Effects of ‘no safe level’ and ‘cutting down’ alcohol messages on problem recognition, defensive processing, and self-efficacy in heavy drinkers: A randomized experimental study.

Morris, J., Tattan-Birch, H. PsyArXiv, 31 Oct. 2025, https://doi.org/10.31234/osf.io/jze7p_v1
Abstract
Background: Public health groups have increasingly been adopting a no safe level of alcohol consumption message. We aimed to test how this message affects psychological outcomes relevant for alcohol use and how it compares with a gain-framed message emphasising the benefits of cutting down.
Methods: Adults who drank alcohol at risky or harmful levels (AUDIT-C ≥ 5; N =485) were randomised to view one of three messages: Control (n = 165), No safe level (n = 163), or Cutting down (n = 157). Outcomes were problem recognition, defensive processing (message derogation, risk recognition, fear), and self-efficacy. Linear regression models adjusted for age, sex, employment status, AUDIT-C score, and drinker identity. Moderation by drinking level was tested using interactions between message condition and AUDIT-C score.
Results: Compared with the control message, no safe level increased problem recognition (estimate = 0.22, 95% CI [0.03, 0.41]), elicited much stronger fear (0.92, [0.73, 1.11]), and was derogated more (0.89, [0.69, 1.09]). It also reduced self-efficacy (−0.23, [−0.43, −0.02]). Cutting down increased fear also modestly (0.30, [0.10, 0.49]) and lowered self-efficacy (−0.28, [−0.48, −0.08]) but had little effect on problem recognition (0.06, [−0.13, 0.26]) or derogation (0.13, [−0.07, 0.33]). Neither message had a clear effect on risk recognition. When compared directly, no safe level generated more fear than cutting down (estimate = 0.62, 95% CI [0.43, 0.82]) and was derogated more strongly (0.76, [0.56, 0.96]), but any effect on problem recognition was uncertain and likely smaller (estimate = 0.15, 95% CI [-0.04, 0.35]). There was no clear evidence that effects differed by AUDIT-C score (all interaction p>.13).
Conclusions: The no safe level message increased problem recognition relative to control, though not when compared with cutting down. Compared to cutting down, the no safe level message elicited much greater fear and message derogation. These mixed responses caution against assuming public health benefits of communicating ‘no safe level of alcohol consumption’ type messages.
ISFAR Summary
In a randomised experimental study, researchers examined how heavy drinkers responded to different alcohol health messages, as they constitute a priority population for harm reduction. Participants were shown either a message emphasising that any level of drinking carries health risks (‘no safe level’), a message highlighting the benefits of reducing alcohol consumption, or a control message neutral to alcohol consumption. The study found that while both alcohol messages increased recognition that drinking can be harmful, the ‘no safe level’ message also provoked stronger defensive reactions and lowered participants’ confidence in their ability to cut down. In contrast, messages focused on reducing consumption were better received and enhanced self-efficacy, which is a key predictor of successful behaviour change.
While raising awareness of risk is essential, it is not sufficient on its own. When messages feel overwhelming or absolute, heavy drinkers may disengage or think that change is impossible. Supportive messages that emphasise manageable reductions appear more effective in building confidence to act.
The findings are significant as governments and health organisations increasingly highlight alcohol’s connection to long-term health problems, including cancer. Although the evidence on the risks of cancers linked to small amounts of alcohol intake has been inconsistent across studies and interpretations, how this information is communicated is crucial, particularly for those at potentially higher risk. Morris and Tattan-Birch (2025) do not challenge the core health evidence, but their study highlights the importance of combining risk information with practical, achievable guidance. As their study focused on immediate psychological responses rather than long-term drinking behaviour, further research with a larger sample size is needed to understand how message framing influences real-world outcomes over time. Nevertheless, the results contribute to the growing evidence that effective alcohol communication should balance clarity about risks with encouragement, support, and practical actions. 
ISFAR Critique
Background
Healthcare systems face enormous challenges as the global population continues to grow and people tend to live longer on average. Throughout life, individuals are exposed to risks that ultimately lead to illness and death. Some of these risks are associated with a person’s behaviour and lifestyle. A healthy lifestyle is important because many diseases can be prevented. Some estimates even suggest that about 80% of chronic diseases and early deaths could be avoided through not smoking, staying physically active, and following a healthy diet. (Katz et al., 2018). Apart from difficulties in precisely quantifying these healthy behaviours across age groups and populations, communication about them is another complicating factor (Scholz et al., 2025).
Communication about alcohol consumption, considered a lifestyle factor, is especially complex because its health effects largely depend on the amount consumed, making scientific reports on the overall impact of alcohol on health potentially confusing. There is debate over public health messages derived from traditional epidemiological methods (Chen et al. 2025), innovative scientific approaches like Mendelian randomisation (Topiwala et al., 2025) and risk models . The latter two methods are more frequently employed by public health bodies such as the World Health Organization, resulting in an alcohol consumption message that is similar to or approaches a ‘no safe level’ message or close to a ‘no safe level’ message (Shield et al., 2024).
The paper by Morris and Tattan-Birch is interesting since the study explores how three public health messages affect psychological outcomes related to alcohol consumption. The three messages were a control message unrelated to alcohol, a no safe level message emphasising that any amount of alcohol is harmful, and a cutting down message emphasising benefits of reducing consumption rather than abstinence. The study posed two research questions. The first question focused on the effects of the three message conditions on four outcomes: problem recognition; defensive processing (e.g., message derogation and fear); risk recognition; and self-efficacy. The second question addressed the moderating effects of alcohol consumption levels on the four outcomes.
Critique
The study’s design involved comparing three groups of approximately 160 UK adults each, who received one of three similarly structured one-page scripts that provided information and virtual statements from an expert. Whereas such an experimental design would, in principle, examine a cause-and-effect relationship, the three groups needed to be similar in all relevant aspects and characteristics. All participants were heavy drinkers according to their AUDIT-C score (Bisschop et al., 2025), which ranged from 5 to 12, indicating increasing levels from high-risk alcohol consumption to dependence. Additionally, participants were screened for drinker identity centrality, a measure of how much an individual considers drinking to be an important part of their overall self-concept. Both AUDIT-C scores and drinker identity centrality did not differ between the three experimental groups, suggesting that the groups were at least comparable in some of their alcohol consumption behaviours and attitudes. Also, and more importantly, the group chosen was a relevant target for public health messages about alcohol consumption.
Participants were shown one-page scripts with a similar structure, including a virtual statement from an academic in the field of addictive behaviour sciences. The script, however, was relatively detailed and more comprehensive than the current public health message, which states that there is no safe level of alcohol consumption that does not affect health. The authors suggested that the length and detail of these script messages might have different effects compared to the current public health message. The authors did not elaborate on these effects, but one might expect that a more detailed explanation could have a greater impact on psychological outcomes as measured in this study.
Five different psychological outcomes were assessed in this study. The results show that the no safe level message increased problem recognition compared to the control, but not when compared to the cutting down message. The no safe level message, however, elicited much greater fear and message derogation than the cutting down message. Alcohol-related ‘self-efficacy’ was lower compared to the control message for both the no safe level and the cutting down messages. Negative ‘self-efficacy’ signifies a low confidence in a person’s ability to manage or abstain from alcohol consumption. The negative self-efficacy resulting from public health messages could indicate another issue related to high-risk alcohol consumption. Individuals with a high-risk consumption pattern may struggle to translate beliefs into actual behaviour. The authors state that it is well-known that increasing knowledge of alcohol-related risks has limited effectiveness in changing drinking behaviour. They rightly point out that alcohol consumers—including those drinking responsibly and those at high risk—value alcohol not only for its health effects but also for its relaxing, socialising, and enjoyable qualities. The positive, relaxing, socialising, and enjoyable aspects of alcohol consumption are experienced by most alcohol consumers, including those who drink lightly, those who drink in moderation, and even those who drink heavily. The positive aspects of alcohol consumption, such as relaxation, socialising, and enjoyment, may be relatively important to alcohol consumers. The results from the author’s research question two, the psychological effects of the two public health messages not differing by AUDIT-C score, may well fit this notion. Morris and Tattan-Birch (2025) concluded that the responses suggest caution against assuming public health benefits from messages stating there is no safe level of alcohol consumption.
This randomised experimental study offers valuable insights into how heavy drinkers, a key group in alcohol harm reduction efforts, perceive different alcohol risk message framings. The results indicate that while both “no safe level” and “cutting down helps” messages raise awareness of alcohol-related risks, they evoke significantly different psychological reactions regarding defensiveness and self-efficacy, with implications for the effectiveness of health communication.
A key strength of this study is its experimental design, which provides stronger causal inference than the observational approaches common in the alcohol communication literature. Random assignment to message conditions boosts confidence that differences in defensiveness and self-efficacy are due to message framing rather than underlying participant differences. Additionally, the focus on heavy drinkers enhances the study’s relevance, as this group is most likely to experience alcohol-related harm and may also be most sensitive to perceived threat or judgment in public health messaging.
The study’s findings align with well-known health behaviour models that highlight defensive processing and self-efficacy as key factors in behaviour change. Although causal mediation could not be confirmed due to unmeasured confounding, the pattern of results helps explain why high-threat, harm-focused messages might increase risk awareness but often fail to motivate change among heavy drinkers.
The policy relevance of the message comparison is another notable strength. ‘No safe level’ messaging has become increasingly prominent in alcohol guideline communication across several countries, including Australia and Canada, reflecting some decision-makers’ interpretations or biases regarding alcohol-related cancer risk, even in negligible to small amounts. This study offers empirically grounded evidence that such messaging, when presented in isolation, may produce unintended effects among heavy drinkers, especially by reducing confidence to cut down consumption.
Several limitations should, however, be acknowledged. First, the study evaluated immediate psychological responses rather than subsequent drinking behaviour or longer-term outcomes. While self-efficacy and defensive processing are well-established predictors of behaviour change, the lack of longitudinal follow-up restricts conclusions about sustained reductions in alcohol consumption. Second, all outcomes depended on self-report measures, which may be influenced by social desirability bias, especially in a population for whom alcohol use is personally significant and potentially stigmatised.
Third, the intervention involved brief, static messages presented in isolation. In real-world settings, alcohol guidelines are typically communicated through multiple channels and may include additional explanations, support, or advice from health professionals. The effects seen in this experimental environment might therefore differ from those produced by more comprehensive or supportive communication approaches. Similarly, the study did not consider the moderating roles of trust in health authorities or prior exposure to guideline messaging, both of which could influence defensiveness and acceptance.
Finally, heavy drinkers form a diverse group with different drinking habits, motivations, and readiness to change. The study did not examine whether message effects differed by demographic factors, drinking patterns, or stages of change, which could be important for tailoring future interventions.
Overall, the findings highlight the potential for harm-focused alcohol messages to cause resistance and decrease perceived ability to change when not combined with supportive, action-oriented guidance. The results emphasise the importance of integrating accurate risk information with messages that promote autonomy, capability, and confidence. For public health policy and practice, this suggests that communication strategies are likely to be more effective when they present risk alongside achievable pathways for reducing consumption, especially among individuals drinking at higher-risk levels.
Indeed, other public health messages might be more effective in promoting a healthy lifestyle related to alcohol intake. Such messaging could emphasise achievable reductions, practical strategies, and the immediate benefits of reduced consumption, such as better sleep, mood and wellbeing, rather than relying solely on harm-focused framing.
Specific comments
Forum member Ellison appreciated seeing the results of this study, but “believes that there are inherent problems in the ‘No Safe Limits’ approach. First, and not discussed in the paper, is that ‘No Safe Limits’ is an incomplete and actually not a true statement if one considers all health effects, both adverse and beneficial, of alcohol consumption. It is clear that regular light-to-moderate drinking with meals is not only usually ‘safe’ in terms of health effects, but actually has beneficial effects, including a considerably lower risk of cardiovascular disease, diabetes, and total mortality.
Advocates of the ‘No Safe Limits’ approach mainly base their argument on a potential increase in the risk of certain cancers (which is actually very minor for light-to-moderate drinkers) without considering the overall health effects. As an analogy, this is like questioning whether driving a motor car should be banned if it were appropriately labelled as having ‘No Safe Limits’. People understand that something with some risk can be acceptable if the overall benefits outweigh the dangers; they are not willing to revert to using a horse and cart, which also poses its own risks.
Instead of using ‘safe’ or ‘unsafe’ terms when advising drinkers, ‘total health effects’ could be a more suitable message, as it allows for not only advising against heavy or inappropriate drinking but also describing a pattern of drinking that may have overall health effects. We should encourage heavy drinkers to reduce their unhealthy drinking patterns and consider the beneficial health effects of adopting a more sensible approach to alcohol consumption.
Furthermore, as other Forum members mentioned, it is unfortunate that this study could not evaluate the effects of these one-time interventions on actual drinking habits; this would have been beneficial, even if only temporarily. Improved drinking behaviour remains the ultimate goal.”
Forum member Skovenborg states that “I agree with the ISFAR critique conclusion that harm-focused alcohol messages cause fear and resistance and decrease perceived ability to cut down heavy alcohol intake. However, the “No safe level” WHO mantra also demands attention for its inherent defect. The references quoted by Morris and Tattan-Birch (2025) (Burton & Sheron 2018; and Stockwell et al. 2024) do not substantiate “that the risk of some diseases, such as cancer, starts at any level of consumption”.
“When it comes to alcohol consumption, there is no safe amount that does not affect health”, WHO warn the public. “It doesn’t matter how much you drink – the risk to the drinker’s health starts from the first drop of any alcoholic beverage” (Anderson et al. 2023). The WHO statement clarifies: “Currently available evidence cannot indicate the existence of a threshold at which the carcinogenic effects of alcohol “switch on” and start to manifest in the human body. To identify a “safe” level of alcohol consumption, valid scientific evidence would need to demonstrate that at and below a certain level, there is no risk of illness or injury associated with alcohol consumption.”
However, science cannot absolutely prove a negative like “zero risk”, but it can provide overwhelming evidence against something, or establish risk levels, by showing something doesn’t happen within observable limits, or that its likelihood is extremely low. Science works by falsifying theories (proving them wrong) and building provisional, evidence-based understandings, rather than achieving absolute certainty. Scientific conclusions are always based on available evidence and are open to change with new data, meaning absolute “proof” doesn’t exist and WHO’s demand of “valid scientific evidence of no risk of illness or injury below a certain level of alcohol intake” is pseudoscientific misinformation.
A recent example of similar pseudoscientific misinformation was published by Centers for Disease Control and Prevention on 19 November 2025 in a statement about autism and vaccines: “The claim ‘vaccines do not cause autism’ is not an evidence-based claim because studies have not ruled out the possibility that infant vaccines cause autism”. The sound scientific conclusion is unchanged that there is no evidence of an association between vaccines and risk of autism, which was reaffirmed by The World Health Organization’s vaccine safety committee on 11 December 2025. It is high time that public health groups and the news media cease to adopt the misleading and fear-eliciting “no safe level of alcohol consumption” message.
Forum member de Gaetano agrees that the statement “ ’There is no safe dose of alcohol’ is misleading and pseudoscientific. I would ask if you know any ‘safe dose of aspirin’, even the 75-100 mg daily dose, that is, cardio-aspirin or baby aspirin. Both low-dose aspirin and low-dose alcohol are associated with significant health benefits, especially on major cardiovascular outcomes, including mortality. Both aspirin and alcohol, even at low doses, may be associated with some harm, such as an increased risk of bleeding for aspirin or of some cancers for alcohol, such as breast cancer. In medicine, the zero value does not exist, but the balance of benefits/risks should be the right guide to follow. “
Forum member Ursini, while he agrees with and endorses all the comments raised, also says that “my point, however, is different: we are no longer dealing with science. This form of neo-prohibitionism carries the unmistakable odour of a pseudo-religious attitude, one that leaves no room for scientific reasoning or evidence-based debate.”
Forum member Mattivi states that “this article falls outside my specific technical expertise, so I’ll limit myself to a few considerations on the difference between cultivating a culture of limits and imposing prohibitions. These may or may not be shared, but they are at least rooted in classical culture and reflected in this interesting paper. In the conclusions of this paper, the ‘no safe level’ message increased problem recognition related to control, though not when compared with “cutting down”. This suggests the validity of the Greek poet Hesiod’s work (c. 700 BC), which offered a moral framework still valid for everyday people: “Observe due measure; moderation is best in all things.” This directly advocates for finding the right balance (sophrosyne, σωφροσύνη), contrasting with absolute prohibitions. The limit is seen here as a positive guide, not just a negative restriction.
The difference between the ‘no safe level’ and ‘cutting down’ messages lies between the choice of an external, legal constraints (requiring rigid obedience) versus an internal, virtuous understanding of the appropriate action (requiring practical wisdom, and interpretation based on context, i.e. the cultivation of character and judgment).
‘No safe level – straight prohibition’ refers to an absolute, unyielding command or law that must be obeyed, and imposed by an external authority (the state, the health system). It assumes the inherent rightness or wrongness of an action regardless of the consequences (in the specific case also scientifically debatable). The emphasis is on simple obedience to a given rule ‘do not consume any alcohol’.
‘Cutting down – sense of the limit’ is an internal, ethical disposition, related to moderation, temperance, and prudence. It involves understanding the appropriate boundaries for human desire and action, derived from reason and from scientific evidence. The focus is not just on not doing something forbidden, but on cultivating in the population a critical character that naturally seeks balance and avoids the risks, consequently, raising awareness of the consequences of excess or inappropriate consumption.
In essence, the ‘no safe level’ message operates like a strict, external stop sign ‘do not consume any alcohol, ever’, while the ‘cutting down’ functions as an internal moral compass that guides one to find the right measure in all aspects of life, avoiding harmful extremes. It would be interesting if future studies also considered the temporal element, that is, not only the different effectiveness, but also the different duration over time of the behaviours adopted depending on the message to which one was exposed.”
Forum member Harding agrees with the comments made by Forum members on the misleading ‘no safe level of alcohol consumption’ message. “It is right that we are critical of the rationale for this.
The study concerns the efficacy of the ‘no safe level’ message when addressing alcohol misuse and addiction. It concludes that it is, in practice, unhelpful and is sceptical of its utility. They accept the validity of this message, but they are a psychiatrist and a statistician, and they don’t have the expertise we do, nor do they claim to. The two papers they cite in support of this message are Burton and Sheron (2028), which is an opinion piece rather than the results of a study, and Stockwell et al. (2024), which examines why some cohort studies find health benefits. Neither paper provides a rationale for the ‘no safe level’ message.
The paper by Burton and Sheron (2018) concerns the Global Burden of Diseases, Injuries and Risk Factors for 195 countries and territories from 1990 to 2016. It contains the telling sentence, ‘The level of consumption that minimises an individual’s risk is 0 g of ethanol per week (correct), largely driven by the fact that the estimated protective effects for ischaemic heart disease and diabetes in women are offset by monotonic associations with cancer (not correct). The associations of alcohol consumption with cancer do not come anywhere near passing the tests for causality, and of course, they don’t ‘offset’ the protective effect for individuals.
The paper by Stockwell et al. (2024) concludes, ‘Studies with life-time selection biases may create misleading positive health associations.’ True, but that does not mean that the associations are not causal.
It is worth listening to Dr Morris’s Alcohol ‘Problem’ Podcast, and in particular to his discussion with Professor Roy Baumeister. Both are in the business of finding effective measures to tackle alcohol misuse and addiction, and both are consumers of alcohol. They accept the veracity of the ‘No safe level’ message but point out that although cancer risk might start at any level of consumption, the risk for most people is very small, and there is not enough focus on the size of the effect. They say that overall alcohol does not cause big problems for most people over their lifetimes. They cite Professor David Spiegelhalter’s point that there is no safe level of driving.”
Forum member Djoussé reflects that “the term ‘no safe level’ is an odd one and I am not sure why people are always trying to reinvent the wheel as the terminology can apply to any FDA-approved drug, because almost any FDA-approved drug can be unsafe for some people, even at very low doses.”
References
Anderson, B.O., Berdzuli, N., Ilbawi, A., Kestel, D., Kluge, H.P., Krech, R., Mikkelsen, B., Neufeld, M., Poznyak, V., Rekve, D., Slama, S., Tello, J., Ferreira-Borges, C. (2023) Health and cancer risks associated with low levels of alcohol consumption. Lancet Public Health, 8(1):e6-e7. https://doi.org/10.1016/S2468-2667(22)00317-6.
Bisschop, J. M., de Jonge, H. J. M., Brunsveld-Reinders, A. H., van de Mheen, D. H., Mathijssen, J. J. P., & Rozema, A. D. (2025). Screening instruments to detect problematic alcohol use among adults in hospitals and their diagnostic test accuracy: A systematic review. Drug and Alcohol Review, 44(2), 505–531. https://doi.org/10.1111/dar.13987
Chen, Y., Yin, X., Wang, X., Zheng, X., Yang, X., Zhou, J., Shi, M., & Zhang, Y. (2025). Associations of alcohol drinking with incident dementia: a prospective study from the UK Biobank. European Journal of Epidemiology. https://doi.org/10.1007/s10654-025-01304-y
Katz, D. L., Frates, E. P., Bonnet, J. P., Gupta, S. K., Vartiainen, E., & Carmona, R. H. (2018). Lifestyle as Medicine: The Case for a True Health Initiative. American Journal of Health Promotion: AJHP, 32(6), 1452–1458. https://doi.org/10.1177/0890117117705949
Morris, J., & Tattan-Birch, H. (2025). Effects of “No Safe Level” and “Cutting Down” messages on problem recognition, defensive processing, and self-efficacy in heavy drinkers: a randomized experimental study. https://doi.org/10.31234/osf.io/jze7p_v1
Scholz, C., Chan, H.-Y., Hendriks, H., & van Reijmersdal, E. A. (2025). The social life of health-promotion messages: Message effects in complex information environments. Social Science & Medicine (1982), 387, 118658. https://doi.org/10.1016/j.socscimed.2025.118658
Shield, K. D., Paradis, C., Butt, P. R., Naimi, T., Sherk, A., Asbridge, M., Myran, D. T., Stockwell, T., Wells, S., Poole, N., Heatley, J., Hobin, E., Thompson, K., & Young, M. M. (2024). New perspectives on how to formulate alcohol drinking guidelines: Response to commentaries. Addiction (Abingdon, England), 119(1), 26–27. https://doi.org/10.1111/add.16402
Topiwala, A., Levey, D. F., Zhou, H., Deak, J. D., Adhikari, K., Ebmeier, K. P., Bell, S., Burgess, S., Nichols, T. E., Gaziano, M., Stein, M., & Gelernter, J. (2025). Alcohol use and risk of dementia in diverse populations: evidence from cohort, case-control and Mendelian randomisation approaches. BMJ Evidence-Based Medicine. https://doi.org/10.1136/bmjebm-2025-113913
Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Henk Hendriks, PhD, Independent consultant and partner of the Nutrition Consultants Cooperative, Netherlands
Creina Stockley, PhD, MBA, Independent consultant and Adjunct Senior Lecturer in the School of Agriculture, Food and Wine at the University of Adelaide, Australia
R. Curtis Ellison, MD, Section of Preventive Medicine/Epidemiology, Boston University School of Medicine, Boston, MA, USA
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
Fulvio Mattivi, MSc, Professor and Scientific Advisor, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy
Pierre-Louis Teissedre, PhD, Faculty of Oenology–ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Monika Christmann, PhD, Head of Institute, Department of Enology and Professorship for Enology, Hochschule Geisenheim University, Germany
Richard Harding, PhD, Formerly Head of Consumer Choice, Food Standards and Special Projects Division, Food Standards Agency, UK
Luc Djoussé, MD, DSc, Dept. of Medicine, Division of Aging, Brigham & Women’s Hospital and Harvard Medical School, Boston, MA, USA
Lynda Powell, MEd, PhD, Chair, Dept. of Preventive Medicine, Rush University Medical School, Chicago, IL, USA
Andy Waterhouse, PhD, Professor Emeritus of Enology, Department of Viticulture and Enology, University of California, Davis, CA, USA
Matilda Parente, MD, Consultant in Molecular Pathology/Genetics and Emerging Technologies, San Diego, CA, USA