(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Impact on alcohol selection and online purchasing of changing the proportion of available non-alcoholic versus alcoholic drinks: A randomised controlled trial [1]

['Natasha Clarke', 'Behaviour', 'Health Research Unit', 'Department Of Public Health', 'Primary Care', 'University Of Cambridge', 'Cambridge', 'United Kingdom', 'School Of Sciences', 'Bath Spa University']

Date: 2023-04

This study provides evidence that substantially increasing the proportion of non-alcoholic drinks—from 25% to 50% or 75%—meaningfully reduces alcohol selection and purchasing. Further studies are warranted to assess whether these effects are realised in a range of real-world settings.

Study limitations include the setting not being entirely naturalistic due to using a simulated online supermarket as well as an actual online supermarket, and that there was substantial dropout between selection and purchasing.

A total of 607 participants (60% female, mean age = 38 years [range: 18 to 76]) completed the study and were included in the primary analysis. In the first part of a hurdle model, a greater proportion of participants in the “75% non-alcoholic” group did not select any alcohol (13.1%) compared to the “25% non-alcoholic” group (3.4%; 95% confidence interval [CI] −2.09, −0.63; p < 0.001). There was no evidence of a difference between the “75% non-alcoholic” and the “50% non-alcoholic” (7.2%) groups (95% CI 0.10, 1.34; p = 0.022) or between the “50% non-alcoholic” and the “25% non-alcoholic” groups (95% CI −1.44, 0.17; p = 0.121). In the second part of a hurdle model in participants (559/607) selecting any drinks containing alcohol, the “75% non-alcoholic” group selected fewer alcohol units compared to the “50% non-alcoholic” (95% CI −0.44, −0.14; p < 0.001) and “25% non-alcoholic” (95% CI −0.54, −0.24; p < 0.001) groups, with no evidence of a difference between the “50% non-alcoholic” and “25% non-alcoholic” groups (95% CI −0.24, 0.05; p = 0.178). Overall, across all participants, 17.46 units (95% CI 15.24, 19.68) were selected in the “75% non-alcoholic” group; 25.51 units (95% CI 22.60, 28.43) in the “50% non-alcoholic” group; and 29.40 units (95% CI 26.39, 32.42) in the “25% non-alcoholic” group. This corresponds to 8.1 fewer units (a 32% reduction) in the “75% non-alcoholic” compared to the “50% non-alcoholic” group, and 11.9 fewer alcohol units (41% reduction) compared to the “25% non-alcoholic” group; 3.9 fewer units (13% reduction) were selected in the “50% non-alcoholic” group than in the “25% non-alcoholic” group.

Adults (n = 737) residing in England and Wales who regularly purchased alcohol online were recruited between March and July 2021. Participants were randomly assigned to one of 3 groups: “25% non-alcoholic/75% alcoholic”; “50% non-alcoholic/50% alcoholic”; and “75% non-alcoholic/25% alcoholic,” then selected drinks in a simulated online supermarket, before purchasing them in an actual online supermarket. The primary outcome was the number of alcohol units selected (with intention to purchase); secondary outcomes included actual purchasing.

Increasing the availability of non-alcoholic options is a promising population-level intervention to reduce alcohol consumption, currently unassessed in naturalistic settings. This study in an online retail context aimed to estimate the impact of increasing the proportion of non-alcoholic (relative to alcoholic) drinks, on selection and purchasing of alcohol.

Funding: This research was funded in whole, or in part, by the Wellcome Trust [ref: 206853/Z/17/Z (TMM, GJH, MRM)]. For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2023 Clarke et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The aim of the current study was to estimate the impact of increasing the proportion of non-alcoholic drink options relative to alcoholic drink options, on the number of alcohol units that are (i) selected (with the intention to purchase) and (ii) purchased. We hypothesised that increasing the availability of non-alcoholic alternatives to alcohol would reduce the number of alcohol units selected and purchased.

In 2020, the UK Government made a commitment with the drinks industry to increase the availability of alcohol-free and low-alcohol products by 2025, although details on what this would involve have not been published [ 17 ]. Currently, most consumers purchase no or low alcohol drinks infrequently, although increased availability of these products is associated with an increase in their sales [ 18 ] and reductions in grams of alcohol purchased [ 19 , 20 ]. Non-alcoholic alternatives to alcohol (i.e., alcohol-free drinks and soft drinks marketed to adults) still only represent a small proportion of the market, however, which combined with their recent increase in popularity, suggests that there is substantial scope for increasing their availability.

There is clear interest in increasing the availability of non-alcoholic drink options, from the perspective of both consumers and policymakers. While the current market for alcohol-free beer, wine, and spirits represents a 3.5% share of the global alcohol industry and is therefore relatively small, it is rapidly growing [ 13 ]. For example, low and no-alcohol beer currently accounts for 3% of the total beer market [ 14 ], but this is forecast to increase by nearly 13% per annum over the next 3 years and is the fastest growing drinks segment in the United Kingdom [ 15 ]. In 2021, the no/low alcohol market grew by 6% globally, and in the UK, sales of non-alcoholic beer increased by 7% [ 16 ].

The promise of so-called “availability” interventions that change proportions of unhealthy (relative to healthier) products is highlighted by an emerging evidence base in relation to food [ 5 ]. A Cochrane systematic review found that reducing the proportion of available food products of a certain type (e.g., unhealthy snacks) resulted in markedly reduced selection of those foods [ 7 ], although the included evidence was limited in both quality and quantity. More recent field trials also suggest that decreasing the proportion of higher energy or meat-based foods reduces their consumption [ 8 – 11 ]. In terms of alcohol products, there is an absence of evidence, with no eligible studies identified in the aforementioned Cochrane review [ 7 ] or in a recent search update [ 5 ]. In what is, to our knowledge, the only previous study that has examined the potential of such an availability intervention applied to alcohol, the proportion of participants selecting an alcoholic drink decreased from 74% when one-quarter of the available drinks were non-alcoholic, to 51% when three-quarters were non-alcoholic [ 12 ]. However, this study only measured hypothetical and mandatory selection of a single drink from a limited range of 8 options. Studies using meaningful outcomes and conducted within more naturalistic contexts that include wider product ranges are necessary to inform the development and implementation of real-world interventions and policies.

Excessive alcohol consumption is one of 4 sets of modifiable behaviours—along with tobacco use, physical inactivity, and unhealthy diet—that make a major contribution to the global burden of non-communicable diseases, including cancer, heart disease, and stroke [ 1 , 2 ]. Given the influence of environmental cues upon consumption and related behaviours, interventions that change physical and economic environments in which these behaviours occur have the potential to reduce alcohol consumption [ 3 ]. Altering the availability of alcohol products has been identified as a particularly potent approach [ 4 ] but has typically been examined in relation to demographic, temporal, or spatial restrictions (e.g., by age, opening hours, or number or density of retail outlets), and not in terms of changing the range of available products. One intervention of this kind, potentially scalable to population-level and currently untested, involves increasing the proportion of non-alcoholic (relative to alcoholic) drink options that are available to select, purchase, and ultimately consume [ 5 ]. This can be achieved by either making more non-alcoholic options available, removing some alcoholic options, or by doing both and so retaining the same overall number of options [ 6 ]; the latter is assessed in the current study.

Methods

The trial was prospectively registered (ISRCTN: 11004483). In addition, both the study protocol (https://osf.io/qfupw) and a detailed statistical analysis plan (https://osf.io/4yuca) were preregistered on the Open Science Framework (OSF). The study was approved by the Faculty of Life Sciences Research Ethics Committee at the University of Bristol (reference no: 116124). Trial reporting follows CONSORT 2010 guidelines.

Study design The study used a parallel-groups randomised controlled design. Individual participants were randomly allocated without stratification to one of 3 groups differing in the proportion (%) of non-alcoholic versus alcoholic drink options available for selection: Group 1: “25% non-alcoholic/75% alcoholic”; Group 2: “50% non-alcoholic/50% alcoholic”; Group 3: “75% non-alcoholic/25% alcoholic”.

Setting The study was conducted online using simulated and real online supermarkets. First, participants completed a simulated supermarket selection task hosted on the Qualtrics online survey platform (see https://osf.io/2cy7t for example task images). The simulated supermarket was designed to look and function as similarly to the actual online supermarket as was possible within Qualtrics. Drinks were displayed in rows of 4 drink options and participants could add these to their basket, which displayed a total price after the selection had been made. Following this, participants were required to purchase drinks in Tesco online supermarket (Tesco.com), the largest national supermarket in the UK.

Participants To be eligible for the study, participants had to be adults aged 18 years and over residing in England or Wales, who consumed beer or wine regularly (i.e., at least weekly), and purchased these drinks at least monthly from Tesco.com, with a minimum spend of £20. Participants had to be willing to complete a shop at Tesco.com following completion of the selection task, book a delivery or click-and-collect slot, and send proof of purchase (their receipt) to the research team. Similar proportions of males and females of a range of ages were recruited via Roots Research (https://rootsresearch.co.uk/), one of the largest research agencies in the UK, with a high-quality panel of over 350,000 participants. Recruitment occurred between March and July 2021. Sample size. A previous online study compared the impact on drink selection of altering non-alcoholic versus alcoholic drink availability [12]. The proportion of participants selecting an alcoholic drink decreased from 74% when non-alcoholic drink availability was low (25% of drink options), to 61% when availability was medium (50% of drink options), and 51% when availability was high (75% of drink options) (i.e., a difference of 13% and 10%, respectively, between adjacent groups). However, only a single drink was selected in this online study and there was no intention to purchase the selected drinks (i.e., decisions were purely hypothetical). As such, to our knowledge, there was no comparable evidence available from which to estimate effects of this intervention on selection or purchasing behaviour of multiple drink options. Available resources allowed recruitment of around 600 participants. As an illustrative calculation, assuming 15% attrition, a sample of 510 participants (170/group) was sufficient to detect an effect of d = 0.3 for the primary outcome for a two-group t test with alpha of 5% and 80% power. Using pretesting data (approximately 5/group), the conservative SD estimate was 12.1 units (i.e., the maximum group variance observed), indicating that the sample size was sufficient to detect a difference of 3.7 alcohol units selected between groups.

Randomisation and masking Randomised assignment of participants was completed via the default algorithm in Qualtrics with a ratio of 1:1:1. Participants were unaware of their group assignment throughout the study. The research team were blinded to allocation until participants had completed the primary outcome; the statistician completing the analysis was blinded to the allocation.

Intervention All participants viewed a total range of 64 drink options. This comprised (i) a range of beers, ciders, alcohol-free beer and cider alternatives, and soft drinks (32 options), and (ii) a range of wines, alcohol-free wine alternatives, and soft drinks (32 options), modelled on the available range of products on Tesco.com. Initial scoping work found that Tesco.com proportions of alcoholic versus non-alcoholic options were roughly 25% non-alcoholic (360 options) and 75% alcoholic (1058 options). Alcohol-free beer, alcohol-free cider, and alcohol-free wine options used in the task were matched as far as possible on brand and size characteristics with the alcohol options available online at Tesco.com. Additional alcoholic beer, cider, and wine options were selected based on the leading brands of lager, ale, mild and stout, cider, and wine [21–23] in Great Britain according to the number of consumers. Adult soft drinks were selected based on options that were commonly displayed next to alcoholic drinks in physical supermarkets and most likely to be consumed as a substitute for alcohol (meaning that children’s soft drinks, milk, and fruit juice were excluded). Participants viewed varying proportions of non-alcoholic and alcoholic drink options depending on their assignment: Group 1: “25% non-alcoholic/75% alcoholic”; Group 2: “50% non-alcoholic/50% alcoholic”; Group 3: “75% non-alcoholic/25% alcoholic”. The proportions used were based on previous food and alcohol studies [12,24]. Within each range of alcoholic drinks, there were the same number of beer as wine options, and within each range of non-alcoholic drinks, there were the same number of soft drinks as alcohol-free options. Participants were randomised to the order in which each subcategory (soft drinks; alcohol free; alcoholic) was presented within each of the beer and wine categories, and the order of drinks within each subcategory was also randomised. Each drink option was displayed as an image, below which was a text description of the drink (identical to Tesco.com), the % alcohol by volume (ABV) for drinks containing alcohol, and its price. Full details of the task, as well as the complete list of drinks, are in the S1 Supporting information (Table A and B). In the Typology of Interventions in Proximal Physical Micro-Environments (TIPPME) [3], this is classified as an “Availability × Product” intervention, while in a detailed conceptual framework specific to availability interventions [6], this is categorised as a “Relative Availability” intervention.

Procedure Participants were initially provided with an information sheet, instructions, and a link to the study via email. Participants were told the study was investigating “Adult drink preferences in England and Wales” and were not made aware of the study aim. Participant instructions outlined the stages of the study in detail, i.e., that participants were required to select the drinks for their next shop from Tesco.com in a simulated online supermarket (Stage 1), then to immediately go to Tesco.com to book a delivery slot and add these drinks to their shopping basket (Stage 2), and finally to send their receipt to the study team on their delivery or collection day (Stage 3). Once they had started the study task, participants were again presented with this information and provided consent. Participants were randomised, and in a simulated online supermarket environment replicating Tesco.com (within Qualtrics), they were shown the available drink selection. They chose all the drinks they wanted to purchase in their next online shop at Tesco.com. They were then shown their total drink selection and price and given the opportunity to amend their selection before continuing. Participants then completed demographic and drinking behaviour measures. After completing the simulated online supermarket task, participants were automatically sent an email detailing their selection. They were prompted to open this email and given further instructions for completing purchasing, alongside a direct link to Tesco.com. Participants placed their selected drinks in their Tesco.com shopping basket, along with any other items, booked their delivery or collection slot, and confirmed this within 48 hours. They were sent a reminder email on their delivery/collection day and requested to send an itemised receipt to the research team within 48 hours. Up to 2 follow-up reminders were sent, 2 and 4 days later. Purchases were recorded from receipts, including any additional drink purchases. Substitutions by the participant or by Tesco that were explained (e.g., not in stock) were marked as the original drink they attempted to purchase. Participants were debriefed via email and reimbursed £25 (approximately $35) for their time taking part in the study (but not the drinks they purchased).

Outcome measures Primary outcome. The primary outcome was the number of alcohol units selected in the context of a stated intention to purchase. In the UK, a unit is a standard measure of pure alcohol in a drink with 1 unit equivalent to 10 ml or 8 g of pure alcohol (this is equivalent to 0.56 of a US standard drink [25]). Participants were aware when selecting drinks in the task that they were required to subsequently purchase the drinks chosen and send proof of this to the research team (otherwise, they were not reimbursed). Units of alcohol were calculated for all drinks that were >0% ABV, i.e., alcoholic and “alcohol-free” drinks (which were defined as containing more than 0% and up to 0.5% ABV). This outcome was preregistered as the primary outcome as it was assessed in all participants who were exposed to the intervention and measured within the same context, i.e., the simulated online supermarket. Secondary and additional outcomes. Secondary outcomes were the number of alcoholic and non-alcoholic drinks selected, the number of alcohol units purchased, and the proportion (i.e., percentage) of total drinks selected and purchased that were alcoholic. Additional outcomes were the total number of drinks selected, and purchased, the number of alcoholic drinks purchased, and the number of non-alcoholic drinks purchased. Selection outcomes were assessed from the simulated online supermarket task, and purchasing outcomes were assessed from receipts after shops at Tesco.com were completed. Purchasing outcomes were calculated to include (i) additional drinks from study categories only (i.e., beer, cider, wine, and adult non-alcoholic drinks) and (ii) all additional drinks (i.e., all alcoholic and non-alcoholic drinks—excluding squash, juice, tea, coffee, and children’s drinks).

Other measures Demographic characteristics. Age, sex, and highest qualification attained (“Higher education or professional/vocational equivalents,” “A levels or vocational level 3 or equivalents,” “GCSE/O level grade A*-C or vocational level 2 or equivalents,” “Qualifications at level 1 and below,” “Other qualifications: level unknown,” or “No qualifications”). Qualifications classifications were based on UK definitions [26,27]. Household members. Number of adults (aged 18+) and of children (aged <18). Drinking behaviour risk. Alcohol Use Disorders Identification Test (AUDIT) [28], a 10-item clinical screening measure for assessing risk associated with participants’ drinking behaviour (low risk drinking: score 0 to 7; medium/hazardous risk drinking: score 8 to 15; high/harmful risk drinking: score ≥16). Baseline weekly unit consumption. Self-reported drinks consumed and purchased over the previous 7 days, used to calculate the number of alcohol units as a continuous variable. Free-text comments. Participants provided comments on the task, such as explaining their choice of drinks.

[END]
---
[1] Url: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1004193

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.