Inclusion criteria.

1. 20–64 years of age
2. At least 4 drinks per week on average
3. Positive screening for probable ALDH2*2 allele

Exclusion criteria.

1. Current alcohol use treatment
2. Lifetime cancer diagnosis
3. Pregnant

Study setting.

The study setting will consist of an internet panel sample in Japan.

Randomization and allocation concealment.

After informed consent is collected, participants will be randomized 1:1 and allocated to one of two groups after screening, informed consent, and baseline assessment. The computer-generated randomization sequence will be stratified by number of standard drinks. Neo Marketing, a third-party panel survey organization, will use a centralized computer program to conduct simple randomization. Once randomized, concealed allocation will be determined via centralized computer programming, and the treatment or sham control intervention groups will be contacted via Neo Marketing Inc LINE accounts. Because randomization and allocation conducted in a centralized computer, allocation concealment is guaranteed.

Blinding.

The principal investigator will be blind to randomization and group allocation until after data analysis. Participants will not be blind to group allocation. However, because there is a sham intervention for the control, participants will be unaware if their treatment is the primary point of investigation. The public title of the study is, “Brief Intervention to Promote Healthy Behavior: Randomized Controlled Trial” as listed in the Informed Consent Document. Care providers are automated educational videos and technically blinded. All outcome measures are self-reported. Thus, the outcome assessor is the participant and blind to the study hypothesis. Data analysts will be blind to randomization and group allocation, as raw data will be coded upon delivery from the third-party panel survey organization.

BIGRE group.

BIGRE consists of an animated educational video with a host medical doctor avatar voiced by a professional voice-over artist (S2 Table). It provides information, animated content, and graphical depictions of graphs, charts, and genetic processes. The components will emphasize that a) the participant’s screening indicates that they are likely an ALDH2*2 carrier, b) ALDH2*2 carriers show increased risk for specific cancers that increases with increased alcohol consumption, c) the association of ALDH2*2 with cancer risk is not deterministic, and d) non- or infrequent drinkers with ALDH2*2 do not show elevated risk for alcohol-related cancers. Research on esophageal cancer risk associated with alcohol use in people with the ALDH2*2 allele is wildly varied with relative risks (RR) ranging from 1.5 times the risk for all drinkers to 89 times the risk for heavy drinkers [10]. To avoid overstating risks, we will use very conservative estimates when explaining the risks to participants. Experimental event rate estimates for people with the ALDH2*2 allele is based on estimates taken from the summary literature meant to demonstrate the exponential dose response with increasing alcohol consumption [10] and a control event rate of 1% [31]. Compared to non-drinkers, we estimated risk to be 2 times greater for light drinkers with the ALDH2*2 allele, 2.4 times for moderate drinkers, and 4.4 times for heavy drinkers, noting that some research suggest the risks are much higher.

The video will last under 6 minutes (5:49) and incorporates the FRAMES approach [16,19–21]:

1. 1) Information: Education on the ALDH2 enzyme, the influence of ALDH2 genetic variations on alcohol metabolism, the relationship between possessing an ALDH2*2 allele and esophageal cancer risk, and the risks of continued alcohol use.
2. 2) Feedback: Informing the participant they are likely an ALDH2*2 carrier
3. 3) Responsibility: Stating that change is up to the participant to achieve the benefits of reducing alcohol use.
4. 4) Advice: Providing pointers on how to reduce use using motivational enhancement techniques
5. 5) Menu of options: Providing suggestions on how to develop a plan to reduce use and to start altering language to support sustain talk.
6. 6) Empathy: Delivery of genetic risk education by supporting in an understanding tone and affirming that this change is not easy, even a small change is great.
7. 7) Supporting Self-Efficacy: The narrator notes that decisions are up to the participant if they choose and provides options for self-help.

QoL sham control group.

The control group will receive a sham video that describes the importance of QoL and health, description of QoL concepts, ways to change health behaviors to improve QoL (Supplemental Table S3). The control sham-intervention consists of an animated educational video with a host medical doctor avatar voiced by a professional voice-over artist. It provides spoken information, and animated content. The video will last just under 6 minutes (5:38).

Primary outcome.

The primary outcome is mean endpoint past 4-week alcohol quantity at 3-months post-randomization. Alcohol quantity will be measured in number of standard drinks with the Daily Drinking Questionnaire (DDQ) [32]. Participants will be provided with a definition of a standard drink. Standard drinks are approximately: 250 ml beer (~5% alcohol), 90 ml wine/sake (~15% alcohol), 45 ml whiskey shot (40% alcohol). The DDQ asks, “Consider a typical week during the last 3 months. How much alcohol, on average (measured in number of drinks), do you drink on each day of a typical week?” The participants will then answer seven items representing each day of the week. For example, “On a typical Monday, I have ___ drinks.” For each day, participants enter a numerical response representing the number of alcoholic beverages consumed on an average day. The responses for a typical week are then summed and averaged over the past 30 days. The DDQ has been used in Asian populations with known ALDH2*2 alleles [23] and has demonstrated good internal validity (Cronbach’s α = 0.83) [33] adequate test-retest reliability [34] and criterion validity [35].

Secondary outcomes.

Secondary outcomes include the quantity in standard drinks at 1- and 2-month follow-ups; alcohol use in grams, improvement, severity, motivation to change, health knowledge retention, and quality of life at 1-, 2-month, and 3-month follow-ups post-randomization. Participant satisfaction will be measured after the intervention program has been delivered.

Anticipated risks and benefits for participants.

The interventions will consist of a screening (2 minutes), survey (15–20 minutes) and brief intervention (approximately 6 minutes) providing education on probable genetic information identified as an alcohol-related cancer risk. Thus, the intervention will therefore be considered as minimally invasive with no serious health risks expected. There are always possible psychological and time burdens associated with participation of health education and responding to questionnaires. Conversely, the only comparable study investigating alcohol use, ALDH2*2, and cancer risk reported no adverse events and high participant satisfaction, interest, and engagement [23]. The risks and benefits will be fully disclosed and explained to potential participants via informed consent. Neo Marketing would manage any participant complaints in line with their reporting procedures. Of which, participants agree to standard participant and private policy agreements.

Remuneration.

Participants will be compensated with Neo Marketing points that can be redeemed in various ways through the company website. Remuneration is aligned with Neo Marketing standards. Points are valued at ¥700 JPY for completing the screening, baseline assessment, and intervention. Participants will be compensated with points valued at ¥200 for completing each of the three follow-up assessments. Those who finish all assessments will receive a completion bonus of ¥1,000. In total, a final remuneration of ¥2,100 (~$13.30 USD) can be awarded in point equivalents.

Methods to decrease protocol deviations.

In this study, a one-time BI will be provided to study participants after randomization. The intervention will be administered through a fully automated computer program, making deviations only due to error in connectivity. Because the intervention consists of a short video, it is unlikely that the actual intervention differs from the method described in the protocol. After the intervention and during the study follow-up phase (3 months post-randomization), participants have no restrictions for concomitant care or co-interventions to reduce alcohol consumption, which is measured at each follow-up. Treatment cannot be discontinued or modified after the completion of the BI (5:49) and this intervention was deemed to be of low risk as it is a general population and of an educational nature. However, participants may request to be removed from the study at any time and for any reason.

Sample size.

Sample size analysis suggests 64 participants (32 per group) would give 80% power (α = 0.05, β = 0.2) to detect a standardized mean difference (SMD) of 0.7. Previous findings indicate an expected 50% reduction for the experimental condition [23]. Our proposed inclusion criteria requires a minimum of 14 standard drinks in the past 30 days. Thus, for our initial study a 7-drink difference is expected from the minimum drinkers, and we will assume large variability from a general population (SD = 10 drinks). Accounting for attrition, the available budget, multiple frequent follow-ups, and analytic methods, the target sample will be 100 participants (50 per group), giving 80% power to detect and SMD of 0.5.

Analyses.

The full analysis set (FAS) will include the largest population for analysis based on the intention-to-treat (ITT) principle, excluding participants who: Withdraw consent and disallow the use of their data after enrollment or have no efficacy-related data available. The safety analysis set (SAS) will include all randomized participants, including those who discontinue or drop out of the study. See the full statistical analysis plan (SAP) for further information (Supplemental Table S6).

Primary analysis.

The primary analysis will use the FAS population and assess the primary outcome (alcohol use quantity with the DDS at 3 months follow-up) with a generalized linear mixed model (GLMM) assuming a negative binomial distribution to address possible overdispersion in the count data [43]. In this model, the treatment group, time point (months 1, 2, and 3), and their interaction will be modeled as fixed effects, and random effects include both the random intercept and the slope for time account for individual variability in baseline outcome levels and temporal trends. The model will provide the estimated marginal means (EMMs) and their 95% confidence intervals (CIs) at 3-months.

A sensitivity analysis using the SAS population will be conducted for the primary outcome with only complete follow-up cases included. We will conduct subgroup analyses for the primary outcome to explore potential interactions between the group and the following pre-specified factors: age, sex, severity, health knowledge retention. These factors will be included as covariates in the model, along with their interaction terms with the intervention group. The p-value for each interaction term will be estimated.

Secondary analyses.

Binary secondary outcomes such as health knowledge retention will be analyzed using a GLMM with a binomial distribution. The model specification will follow the same structure as that used for the primary outcome analysis. For continuous secondary outcomes such as alcohol use in grams, readiness to change, and QoL, we will use mixed model for repeated measures (MMRM). These models will include intervention group, time points (months 1, 2, and 3), and the interaction between intervention group and time as covariates. If the sample size is too small to achieve convergence, simpler models will be applied.

Reporting.

In both GLMM and MMRM models missing data are assumed to be “missing at random” (MAR), and analyses incorporate all available repeated measures from all participants and account for within-subject correlations over time. Although missing values are not directly imputed, the observed data contribute to estimating parameters that implicitly inform inference in the presence of missing data. Secondary outcomes including participant satisfaction and safety will be summarized descriptively.

Alpha will be set a p < 0.05. Effect sizes will be represented with a standardized mean difference (SMD) using baseline SDs for continuous outcomes, and odds ratio, risk difference and number needed to treat (NNT) for binary outcomes. All analyses will be conducted using RStudio version 2024.12.0 Build 467.