

The Global Alcohol Strategy of the World Health Organization (WHO, 2010) and several national alcohol strategies (National Alcohol Strategy Working Group, 2007; the Scottish Government, 2009) call for a comprehensive set of complementary population-level alcohol policies and programs to reduce the harmful consumption of alcohol and attendant harms (Paradis, 2016; WHO, 2010). Several of these interventions require individuals to track their alcohol consumption. For example, several countries have set low-risk alcohol drinking guidelines (LRDGs) to help individuals monitor their drinking and reduce alcohol-related harms (Butt, Beirness, Gliksman, Paradis, & Stockwell, 2011; Department of Health, U.K., 2015; National Health and Medical Research Council, 2009). LRDGs typically provide upper limits on the number of standard drinks (SDs) that individuals should not exceed in a day and/or week. So that these guidelines can be meaningfully applied, they are presented in terms of SDs. An SD is an unvarying unit of measurement that takes into account the strength and volume of a beverage and allows for meaningful comparisons across different alcohol products. For individuals to accurately track consumption against these guidelines, knowledge of the concept and definition of an SD is required.

Similarly, screening, brief intervention and referral (SBIR) activities conducted by health professionals often use SDs as a means of assessing individual risk, and communicating information and advice on alcohol consumption (College of Family Physicians of Canada, 2012). SBIR activities require both the health professional and the client to have some knowledge of SDs so that alcohol consumption can be accurately communicated, risk assessments accurately conducted, and advice or interventions appropriately applied (European Alcohol Policy Alliance, 2011). This knowledge might be especially important for higher-risk populations (U.K., Office for National Statistics, 2010) or for women during pregnancy (Kaskutas & Graves, 2000) or thinking of becoming pregnant.

Many initiatives to prevent impaired driving depend upon individuals' ability to track their alcohol consumption. Several countries impose “per se” laws whereby it is illegal to operate a vehicle with a concentration of alcohol in the body in excess of a specified threshold (Canadian Centre on Substance Abuse, 2016). As a result, drinkers have an interest in determining how much they can drink and still remain below the legal blood alcohol concentration (BAC) limits. BAC monitoring requires drivers to track their consumption using SDs.

In addition to supporting public health policy, there is increasing demand from consumers around the world for more information on food labels, and alcohol is no exception. While food labelling is becoming increasingly comprehensive, alcohol is often exempt from specific labelling components (O'Brien, 2014). Alcohol is not an ordinary food commodity; it has intoxicating effects, can lead to dependence and is linked with acute and long-term harms. Because of these characteristics, individuals have a right to know how much alcohol they are consuming.

The concept of an SD is not one with which many drinkers are familiar or that they can accurately apply in a drinking context (Kerr & Stockwell, 2012; Osiowy, Stockwell, Zhao, Thompson, & Moore, 2015; Devos-Cromby & Lange, 2008; Zandy, Pang, Ho, & Matthews, 2013). This observation is also true for health professionals (Webster-Harrison, Barton, Sanders, Anderson, & Dobbs, 2002; Welsh et al., 2014), as well as those in the alcohol service industry (Devos-Cromby & Lange, 2008). Knowledge of and ability to pour an SD varies by beverage type (Kerr & Stockwell, 2012; Foundation for Alcohol Research and Education, 2012) and actual pours typically contain more alcohol than an SD, leading to individuals underestimating their alcohol consumption (De Visser & Birch, 2012; Stockwell, Zhao, & MacDonald, 2014; Zhao, Stockwell, & Thomas, 2015).

Many factors can pose barriers to pouring an SD. The range of alcohol content within and across beverage types presents a challenge (Stockwell & Honig, 1990); as alcohol content rises, the discrepancy between actual pours and an SD increases (Devos-Cromby & Lange, 2008). The shape of the glassware (Walker, Smarandescu, & Wansink, 2014), size of the pouring container (Devos-Cromby & Lange, 2008) and glassware (Devos-Cromby & Lange, 2008; De Visser & Birch, 2012; Stockwell & Honig, 1990; Kerr, Patterson, Koenen, & Greenfield, 2009), the color contrast of the beverage against the glassware (Walker et al., 2014), as well as social norms (Zandy et al., 2013), can all influence pour sizes. Finally, many formats of labelling alcohol content, such as original gravity, original proof and percent alcohol by volume (% abv), are confusing to the consumer due to the calculation required to determine the alcohol content of the container in terms of SDs. These labelling formats can preclude the use of this information for tracking consumption accurately (Stockwell & Honig, 1990; Kerr & Stockwell, 2012).

SD information can provide individuals with a better understanding of the “dose” of alcohol they have consumed so that they can make informed choices. Loxey et al. (2004) report that general awareness of the concept of an SD had risen since the implementation of SDLs in Australia in the 1990s; however, there has been no formal evaluation of the Australian labelling strategy since it was implemented on December 22, 1995. Loxey et al. (2004) indicate that the people who can recall the label information are those who drink the most, making labels an efficient way of providing this information to higher-risk drinkers.

There have been a number of calls to action for enhancing alcohol content labelling and including SD information on alcohol packaging (Webster-Harrison & Barton, 2002; Marjot, 2012; “Getting tight on units of alcohol,” 2001; Knai, Petticrew, Durand, Eastmure, & Mays, 2015). This paper seeks to synthesize what is known about the impact of SDLs, specifically with regards to knowledge and awareness of SDs and drinking-related behaviors such as SD pours and drinking patterns.

Methods

A literature search was conducted of research published from January 1990 to January 2016 using Pubmed, PsychInfo and Scopus, and the following keywords: alcohol^* and “standard drink” or unit and label^*. To identify grey literature, a Google search was undertaken using the keywords included in the primary search strategy. Additionally, policy makers and academics in countries where SDLs have been implemented were consulted to identify reports or evaluation materials that assess the impact of SDLs. All records (n = 408) were assessed for eligibility by title and abstract screening. Records were selected for inclusion if they were original studies that focused on the impact or probable impact of SDLs, contained outcome data and were published in English between January 1990 and January 2016 inclusive.

To supplement the original search strategy, reference lists of identified records (n = 37) were hand-searched. The selected records were reviewed by an expert working group to ensure each selection met the criteria for inclusion outlined above. The working group (n = 3) consisted of three alcohol policy researchers with backgrounds in both quantitative (n = 3) and qualitative research (n = 2) with expertise in alcohol (n = 2) and nutrition labelling (n = 1) research and implementation. Full-text articles were retrieved and reviewed, and relevant information was extracted. See Figure 1 for a flowchart of the search strategy and its yield.

Figure 1. Search strategy flowchart.

Results

Search yield

The search strategy identified seven articles, two reports, one presentation, and one letter to the editor that met the inclusion criteria and were selected for analysis. See Table 1 for a summary of the selected articles. The search strategy also returned six literature reviews that included the topic of SDLs (Knai et al., 2015; Public Health Ontario, 2014; Esser & Seigel, 2014; Martin-Monero et al., 2013; Kerr & Stockwell, 2012; Stockwell & Single, 1997). While these reviews were not selected for inclusion, the reference lists were hand searched for relevant articles. A report by the Commonwealth Department of Health in Australia was identified in Loxey et al. (2004); however, complete citation information was not provided and the original report could not be accessed.

Table 1. The impact of standard drink labels on awareness, knowledge, and drinking-related behaviors.

Study	Sample size	Population	Jurisdiction	Outcome measures	Results
Experimental research
Stockwell, Blaze-Temple, & Walker (1991a)	n = 72; 24 subjects in each of the three experiments	Convenience sample of drinkers above the legal drinking age	Australia	Exp 1: Ability to pour a SD of beer across a range of glass sizes, with SDL vs. % abv label	Exp 1: There were significant advantages of SDLs over % abv labels in pouring a SD of beer, especially for unusual glass sizes
				Exp 2: Ability to pour a SD of wine across a range of glass sizes, with SDL vs. % abv label	Exp 2: There were no significant advantages for SDLs over % abv labels for pouring a SD of wine
				Exp 3: Ability to pour a SD of beer across a range of different strength products, with SDL vs. % abv label	Exp 3: There were significant advantages for SDLs over % abv labels when pouring an SD of low alcohol content beer
Webster-Harrison, Barton, Sanders, Anderson, & Dobbs (2002)	n = 196	Convenience sample of drinkers above the legal drinking age	UK	Ability to measure alcohol content in SDs using SDLs vs. % abv labels across different beverage types	There was a significant advantage of SDLs over % abv labels in estimating the alcohol content of alcohol container for all beverage types (beer, wine, and gin)
Osiowy, Stockwell, Zhao, Thompson, & Moore (2015)	n = 305	Convenience sample of drinkers above the legal drinking age	Canada	Ability to measure hypothetical consumption in SDs using SDLs vs. % abv labels across different beverage types of differing strengths	There was a significant advantage of SDLs over % abv labels in estimating hypothetical consumption across different beverage types and strengths except for regular strength beer
Hobin et al., (2015)	n = 2016	General adult population of Ontario, Canada	Canada	Ability to (a) define an SD, (b) estimate alcohol content in SDs, and (c) estimate consumption relative to the LRDGs in SDs, using SDLs vs. % abv labels presented in varying formats	There was a significant advantage of SDLs over % abv labels in estimating the amount of alcohol in a SD, estimating the amount of alcohol in a container and tracking hypothetical consumption against the LRDGs
Survey research
Chan, Chan, P'ng, & Segarajasingam (1997)	n = 164	Convenience sample of drinkers above the legal drinking age	Australia	Awareness of the SD concept, awareness of SDLs on containers, and ability to use SDLs to measure consumption	89% were familiar with the SD concept; 44% were aware of SDLs on containers, with fewer young drinkers having noticed the labels; and the majority underestimated the number of SDs in a container
Gill & O'May (2006)	n = 180	Female undergraduate students in the UK	UK-Scotland	Ability to estimate the number of SDs in a container and attitudes toward SDLs	The majority did not identify the number of SDs in a container of their preferred beverage; the majority expressed positive attitudes toward enhanced alcohol labels with LRDG and SD information
Haines & Stockwell (2003)	n = 154	Convenience sample recruited from a local shopping mall	Australia	Awareness of the SD concept, awareness of SDLs on containers, and ability to measure the alcohol content in SDs of their preferred beverage and random beverages using SDLs	Awareness of the SD concept had increased from 59% in 1993 to 97% in 2003; the majority were aware of SD information on containers; the vast majority were able to correctly measure the alcohol content in SDs of their preferred beverage and randomly selected beverages
Qualitative research
Jones & Gregory (2009)	n = 44; 6 focus groups of 6–10 participants each	University students 18 years and older	Australia	Knowledge of SDs, knowledge of and attitudes toward SDLs, use of SDLs by self and others	The majority understood the concept of a SD; all were aware of SDLs and perceived them to be a legal requirement designed to inform people about how much they should drink; participants indicated they and others used SDLs to choose stronger or more cost effective drinks and to monitor or control their drinking in specific contexts
European Commission & CIROC (2011)	n = 126	Young adults 18–25 years of age	Belgium, France, Romania, Lithuania, Spain, Hungary	Perceptions and behaviors regarding alcoholic drink labelling	Brand, alcohol content and price have the most influence on youth purchase decisions. Most participants were not familiar with the SD concept and valued other sources of information over labelling
Mixed-methods research
Stockwell, Blaze-Temple, & Walker (1991b)	Part 1 (experiment): n = 102	Convenience sample of drinkers above the legal drinking age	Australia	Exp: Ability to measure alcohol content in SDs across a range of different beverage types and strengths using SDLs vs. % abv labels	Exp: There was a significant advantage of SDLs over % abv labels in measuring alcohol content in SDs across all beverage types and strengths
	Part 2 (survey): n = 157			Survey: Awareness, knowledge and use of SDLs	Survey: Two thirds of participants were familiar with the SD concept, less than one fifth of participants could define a SD, the vast majority of participants had not used SDs to track their drinking
Thompson, Vanderberg, & Fitzgerald (2009)	Part 1 (survey): n = 1523	General population (survey); teenagers, young adults and parents of teenagers (focus groups)	Australia	Survey: Attitudes toward SDLs	Survey: there was high support for both SDLs and % abv label information (95% and 96%, respectively)
	Part 2 (focus group): n = 45			Focus Groups: use of SDLs and % abv information	Focus Groups: younger participants used alcohol content information to select stronger or more cost effective drinks. Older participants used alcohol content information to select lower strength products and to monitor their consumption

Research design

The majority (n = 9) of the selected records implemented quantitative methods, with five papers implementing an experimental design and five papers including a survey component. Three of the selected records included a qualitative focus group component. Two papers implemented mixed-methods design incorporating a combination of methods and measures.

Sample description

All selected records included both genders with the exception of one article that included female students only. The majority of the selected records (n = 6) made use of convenience samples of drinkers recruited outside local shopping areas. Four of the records focused on younger populations including teenagers and young adults.

Location

The research was concentrated in Australia (n = 6). Two studies were conducted in the United Kingdom (U.K.), two in Canada and one across multiple countries (Belgium, France, Romania, Lithuania, Spain, and Hungary).

Research synthesis

Experimental research

Stockwell, Blaze-temple, and Walker (1991a) conducted three experiments to test participants' ability to pour an SD of their preferred beverage (beer or wine). For each of the experiments, 24 adult (18 years or older) beer or wine drinkers were recruited at a local shopping center, provided they could correctly answer two screening questions about SDs and % abv.

In the first two experiments, participants were asked to pour an SD of beer (exp. 1) or wine (exp. 2) into a glass on six occasions. For three of the occasions the container was labeled with % abv information and the other three occasions the container was labeled with SD information. For each label type, three different sized glasses and two brands of the same alcohol concentration (5% for beer and 12% for wine) were balanced across conditions. In the third experiment, participants were also asked to pour an SD (beer) on six occasions with the balanced label conditions. However, this time the glass size was held constant while the strength varied (2.2%, 5%, and 7.9%).

In each experiment, the amount poured was measured against an SD. The authors found a significant advantage of SDLs over % abv labels in assisting participants in pouring an SD of beer, especially for very small or large glassware. When using SDLs, 48% of beer drinkers accurately poured (within 10%) an SD; when using the % abv labels, 37% accurately poured (within 10%) an SD. These findings did not hold true for wine pours. Participants were equally inaccurate in pouring an SD of wine into varying glass types with SDL and % abv labels. Experiment 3 showed that participants were significantly more accurate in pouring an SD of beer with SDL compared to % abv labels. However, this advantage of the SDL was only present when pouring beer with a lower alcohol concentration (2.2% abv) (Stockwell et al., 1991a). Overall the authors concluded that SDLs offer an advantage in assisting consumers to pour an SD. While there was no significant difference in the mean pours for SDLs and % abv labels across the three experiments, when using SDLs the range of the mean values were narrower compared to % abv labels.

Webster-Harrison et al. (2002) asked 196 drinkers from the United Kingdom whether they use the voluntary SDLs to monitor their drinking and to participate in an alcohol estimation task. Participants were presented with a range of products labeled with % abv information and asked to estimate the alcohol content in SDs. These containers were followed by SD labeled products matched by type of beverage, strength and volume, and again participants were asked to estimate the alcohol content in SDs.

Results indicated that 59% of participants use the alcohol label to determine alcohol content. With the % abv labels, participants substantially underestimated the alcohol content, especially for beer, wine, and gin. Participants were much more accurate in estimating the number of SDs in the SD-labeled container compared to containers labeled with % abv information. For example, 94.4%–99.5% of the total sample correctly identified the number of SDs in a container for beer, wine, and gin when the label displayed SD information, compared to 17.7% (males) and 17% (females) for beer, 11.5% (males) and 8% (females) for wine, and 13.5% (males) and 8% (females) for gin, when the label displayed % abv information. The authors noted that the small print appeared to be a limiting factor in correctly identifying the alcohol content in SD labeled drinks.

Two Canadian studies implemented a method similar to that used by Stockwell et al. (1991b) to investigate whether SDLs could improve estimates of personal alcohol consumption. In the first of these studies, conducted by Osiowy et al. (2015), following their successful completion of two screening questions about SDs and % abv, 305 drinkers of legal drinking age (19 years and over) were invited to participate in a brief survey and six alcohol estimation tasks. Participants were presented with six different containers of their preferred beverage (beer, wine, or spirits). The containers had two different label types (SD vs. % abv) and three different strengths (low, regular, and high). Participants were asked to identify how many SDs they would have consumed if they drank X amount of each beverage (i.e., beer: three bottles; wine: half of the bottle; spirits: a quarter of the bottle).

Participants were significantly more accurate in estimating the number of SDs consumed when SD information was displayed on the label compared with % abv information. The advantage of SDLs was found across all beverage types and strengths with the exception of regular strength beer. The authors concluded the SDLs did not provide an added benefit in estimating the number of SDs in a bottle of regular strength beer because of the ease of this calculation given a bottle of regular strength beer is equivalent to one SD. Overall the advantage of SDLs was strongest for beer and higher strength beverages. In general, drinkers who preferred beer were younger and had a higher level of education were most accurate in estimating alcohol content. The majority of participants (68.3%) indicated that SDLs would help them follow the Canadian LRDGs. Almost three-quarters indicated they would use SDLs to monitor BAC levels prior to driving and 46.4% said they would use this information to plan for a party or select the best value products (Osiowy et al., 2015).

The second study using the method similar to Stockwell et al. (1991b) was conducted by Hobin et al. (2015). The researchers recruited participants through Neilsen and randomly assigned them to 1 of 11 different label conditions where the label content (% abv vs. SDL vs. LRDG vs. SDL and LRDG combined), size (50% of back panel vs. 100% of back panel), and format (LRDG information as pictogram vs. text) of the labels varied between conditions. This study assessed the impact of the various labels on participants' ability to define an SD, estimate the number of standard drinks in a container and estimate the number of SDs it would take to reach the daily limits defined in the Canadian LRDGs. Participants were also asked their opinion of the labels and how they would use the label information, if at all (Hobin et al., 2015).

When participants were exposed to % abv labels, the majority underestimated the amount of alcohol in an SD and the number of SDs per container. In addition, 80% of participants exposed to % abv labels were unable to accurately estimate the number of SDs required to reach the daily limit of the LRDGs. Accuracy of estimating intake increased when participants were exposed to SDLs. Those exposed to SDLs and labels containing SDL and LRDG information were significantly more likely to correctly estimate the number of SDs required to reach the daily LRDG limit. Label size was also an important element in facilitating accurate estimates of consumption: participants were better able to estimate alcohol consumption with larger labels. The majority of the sample were supportive of SDLs and indicated they would use this information to help them monitor their alcohol intake.

Overall, the experimental studies demonstrated an advantage of SDLs in assisting drinkers to estimate alcohol content and pour an SD of their preferred beverage. The advantage of SDLs over % abv labels held true across nonstandard conditions such as when the glass size varied and when alcohol content of the beverage deviated from the norm.

Survey research

In a letter to the editor of the Medical Journal of Australia, Chan, Chan, P'ng, & Segarajasingam (1997) summarize a study that assessed public awareness of SDLs in Australia, the visibility of the labels and whether drinkers were able to use this information to monitor their consumption. Approximately 7 months following the implementation of mandatory SDLs in Australia, 89% of the sample was familiar with the concept of an SD; however, only 44% had noticed the SDLs on alcohol containers, with younger drinkers being less likely to have noticed the labels. The authors also indicated that the majority of respondents underestimated the number of SDs in a bottle of beer, wine, or cooler, and found that participants took longer to locate SDLs on the back panel of containers compared to the front panel. The authors conclude that increasing the size of the label and placing this information on the front panel is preferred and increases accuracy.

Gill and O'May (2006) conducted a survey among 180 first-year, female university students who had completed their secondary education in the United Kingdom in the previous 4 years. Participants were asked about their drink preferences and drinking habits, as well as their ability to identify the number of SDs in their preferred beverage and their attitudes toward drink labels. The majority of the sample (91%) was spirit (44%), cooler (27%), or wine (20%) drinkers. When asked to identify the SD content of their preferred beverage, the majority of the sample, including 52% of all drinkers, 45% of wine drinkers, 61% of spirits drinkers, and 43.2% of cooler drinkers, offered no response. Overall, 13% of drinkers significantly underestimated the number of SDs contained in their preferred beverage type. When asked about their attitudes toward wine labels displaying drinking guidelines and SD information, nondrinkers were slightly more likely to hold favorable attitudes than drinkers. Overall, 55% of drinkers and 60% of nondrinkers expressed positive attitudes toward these label components; 40% of drinkers and 40% of nondrinkers expressed some form of negative attitude, such as they would not be noticed or would have no impact; and 5% of drinkers expressed neutral attitudes (Gill & O'May, 2006).

Approximately 7 years after the implementation of mandatory SDLs in Australia, Haines and Stockwell (2003) asked a convenience sample of survey participants whether they had heard of the concept of an SD and where they could find this information. They found that awareness of the term “standard drink” increased significantly from 1993 (58.8%) to 2003 (97.4%). When asked about possible source of SD information, 55.2% of participants identified labels as a source of this information without prompting. Participants were also asked to complete a timed SDL recognition task. Participants were presented with their preferred beverage and a beverage selected at random and asked to locate the SD information on the label. On average, participants took 8.1 s to locate this information on their preferred beverage and 6.9 s for the randomly selected beverage, with times overall ranging from 0.1 s to 42.8 s (Haines & Stockwell, 2003). Participants were then asked to identify the SD content of both their preferred beverage and the randomly selected beverage: 97.4% of participants correctly identified the SD content of their preferred beverage (93.5% for the beverage randomly selected at random). Many participants found this task difficult and 60.4% indicated the labels were too small.

The findings from the survey-based research suggest that individuals are relatively aware of the concept of an SD (89%–97.4%); however, they are less aware that this information is captured on alcohol containers in Australia (44%–55.2%). While the findings were mixed in terms of participants' ability to estimate the number of SDs in their preferred beverage, the results consistently indicated that the size and position of the label are important factors that influence drinkers' ability to use this information.

Qualitative research

Jones and Gregory (2009), the European Commission and the Centre de recherche et d'information des organismes de consommateurs (CRIOC) (2011) conducted focus groups to test different label concepts, including SDLs. Approximately 13 years following the implementation of mandatory SDLs in Australia, Jones and Gregory (2009) conducted focus groups with 44 Australian college students to determine how they use SDLs. Overall, participants understood the concept of an SD and were aware of SDLs in Australia. Participants indicated that they mainly used SDLs to help them choose stronger drinks and to get better value for their money. When prompted, participants also indicated that they might use SD information to make safer drinking choices, such as limiting consumption before driving, in certain social contexts and for medical reasons (Jones & Gregory, 2009).

The research by the European Commission and CRIOC involved focus groups with 126 youth and young adults across six European countries to evaluate different components of alcohol labels. They found that youth pay little attention to the label when selecting an alcohol beverage; however, when selecting a beverage based on the label, youth mostly based their decision on brand, and then strength and price. Several youth mentioned that they used the % abv information to select lower-strength products as a means of controlling intoxication. When asked specifically about SDLs, most did not understand the concept of an SD or the purpose of this information. Participants seemed to favor other sources of information (parents, school curriculum) over labels and suggested increases in price and limitations on availability as alternative alcohol control policies (European Commission & CRIOC, 2011).

The qualitative research highlights the possibility that younger drinkers might be more heavily influenced by product brand and price than SD information when making beverage selections. Given this possibility, youth might use SD information to help them select products that provide them with the best value for their dollar.

Mixed-methods

Stockwell, Blaze-Temple, and Walker (1991b) conducted two studies to test the effectiveness of SDLs as a means of communicating alcohol content in a way that helps individuals estimate the number of SDs in a container and individuals' preferences for different types of alcohol labelling.

In the first study, local shoppers were approached and asked to estimate the number of SDs in a container using a % abv label and a SDL. Beer and wine drinkers 18 years and older were given a brief tutorial on SD and % abv information. Participants who responded correctly to two skill-testing questions were invited to participate in a task of estimating the number of SDs in their preferred beverage (beer or wine). Beer drinkers (n = 52) were asked to estimate the total number of SDs in three 375 mL cans of either 2%, 5%, or 7.9% beer. Wine drinkers (n = 52) were asked to estimate the number of SDs presented in half of a 750 mL bottle of 14% wine. Participants made their first estimate using a % abv label and subsequently with an SDL.

Results from this experiment indicated a significant advantage of SDLs over % abv labels in correctly estimating the number of SDs in a container for both beer and wine. In comparison to the SDL, % abv labels consistently led to significantly lower and less accurate estimations of the number of SDs in a container for both beverage types. For example, beer drinkers underestimated the number of SDs in a container by 26% with the % abv label and by 4% when using the SDL. Similarly, wine drinkers underestimated the number of SDs by 25% with the % abv label and by 6% when using the SDL (Stockwell et al., 1991b).

In the second study, Stockwell et al. (1991b) asked adult drinkers (n = 257) about their awareness, knowledge and use SD information. Approximately 67% of participants had heard of the SD concept, 17.9% were able to correctly define an SD and 11.7% indicated that they had used SDs to track their own drinking. The majority of beer (62.6%) and wine drinkers (66.7%) indicated SDLs would be easiest to use compared to other alcohol content presentations.

In 2009, Thomson, Vanderberg, and Fitzgerald surveyed 1523 participants and conducted focus groups with 45 individuals to identify specific features of alcohol labels that could inform a new labelling regime in Australia. Amongst the survey participants, 95% supported having SD information and the % abv (96%) displayed on the label; both of which were mandatory components of alcohol labels in Australia at the time of the study.

The focus groups were segmented by life-stage (teenager, young adults, and parents with teenaged children), gender, and socioeconomic status. Younger participants reported using the SD and % abv information to identify high-strength products that offered the best value for their money, while the parents indicated they used this information to select lower-strength beverages and monitor their consumption prior to driving.

The findings from these mixed-method studies support the findings that the majority of drinkers are aware of the concept of an SD and that SDLs offer an advantage over % abv labels in estimating alcohol content. A consistent finding across all qualitative research components was that SDLs are unlikely to be effective in influencing drinking behaviors in isolation, particularly for youth. SDLs need to be supported by educational campaigns and a comprehensive policy package that include controls on price and alcohol availability (Jones & Gregory, 2009; European Commission & CIROC, 2011; Loxey et al., 2004; Thomson et al., 2012).

Conclusions

SDLs are a tool with the potential to increase awareness of SDs and facilitate alcohol consumption monitoring. Although no formal evaluation has been conducted, since the implementation of SDLs in Australia in 1995 people's awareness of the concept of an SD has increased from 59% in 1993 (pre-SDL implementation) to 97% in 2003 (post-SDL implementation) (Haines & Stockwell., 2003) and survey research suggests drinkers make use of the SDLs to ascertain the alcohol content of a beverage (Webster-Harrison et al., 2002). SDLs are the preferred labelling format among drinkers (Stockwell et al., 1991b; Hobin et al., 2015) and provide an advantage over % abv labels in assisting drinkers to estimate the alcohol content of their beverages (Stockwell et al., 1991b; Webster-Harrison et al., 2002; Osiowy et al., 2015; Hobin et al., 2015) and facilitate more accurate pouring of an SD (Stockwell et al., 1991a). The advantages of SDLs were shown to vary depending on beverage type and strength of the beverage (Stockwell et al., 1991a; Webster-Harrison et al., 2002; Osiowy et al., 2015). Furthermore, the visibility of the labels, determined by the size and location of the label, is an important factor that influences their utility (Chan et al., 1997; Haines & Stockwell, 2003; Hobin et al., 2015).

The impact of SDLs on drinking behaviors requires further exploration, especially amongst high-risk populations. While there is evidence from the warning label literature to suggest that high-risk drinkers are more likely to receive and recall information provided on a label (Greenfield, Graves, & Kaskutas, 1993), it is unlikely that the impact SDLs have on awareness and knowledge of SDs will be sufficient to bring about immediate positive changes in drinking behaviors (Thomson et al., 2012). Changing drinking patterns can be a very complex process. SDLs should be supported by education campaigns that explain how to use the information (Chan et al., 1997; Haines & Stockwell, 2003) and a comprehensive alcohol strategy that includes other targeted and population-level alcohol policies and interventions (Jones & Gregory, 2009; European Commission & CIROC, 2011; Loxey et al., 2004; Thomson et al., 2012).

Special consideration might be given to the varying impact of SDLs by subpopulation and beverage type. Older drinkers are more likely to use SDLs to select lower-strength beverages and manage their consumption (Thomson et al., 2012; Osiowy et al., 2015); however, younger drinkers might be less likely to notice (Chan et al., 1997; Gill & O'May, 2006) or pay attention (European Commission & CIROC, 2011) to SDLs and might use this information to select products of higher strength and lower price (Jones & Gregory, 2009; Thomson et al., 2012). While using SDLs to select products that offer the best value might run contrary to public health goals, when asked about factors beyond SDLs that influenced their purchasing decisions younger drinkers indicated they were more likely to look at product branding and price over alcohol content (European Commission & CIROC, 2011). Additionally, there is evidence to suggest that the efficacy of SDLs might not vary by age or drinking patterns (Hobin et al., 2015).

The advantages SDLs offer in estimating alcohol content vary depending on the beverage type and strength (Stockwell et al., 1991a; 1991b; Osiowy et al., 2015). Overall, drinkers appear to more accurately pour an SD for lower-strength beverages, even with the aid of SDLs. These findings suggest a possible need for different labels or approaches for different beverage types (Stockwell et al., 1991a) and subpopulations (Thomson et al., 2012; European Commission & CIROC, 2011).

There could also be practical barriers to implementing SDLs. Implementing SDLs might be challenging due to the fact that different jurisdictions have varying definitions of an SD and use different terminology (i.e., alcohol units). Furthermore, while the research shows that the size (Hobin et al., 2015) and placement of the SD information are important (Chan et al., 1997), label space is prime real-estate and there is growing competition for this space (O'Brien, 2014). Labelling could be even more challenging for multilingual societies where this information would need to be presented in multiple languages. Finally, label information is only useful if visible to the consumer. In on-premise establishments, the label might not be visible if the alcohol is consumed from a glass rather than the original container. Therefore, SDLs might need to be supported by other means of conveying this information, such as SD menu labelling and SD glassware markings to assist both service staff and consumers in pouring and tracking SDs.

To date, there has been very little evaluation following the implementation of mandatory SDLs in Australia and New Zealand in December 1995 and voluntary SDLs in the United Kingdom.¹ The research included in this synthesis should be interpreted with caution. Label size can affect participants' ability to use the label information (Haines & Stockwell, 2003; Hobin et al., 2015). Several of the studies were conducted with significantly larger labels than what was implemented in Australia (Stockwell et al., 1991a; 1991b; Osiowy et al., 2015; Hobin et al., 2015). Furthermore, much of the research was conducted with convenience samples and might not be representative of the general population; the efficacy of SDLs might vary by subgroup, although the findings are mixed. Finally, much of the research focused on beer and wine, but the ability to apply the SD concept varies by beverage type and strength (Stockwell et al., 1991a; 1991b; Osiowy et al., 2015).

Future research should focus on a comprehensive evaluation of SDLs that assesses both implementation and outcome with precise indicators of uptake and impact on awareness and understanding of the SD concept, knowledge of SDs, attitudes, as well as drinking and monitoring behaviors. For example, the majority of SDL research is based on cross-sectional survey findings or experimental designs that make use of SDL mock-ups. A natural experiment, examining the impacts of real-world SDL implementation, could make use of pre- and post-intervention measures, as well as a control state to more rigorously assess the impact of SDLs on awareness and understanding of the concept of SDs, ability to track consumption, and their impact on consumption patterns themselves. Future studies could also examine the efficacy of different SDL formats (e.g., size and placement), as current research has identified size and placement of SDLs as important factors that can influence consumers' ability to notice and make use of this information. Finally, as an education based intervention, future research on SDLs should be interpreted within the context of other alcohol policies and programs.

Declaration of interest

The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.

Additional information

Funding

Canadian Centre on Susbtance Abuse

Notes

1 In March 2011 the English Government announced a Public Health Responsibility Deal that aimed to have 80% of alcoholic products voluntarily labeled with the unit content (SDLs) by December 2013.