The current study presents expanded data from our recent study on how the free provision of non-alcoholic beverages impacted alcohol consumption [3, 5]. That study was conducted after approval by the ethics committee of the University of Tsukuba (Notification Number G299) and was registered as UMIN000047949 (Randomized controlled study on the impact of serving non-alcoholic beverages on alcohol consumption). We adopted a single-center, open-label, randomized, parallel-group design. Individuals who are at least 20 years old and drank on 4 or more days per week, with alcohol consumption of at least 40 g for men or 20 g for women on each of those days, were enrolled. This volume of alcohol consumption was defined as medium risk of alcohol-related health problems according to the definition of the World Health Organization (WHO) [6]. Although this is the first RCT study whether non-alcoholic beverage provisions on alcohol consumption, we could not ignore the possibility that non-alcoholic beverage provision exacerbated heavy drinking in the drinkers. Depending on the case, the intervention would exacerbate the symptoms of liver diseases which often occurred by heavy drinking. Therefore, it was necessary to exclude in advance those at risk so that even if adverse effects from the provision of non-alcoholic beverages were to occur, they would not cause irreversible adverse effects on the body. Considering these concerns and smooth implementation of research, the exclusion criteria were consumption of non-alcoholic beverages at least twice per month, past history of liver disease, current pregnancy or nursing, alcohol dependence, lack of consent for the use of LINE® (a messaging application adopted widely throughout Japan that can be used on personal computers or smartphones; LINE Corp., Shinjuku-ku, Tokyo, Japan), and inability to understand the study explanation or answer the pre-intervention online survey, both of which were written only in Japanese. Prior to the randomization, a 2-h face-to-face orientation session was held that included a well-experienced physician interview regarding alcohol dependence, based on the International Statistical Classification of Diseases and Related Health Problems-10 (ICD-10). We then excluded individuals with alcohol dependence because it has been suggested that their use of non-alcoholic beverages may enhance alcohol craving and stimulate their desire to drink, which may increase the risk of drinking relapse [4].

Written informed consent was obtained from those who passed the screening during the orientation session. During this briefing, participants were asked to complete a questionnaire regarding factors such as age, sex, race, marital status, highest level of education, employment status, household income, smoking history, and subjective view of health, as well as the Alcohol Quality of Life Scale (AQoLS) [7] and questions related to drinking, specifically the number of binge-drinking episodes within the past month and the items in the AUDIT [1]. In addition, height and body weight were measured and a saliva test was administered to assess the activity of genes related to alcohol metabolism, such as alcohol dehydrogenase 1B (ADH1B) and aldehyde dehydrogenase 2 (ALDH2) [8].

Following the briefing, simple randomization using a random number table was used to randomly allocate the participants to the non-alcoholic beverage provision (intervention) group or the control group [9]. Free non-alcoholic beverages were provided once every 4 weeks (three times in total). Each case included 24 350-mL bottles. Up to three cases were provided at a time. Thus, the participants in the intervention group received a maximum of 216 350 mL non-alcoholic beverages. In this study, we used non-alcoholic beverages that contained 0.00% alcohol by volume, had a taste similar to that of alcoholic beverages, and were designed or suggested for individuals aged 20 years or older as defined by the Japanese Alcoholic Beverage Advertising Review Committee. The beverages were selected separately by each participant depending on their preferences. Since the purpose of this study was to investigate whether increased availability of non-alcoholic beverages would change the amount of alcohol consumption, there were no stipulations regarding how to drink (amount or frequency). No other (psychological and/or educational) interventions were not conducted for both groups in this study. Participants in both groups were asked to record their consumption of alcoholic and non-alcoholic beverages in a calendar-format drinking diary every day, from 2 weeks before the start of the 12-week intervention to 8 weeks after its completion. All participants were also required to submit the drinking diary to the research staff every 4 weeks using LINE®. After the briefing, the study participants were contacted only by phone or via internet. At the end of the study, a gift card worth 10,000 yen (approximately 73.67 US dollars) was given to all participants as a reward. In addition, each participant in the control group received up to five cases of non-alcoholic beverages of their choice, but not received any intervention throughout the study period.

Although the amounts of alcoholic and non-alcoholic beverages consumed were obtained every 4 weeks, for practical purposes the mean amounts of alcoholic and non-alcoholic beverages consumed per 4 weeks were used in the analyses. We calculated the amount of alcohol consumption from the drinking diary data using the following formula: “consumption (mL) × alcohol concentration (%, v/v) × specific gravity (0.8)/100” [10]. We also calculated drinking frequency (drinking days per 4 weeks) and alcohol consumption on the day of drinking [5]. AUDIT is used as a screen for the full spectrum of risky drinking [11]. The WHO defines AUDIT scores of 7 or lower as a low risk of alcohol-related health problems, 8–15 as medium risk, and 16 or higher as high risk, but the relationship between AUDIT scores and alcohol-related problems is not universal and varies by ethnicity and race [1]. Previous studies in Japan have suggested that the AUDIT cutoff points for dangerous drinking is 8, harmful drinking is 12, and potential alcohol use disorder is 15 [12,13,14,15]. Thus, we divided the participants into four groups for secondary analysis according to their AUDIT scores. Briefly, we defined an AUDIT score of 7 or lower as group 1, 8–11 as group 2, 12–14 as group 3, and 15 or higher as group 4 (Fig. 1). The participants in group 4 may have had alcohol use disorder in terms of AUDIT score, but they were not diagnosed with alcohol dependence in physician interviews based on the ICD-10 prior to the start of this study. Therefore, this secondary analysis means that was an examination of the differences in classification based on AUDIT scores on the effect of non-alcoholic beverage provision on reducing alcohol consumption in heavy drinkers without alcohol dependence. In both the control and intervention groups, we expressed changes in alcohol consumption between the intervention and follow-up periods as percent changes from baseline levels [5]. Percent changes in non-alcoholic beverage consumption could not be calculated if participants did not consume non-alcoholic beverages at baseline. Hence, the consumption of non-alcoholic beverages was calculated based on the amount (mL) of non-alcoholic beverages consumed.

Furthermore, to examine whether alcohol consumption was reduced by replacing alcoholic beverages with non-alcoholic ones, the correlation between changes in alcohol consumption and the consumption of non-alcoholic beverages was analyzed.

The available data of all participants were included according to the original allocation in an intention-to-treat analysis. The unsubmitted data of the participants who dropped out were treated as missing data. The normality of the data was evaluated by the Kolmogorov–Smirnov test. Intergroup comparisons of baseline data in each AUDIT group were performed by the t-test in cases of normal distribution and by the Mann–Whitney U test in cases of non-normal distribution. The chi-square test or Fisher’s exact probability test was used to compare categorical variables. Two-way repeated measures analysis of variance (ANOVA) was used to analyze longitudinal changes in the consumption of alcoholic and non-alcoholic beverages within each AUDIT group in both control and intervention participants. The normality assumption was not met for all variables. However, the homogeneity of the variance of the data was confirmed for both variables using Levene’s test, and therefore the ANOVA results were accepted with reference to a preceding study [16]. The post hoc Bonferroni’s test was performed to analyze time course changes within a single group, as well as intergroup differences at individual time points. The Kruskal–Wallis test was used to analyze changes from baseline levels in the consumption of alcoholic and non-alcoholic beverages between intervention groups in the four AUDIT groups during the study, including the intervention periods. Correlations were evaluated using Spearman’s rank correlation coefficient. The significance level was 5%. The effect size was described as partial η squared (ηp2) and 0.01, 0.06, and 0.14 were considered as small, medium, and large effect sizes, respectively [17]. GraphPad Prism v. 9.0 (GraphPad Inc., La Jolla, CA, USA) and Stata 18 SE for Windows (Stata Corp., College Station, TX, USA) were used for all analyses.