Impact of Alcohol Health Messages in a Real-world Online Retail Setting

Alcohol units will be calculated based on product-specific information on alcohol by volume (ABV) and volume purchased. The calculations will be made for the overall purchases and separately by type of beverage (beer, wine, spirits, other).

For overall purchases and separately by type of beverage (beer, wine, spirits, other, non-alcoholic beverages)

Will be assessed with question "Which of the following diseases and conditions does alcohol consumption increase the risk of? (select all that apply)", with the possible choices being [Cancer, Heart Disease , Liver Disease, Respiratory Disease, Alcohol use disorder, Injuries, Diabetes, Pancreatitis, Multiple sclerosis, Cerebral haemorrhage, Fetal alcohol spectrum disorder] displayed randomly, and "I don't know" and "None" as exclusive options. The question format and options were adapted from a previous study [18], but expanded with a longer list of possible outcomes described in the used health messages. Two outcomes will be derived from this question: (i) the correct identification of the conditions mentioned in the messages, with the possible score range being between 0 and 7 points, and (ii) the correct identification of cancer as condition that alcohol increases risk of.

Alcohol consumption behaviour will be based on question on whether they have foregone an alcoholic drink because they wanted to drink less and whether they have ever replaced alcoholic beverage with a non-alcoholic alternative because they wanted to drink less alcohol.

Perceived personal cancer risk will be measured with the question "If I consume alcohol on a regular basis, I am at greater risk of getting cancer." Perceived general health risk will be measured with the question "If I consume more alcohol, I am at greater risk for health harm". Both questions will be measured on 5-point Likert scale (Strongly Disagree to Strongly Agree).

Intention will be measured with the question "I intend to cut down on the number of alcohol units that I drink in the next month" on 5-point Likert scale (Strongly Disagree to Strongly Agree).

Emotional response will only be measured among those reporting to have noticed the warnings, "Thinking about messages I've seen as part of this study, I felt: disgusted / afraid / uncomfortable / worried / excited / pleased, with Likert answers scale from 1 to 5 (1 = Not at all, 3 = Moderately, and 5 = Very).

• Noticing, reading and recalling the warnings: Participants will be asked if they noticed health-related messages while using the Systembolaget app in the past two months. Participants will also be presented with a list of statements and asked to mark which of those have they been exposed to during participation in the study.
• Information seeking behaviour: If participants noticed the health messages, they will be asked if they took any actions such as searching for more information about health risks, discussing the information with others, or looking for ways to reduce their alcohol consumption.
• Message perceptions: Participants who noticed the health messages will also be asked to indicate their level of agreement with statements about the impact of the information on their alcohol consumption, purchasing behaviour, and their perception of the information's acceptability, novelty, relevance, and believability. Responses will be measured on a 5-point Likert scale.