Support for a browser plug-in blocking online alcohol imagery among Australian participants: A qualitative interview study.

ISSUES ADDRESSED: Alcohol depictions are extremely common online, and there is a reported relationship between alcohol exposure and alcohol use. A browser plug-in specifically designed to block online alcohol depictions may be helpful to prevent the uptake of alcohol or increased alcohol use. METHODS: Twenty-five semi-structured interviews were conducted, examining participants' beliefs about the effects of online alcohol exposure and their support for developing the browser plug-in. RESULTS: Using reflexive thematic analysis, we found participants highlighted a clear impact of viewing alcohol online and offline alcohol use. Participants believed a browser plug-in that blocked alcohol was acceptable and would be especially useful for minors (to prevent alcohol initiation) and those who are aiming to reduce their alcohol use. CONCLUSIONS: Participants emphasised that viewing online alcohol exposure had an impact on drinking behaviours, such as increased craving and temptation. The browser plug-in was considered an easy intervention tool for both parents and people who are experiencing alcohol-related problems or trying to reduce their drinking. SO WHAT?: Participants' continuous support of an alcohol-blocking browser plug-in suggests that future health promotion strategies should consider the development of a prototype plug-in.

Exploring the relationship between solitary drinking and living without other adults on alcohol consumption during the COVID-19 pandemic.

AIMS: The COVID-19 pandemic presents the opportunity to learn about solitary drinking as many people were forced to spend time at home. The aim of this study is to examine the relationship between solitary drinking and living without other adults on alcohol consumption. METHODS: A longitudinal study with four survey waves (between May and November 2020) obtained seven-day drinking diary data from Australian adults living in New South Wales. In May, a convenience sample of 586 participants (Mage = 35.3, SD = 14.8; 65.3% women) completed the first wave. Participants then completed a survey in June (n = 319, 54.4% response rate), July/August (n = 225, 38.4% response rate), and November (n = 222, 37.9% response rate). Information about alcohol consumption including risky drinking (more than four drinks on one occasion), household structure, solitary drinking, and demographics were collected. We conducted random-effects panel bivariate and multivariable regression analyses predicting the number of standard drinks and risky drinking. RESULTS: Participants with solitary drinking occasions consumed more and had more risky drinking occasions than participants with no solitary drinking occasions, which was also found to be the case during lockdown. Living without other adults was associated with less consumption and less risky drinking than living with other adults. However, participants who lived without other adults and had frequent solitary drinking occasions (solitary drinking in >50% drinking occasions) reported more consumption than participants without a solitary drinking occasion. CONCLUSIONS: Individuals who consume alcohol alone and live without other adults or spend long periods of time at home may be more at risk of alcohol-related harm.

The prevalence of alcohol references in music and their effect on people's drinking behavior: A systematic review and meta-analysis.

Understanding the prevalence of alcohol references in music and their impact on alcohol drinking behavior is important given the increased accessibility to daily music listening with the proliferation of smart devices. In this review, we estimate the pooled prevalence of alcohol references in music and its association with drinking behavior. Systematic searches were conducted across four major databases (MEDLINE, PsycINFO, EMBASE, and CINHAL). Articles were selected following duplicate checking, title and abstract screening, and full-text review. Studies reporting the prevalence of alcohol-referencing music and/or investigating its association with drinking behavior were included. Pooled prevalence with 95% confidence intervals (CIs) were computed using a random effects model. Of 1007 articles identified, 26 met inclusion criteria and 23 studies comprising 12,224 songs were eligible for meta-analysis. The overall pooled prevalence of alcohol references in music (including lyrics and videos) was 24.0% (95% CI: 19.0%-29.0%). The pooled prevalence was 22.0% (95% CI: 16.0%-29.0%) for only lyrics, 25.0% (95% CI: 18.0%-33.0%) for only the visual elements of music videos, and 29.0% (95% CI: 21.0%-38.0%) for both the lyrical content and the visual components. Only three studies assessed the relationship between listening to music with alcohol references and drinking behavior, and all three reported a positive association. Whereas almost a quarter of all songs included references to alcohol, public health preventive measures are needed to reduce alcohol exposure from music. Future research is needed to understand fully the effect of music with alcohol references on drinking behavior.

How to apply zero-shot learning to text data in substance use research: An overview and tutorial with media data.

A vast amount of media-related text data is generated daily in the form of social media posts, news stories or academic articles. These text data provide opportunities for researchers to analyse and understand how substance-related issues are being discussed. The main methods to analyse large text data (content analyses or specifically trained deep-learning models) require substantial manual annotation and resources. A machine-learning approach called 'zero-shot learning' may be quicker, more flexible and require fewer resources. Zero-shot learning uses models trained on large, unlabelled (or weakly labelled) data sets to classify previously unseen data into categories on which the model has not been specifically trained. This means that a pre-existing zero-shot learning model can be used to analyse media-related text data without the need for task-specific annotation or model training. This approach may be particularly important for analysing data that is time critical. This article describes the relatively new concept of zero-shot learning and how it can be applied to text data in substance use research, including a brief practical tutorial.

Development of the lyrics-based deep learning algorithm for identifying alcohol-related words (LYDIA).

BACKGROUND: Music is an integral part of our lives and is often played in public places like restaurants. People exposed to music that contained alcohol-related lyrics in a bar scenario consumed significantly more alcohol than those exposed to music with less alcohol-related lyrics. Existing methods to quantify alcohol exposure in song lyrics have used manual annotation that is burdensome and time intensive. In this paper, we aim to build a deep learning algorithm (LYDIA) that can automatically detect and identify alcohol exposure and its context in song lyrics. METHODS: We identified 673 potentially alcohol-related words including brand names, urban slang, and beverage names. We collected all the lyrics from the Billboard's top-100 songs from 1959 to 2020 (N = 6110). We developed an annotation tool to annotate both the alcohol-relation of the word (alcohol, non-alcohol, or unsure) and the context (positive, negative, or neutral) of the word in the song lyrics. RESULTS: LYDIA achieved an accuracy of 86.6% in identifying the alcohol-relation of the word, and 72.9% in identifying its context. LYDIA can distinguish with an accuracy of 97.24% between the words that have positive and negative relation to alcohol; and with an accuracy of 98.37% between the positive and negative context. CONCLUSION: LYDIA can automatically identify alcohol exposure and its context in song lyrics, which will allow for the swift analysis of future lyrics and can be used to help raise awareness about the amount of alcohol in music. Highlights Developed a deep learning algorithm (LYDIA) to identify alcohol words in songs. LYDIA achieved an accuracy of 86.6% in identifying alcohol-relation of the words. LYDIA's accuracy in identifying positive, negative, or neutral context was 72.9%. LYDIA can automatically provide evidence of alcohol in millions of songs. This can raise awareness of harms of listening to songs with alcohol words.

Digital health interventions for all? Examining inclusivity across all stages of the digital health intervention research process.

Digital interventions offer many possibilities for improving health, as remote interventions can enhance reach and access to underserved groups of society. However, research evaluating digital health interventions demonstrates that such technologies do not equally benefit all and that some in fact seem to reinforce a "digital health divide." By better understanding these potential pitfalls, we may contribute to narrowing the digital divide in health promotion. The aim of this article is to highlight and reflect upon study design decisions that might unintentionally enhance inequities across key research stages-recruitment, enrollment, engagement, efficacy/effectiveness, and retention. To address the concerns highlighted, we propose strategies including (1) the standard definition of "effectiveness" should be revised to include a measure of inclusivity; (2) studies should report a broad range of potential inequity indicators of participants recruited, randomized, and retained and should conduct sensitivity analyses examining potential sociodemographic differences for both the effect and engagement of the digital interventions; (3) participants from historically marginalized groups should be involved in the design of study procedures, including those related to recruitment, consent, intervention implementation and engagement, assessment, and retention; (4) eligibility criteria should be minimized and carefully selected and the screening process should be streamlined; (5) preregistration of trials should include recruitment benchmarks for sample diversity and comprehensive lists of sociodemographic characteristics assessed; and (6) studies within trials should be embedded to systematically test recruitment and retention strategies to improve inclusivity. The implementation of these strategies would enhance the ability of digital health trials to recruit, randomize, engage, and retain a broader and more representative population in trials, ultimately minimizing the digital divide and broadly improving population health.

Support for policies restricting alcohol exposure in films: Does feeding back the amount of alcohol in films increase support?

INTRODUCTION: Alcohol exposure is common in popular films, and research has demonstrated a link between alcohol exposure and use. The likelihood of implementing specific policies to reduce the amount of film exposure is dependent on the level of public support; however, evidence is currently lacking. This study investigated how supportive people are of film-related alcohol policies and whether providing information about the amount of film exposure increased support. METHODS: Australian adults (N = 252) first provided estimates of how much alcohol they thought were in popular films and then were randomised to either see an infographic about the amount of alcohol in films or not. All participants rated how supportive they were of eight policies. RESULTS: The items 'alcoholic beverages and consumption should not be shown in G or PG rated films' (M = 3.54) and 'alcohol should not be glorified in films' (M = 3.49) were rated significantly higher than the scale's midpoint of 3 (p < 0.001). Participants who were older, female or reported lower alcohol use were more supportive of the policies. Only one policy item, 'information about alcohol sponsorship should be provided' received higher support from those who received the infographic compared to those who did not (M = 3.53 vs. M = 3.05; t(250) = -3.09, p = 0.002). DISCUSSION AND CONCLUSION: Participants were relatively supportive of film alcohol policies. However, providing information about the amount of alcohol in films did not make a difference on the level of support for most film alcohol policies.

Reducing intervention- and research-induced inequalities to tackle the digital divide in health promotion.

Social inequalities are an important contributor to the global burden of disease within and between countries. Using digital technology in health promotion and healthcare is seen by some as a potential lever to reduce these inequalities; however, research suggests that digital technology risks re-enacting or evening widening disparities. Most research on this digital health divide focuses on a small number of social inequality indicators and stems from Western, educated, industrialized, rich, and democratic (WEIRD) countries. There is a need for systematic, international, and interdisciplinary contextualized research on the impact of social inequality indicators in digital health as well as the underlying mechanisms of this digital divide across the globe to reduce health disparities. In June 2023, eighteen multi-disciplinary researchers representing thirteen countries from six continents came together to discuss current issues in the field of digital health promotion and healthcare contributing to the digital divide. Ways that current practices in research contribute to the digital health divide were explored, including intervention development, testing, and implementation. Based on the dialogue, we provide suggestions for overcoming barriers and improving practices across disciplines, countries, and sectors. The research community must actively advocate for system-level changes regarding policy and research to reduce the digital divide and so improve digital health for all.

Strategies to support midlife women to reduce their alcohol consumption: an Australian study using human-centred design.

Alcohol consumption is causally associated with long-term health-related consequences, such as cancer and cardiovascular disease, and short-term harms, such as accidents and injuries. Alcohol consumption has increased among midlife women (aged 40-65) over the last two decades in high-income countries. This study aimed to centre women's voices by using co-design methodologies to investigate what women identify as strategies that could assist them and other women their age to reduce their alcohol consumption. Human-centred design workshops were undertaken with 39 women, and conventional qualitative content analysis was used to analyse information from written workshop materials to develop categories in the data and count their occurrence. Six categories, or strategies, emerged, listed here from most to least represented: 'Participate in alternative activities to drinking alcohol', 'Track alcohol consumption and set goals', 'Seek support from family and friends', 'Drink alcohol-free beverages', 'Reduce supply of alcohol in the home' and 'Seek professional support'. Our findings identify strategies that are realistic and feasible to midlife women; our sample, however, likely reflects a more affluent subsection of this group, and as such, any focus on individual-level strategies must be complemented by policies that increase equitable access to healthcare and act on the social and commercial determinants of health. An intersectional approach to alcohol and other drug research is required to examine how the interplay of gender and other markers of social identities shape alcohol consumption.

Social and contextual factors associated with drinking before, during and after watching Australian Football League games: A pilot ecological momentary assessment study.

INTRODUCTION: The aim of this study was to: (i) determine the feasibility of using ecological momentary assessment to collect data from Australian Football League (AFL) fans; (ii) explore pre-game, during-game and post-game consumption patterns of AFL fans; and (iii) explore the social and setting-related factors associated with risky single occasion drinking (5+ drinks) among AFL fans. METHODS: Thirty-four participants completed up to 10 ecological momentary assessment surveys before, during and after 63 AFL games (n = 437 completed surveys). Surveys collected data about their drinking, and their social and environmental milieu (e.g., location, company). Binary logistic regression analyses clustered by participant identified which game-day characteristics were associated with higher odds of risky single occasion drinking. Significant differences between pre-game, during-game and post-game drinking on social and environmental factors were explored using pairwise comparisons. RESULTS: Risky single occasion drinking was more likely when games began in the early-afternoon (1-3 pm) than late-afternoon (3-6 pm), when participants watched the game at a stadium or pub compared to home, and when participants watched the game with friends compared to family. Pre-drinking was more likely before night games and post-drinking was more likely after day games. Drinking during the game was heavier when watching the game at a pub and when watching with a combined group of friends and family. DISCUSSION AND CONCLUSIONS: Preliminary findings suggest that social and contextual factors matter in the way alcohol is consumed while watching AFL games. These findings require further investigation in larger samples.

The promise of zero-shot learning for alcohol image detection: comparison with a task-specific deep learning algorithm.

Exposure to alcohol content in media increases alcohol consumption and related harm. With exponential growth of media content, it is important to use algorithms to automatically detect and quantify alcohol exposure. Foundation models such as Contrastive Language-Image Pretraining (CLIP) can detect alcohol exposure through Zero-Shot Learning (ZSL) without any additional training. In this paper, we evaluated the ZSL performance of CLIP against a supervised algorithm called Alcoholic Beverage Identification Deep Learning Algorithm Version-2 (ABIDLA2), which is specifically trained to recognise alcoholic beverages in images, across three tasks. We found ZSL achieved similar performance compared to ABIDLA2 in two out of three tasks. However, ABIDLA2 outperformed ZSL in a fine-grained classification task in which determining subtle differences among alcoholic beverages (including containers) are essential. We also found that phrase engineering is essential for improving the performance of ZSL. To conclude, like ABIDLA2, ZSL with little phrase engineering can achieve promising performance in identifying alcohol exposure in images. This makes it easier for researchers, with little or no programming background, to implement ZSL effectively to obtain insightful analytics from digital media. Such analytics can assist researchers and policy makers to propose regulations that can prevent alcohol exposure and eventually prevent alcohol consumption.

Drinking before sporting events in Australia: An ecological momentary assessment study.

This study investigates the individual and event-level correlates of drinking prior to attending Australian Football League (AFL) games among a sample of Australian spectators.A total of 30 adults (20% female, mean age = 32) completed a series of questionnaires (n = 417) before, during, and after an AFL match on a Friday, Saturday, or Sunday. Cluster-adjusted regression analyses were conducted to examine the impact of individual-level (age, gender, drinking habits) and event-level factors (time and day of game, location of viewing the game, viewing with friends or family) on drinking prevalence and the number of drinks consumed prior to the game.41.4% of participants engaged in drinking before attending an AFL match with a mean of 2.3 drinks consumed by those who reported pre-game consumption. Those aged 30 and over were significantly more likely to engage in pre-game consumption (OR = 14.44, p = 0.024) and consumed significantly more pre-game (B = 1.39, p = 0.030). Drinking before the game was significantly more likely before night games than daytime games (OR = 5.24, p = 0.039). Those who watched the game on-premise consumed significantly more before the game than those who watched the game at a private residence or at home (B = 1.06, p = 0.030). Those who watched games with family also drank significantly less prior to the game than those who attended without family (B=-1.35, p = 0.010).Addressing the contextual factors associated with drinking before the sporting events, such as the time of the game, may assist with efforts to reduce risky alcohol consumption and related harm.

Measurement of transdermal alcohol concentration using a wrist-worn enzymatic transdermal monitor.

BACKGROUND: New-generation transdermal monitors such as the ION Research Alpha Prototypes (ION RAP) hold promise for real-time alcohol measurement, with improvements in design features such as sampling frequency, size, and comfort. This paper aims to provide the first comparisons of the wrist-worn enzyme-based ION RAP and the fuel cell-based SCRAM-CAM against breath alcohol concentration (BrAC) readings. METHODS: Participants (N = 23) completed a total of 69 laboratory alcohol administration sessions while wearing both a prototype of the ION RAP wristband and a SCRAM-CAM ankle monitor; they also gave breath samples each 10 min. Analyses focused on latencies of transdermal alcohol concentration (TAC) after alcohol ingestion, correlations, and cross-correlations between BrAC and TAC measurements. RESULTS: A high failure rate of the ION RAP was observed (61.5% of the sessions were excluded due to the sessions not containing enough valid data). On average, the SCRAM-CAM and ION RAP detected alcohol 43 (SD = 21) and 50 (SD = 27) minutes after the first drink, with peak values reached after 138 (SD = 47) and 154 (SD = 56) minutes, respectively. SCRAM-CAM TAC peak (r = 0.185, p = 0.375) and area under the curve (AUC; r = 0.320, p = 0.118) showed small- and medium-sized correlations to BrAC. ION RAP TAC peak (r = -0.082, p = 0.698) and AUC (r = 0.040, p = 0.852) correlations to BrAC were close to zero. CONCLUSIONS: In this study, the new-generation ION RAP and the traditionally used SCRAM-CAM show similar delays in detection and similar TAC curves over time, despite using either enzyme- or fuel cell-based technologies, respectively. Due to high failure rates of the ION RAP prototypes and close to zero correlations to BrAC, further developments and improvements of these TAC wristbands are required for reliable and valid use in real-time alcohol measurement.

The Effects of a Web-Based Intervention to Reduce Alcohol Consumption Among Middle-Aged Women: Protocol for a Randomized Controlled Trial.

BACKGROUND: In the last decade, alcohol consumption among middle-aged women (40-65 years old) in Australia increased, despite declines in overall population consumption. Web-based, brief interventions are promising for reducing alcohol consumption, with efficacy shown in a wide range of populations. However, no published interventions have been designed specifically for and tested with middle-aged women. OBJECTIVE: This study aims to design and implement a web-based intervention intended to reduce alcohol consumption among middle-aged women. METHODS: The study is a 3-arm randomized controlled trial with a web-based intervention plus ecological momentary assessment (EMA) group compared to an EMA-only and a pre-post only control group. The study is aimed at middle-aged women, defined as women aged between 40 and 65 years, who consume alcohol at least weekly or who have consumed 4 or more drinks on 1 occasion in the last month. The intervention aims to reduce alcohol consumption through 4 modules that provide information on the health impacts of alcohol, mindfulness, social influences, and alcohol marketing. Intervention participants will also fill out biweekly EMA assessments. The comparators are EMA-only and pre-post control only. The primary outcome is alcohol consumption at 8 weeks compared between groups. Secondary outcomes are awareness of alcohol-related harms, readiness to change alcohol consumption, health status, mental health, and social support. RESULTS: Ethics approval for this project was received on September 11, 2019. The trial was registered on August 14, 2020. Recruitment has commenced, and the expected results will be available in 2022. CONCLUSIONS: This web-based intervention aims to reduce alcohol consumption among middle-aged women, a currently understudied cohort in alcohol research. TRIAL REGISTRATION: Australia New Zealand Clinical Trials Registry (ANZCTR) ACTRN12620000814976; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=12620000814976. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/34842.

Sex differences in young children's perceptions of situational drinking norms over time.

OBJECTIVES: To date, there have been no previous studies examining sex differences in the development of alcohol-related perceptions over time, a potential avenue for targeted prevention and early intervention efforts. This article examines any potential sex differences in young children's development of situational drinking norms over time. METHOD: Two hundred ninety-eight children (aged 4-6 years at baseline) completed the Dutch electronic Appropriate Beverage Task-which involves attributing alcoholic beverages to adults in varying situational contexts-annually over 3 years (2015, 2016, 2017). Three-level regression models were estimated examining whether perceptions of situational drinking norms varied as a function of the sex of the participant and whether there were any changes over time. RESULTS: Over time children did not attribute more alcoholic beverages to adults in various situations, instead both boys and girls became more accurate at correctly identifying situations in which drinking is more common. Over time, both boys and girls attributed more alcoholic beverages in common situations and less in uncommon situations, with no significant sex differences in attributions of alcohol found. CONCLUSIONS: We identified no significant sex differences in the development of situational drinking norms over time, suggesting that education and prevention campaigns can be unified/nontargeted by sex. However, given our findings differ from previous studies that identify significant sex differences between children on the development of other alcohol-related cognitions beyond situational drinking norms, there is a need for more international research in this space to understand the importance and nature of the development of alcohol-related perceptions over time. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Validation of the revised Alcohol Expectancy Task [rAET].

Alcohol expectancies are well-established determinants of alcohol consumption. Traditionally, expectancies were assessed using self-report questionnaires. However, researchers have increasingly begun to explore the use of pictographic assessments, for example, the revised Alcohol Expectancy Task [rAET]. The current research aimed to examine the factor structure of the task in relation to the hypothesised expectancy dimensions, participants' endorsement of these dimensions, and whether rAET scores are associated with drinking patterns. The rAET presents participants with several illustrated scenarios in which people are displaying emotions, following the Circumplex Model of Affect. For each scenario, participants select which type of drink the person presented in the illustration was most likely to have drunk. The rAET was administered online to a convenience sample (n = 1,192, female: 50.7%, Mage = 36.8, SD = 13.7). The confirmatory factor analysis demonstrated adequate fit of the hypothesised four-factor model. Mean comparisons demonstrated that positive expectancies were more pronounced than negative expectancies, and a distinction between arousal and sedation expectancies was found. Results of a structural equation model found that positive sedation and positive arousal were related to high drinking frequency. There were no links between usual quantity of alcohol use or binge drinking. It appears that the rAET can be successfully used to assess alcohol expectancies in terms of the emotions that are expected to occur from alcohol consumption. Future research is needed to establish the tool's assessment properties in different settings, and its utility in predicting alcohol consumption in different age groups, particularly children and young adolescents. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Audio-based Deep Learning Algorithm to Identify Alcohol Inebriation (ADLAIA).

BACKGROUND: Acute alcohol intoxication impairs cognitive and psychomotor abilities leading to various public health hazards such as road traffic accidents and alcohol-related violence. Intoxicated individuals are usually identified by measuring their blood alcohol concentration (BAC) using breathalyzers that are expensive and labor intensive. In this paper, we developed the Audio-based Deep Learning Algorithm to Identify Alcohol Inebriation (ADLAIA) that can instantly predict an individual's intoxication status based on a 12-s recording of their speech. METHODS: ADLAIA was trained on a publicly available German Alcohol Language Corpus that comprises a total of 12,360 audio clips of inebriated and sober speakers (total of 162, aged 21-64, 47.7% female). ADLAIA's performance was determined by computing the unweighted average recall (UAR) and accuracy of inebriation prediction. RESULTS: ADLAIA was able to identify inebriated speakers - with a BAC of 0.05% or higher - with an UAR of 68.09% and accuracy of 67.67%. ADLAIA had a higher performance (UAR of 75.7%) in identifying intoxicated speakers (BAC > 0.12%). CONCLUSION: Being able to identify intoxicated individuals solely based on their speech, ADLAIA could be integrated into mobile applications and used in environments (such as bars, sports stadiums) to get instantaneous results about inebriation status of individuals.