Evaluating Alcohol Industry Action to Reduce the Harmful Use of Alcohol

Abstract

For the year 2013, it is estimated that alcohol was the world's sixth leading risk factor for disability-adjusted life years (DALYs), after high blood pressure, smoking, high body mass index, childhood undernutrition and high fasting plasma glucose (GBD 2013 Risk Factors Collaborators, 2015). The numbers of age-standardized alcohol-attributable deaths and DALYs were 11.1 and 13.6% higher in 2013 than in 1990, respectively. Reduction in alcohol consumption is essential to achieve global targets of reducing deaths from non-communicable diseases by 25% between 2010 and 2025 (Kontis et al., 2014), and WHO has set a target of reducing the harmful use of alcohol by 10% between 2010 and 2025 (WHO, 2014a,b), largely operationalized by measuring levels of per capita adult alcohol consumption. The evidence base for effective measures to reduce alcohol consumption is robust (see Anderson et al., 2009,, 2012,, 2013). It has been pointed out that improvements in alcohol-related health cannot be done by ministries of health alone, but require whole of government and whole of society approaches (see World Health Organization (WHO) publications: Kickbusch and Gleicher, 2012; Kickbusch and Behrendt, 2013), including action by the alcohol industry (OECD, 2015). Indeed, WHO's Global Strategy to reduce the harmful use of alcohol (WHO, 2010) states: ‘(p. 20) Economic operators in alcohol production and trade are important players in their role as developers, producers, distributors, marketers and sellers of alcoholic beverages. They are especially encouraged to consider effective ways to prevent and reduce harmful use of alcohol within their core roles mentioned above, including self-regulatory actions and initiatives.’ On 9 December 2015, AB InBev, the world's largest producer of beer, and soon likely to become even larger (Collin et al., 2015), launched its four ‘drinking goals’ 2015–2025, aiming to reduce the harmful use of alcohol (http://www.ab-inbev.com/social-responsibility/smart-drinking/global-goals.html). Given the future potential of the company producing one-third of the world's beer (with beer responsible for more than one-third of all recorded alcohol consumed in the world (WHO, 2014a,b)), the nature and potential impact of AB InBev's drinking goals are of paramount concern for global public health. The goals, one of which has, as its sub-text, an anticipated reduction of the average alcohol concentration of beer products by 10% between 2015 and 2025, may be seen as responding to WHO's Global Strategy. In its discussion of the place of the alcohol industry in reducing harmful alcohol use, OECD (2015) concluded that ‘limited evidence is available today in support of, or against, the effectiveness of business-led initiatives as contributors to the fight against harmful alcohol use.’ The report ended with: ‘There remains scope for broader and more incisive private sector actions, and for a more thorough and policy-relevant evaluation of the impacts of such actions.’ In this editorial, we do not discuss the merits or demerits of alcohol industry action to reduce the harmful use of alcohol, whether or not the industry is a ‘friend or foe’ to public health (Yach, 2014), or public–private partnerships in general to reduce the prevalence of non-communicable diseases (see Johnston and Finegood, 2015). Rather, taking each of the four goals in turn and the background information behind them, we propose the required evaluation needed to provide the evidence to answer this question: ‘By the year 2025, have AB InBev's goals positively or negatively impacted on reducing the harmful use of alcohol and subsequent public health’. This goal has as its sub-text an anticipated reduction of average alcohol by volume (ABV) of at least 10% by end of 2025. The evaluation of such industry pledges, such as the UK alcohol industry's ‘billion units pledge’ is not without problems (Department of Health, 2014; Holmes et al., 2015). From a public health point of view, it is important that the reduction of alcohol concentration of beer products by 10% between 2015 and 2025 is defined in a way that drinkers, and, in particular heavy drinkers, consume less alcohol and thus incur less harm (on the relationship between intake, reduction of intake and harm, see Rehm et al., 2010a; Rehm and Roerecke, 2013). Imagine the following scenario: nothing changes in the distribution of ABV for beer between 2015 and 2025, except that the share of alcohol-free beer is increased dramatically in, say, countries with predominant Islamic populations. From a public health point of view, this is not positive even though the global goal may be reached, because it is reached solely via a definition of beer products which include alcohol-free beer, and by changes in the markets for non-drinkers. Thus, from a public health perspective, the sub-text is more important. In order to evaluate the effect of the goal on drinking and thus on harm the following parameters must be controlled for: For the evaluation, the global market must be disaggregated into national or sub-national (e.g., for countries like India) markets, and the evaluation models must control for proportion of sex- and age-specific lifetime, and 12-month abstainers. These data are available, as there is regular monitoring of alcohol consumption for all countries as a result of the WHO Global Strategy for the reduction of the harmful use of alcohol (Poznyak et al., 2013; Shield et al., 2013; see also the regularly updated WHO Global Status Reports on Alcohol and Health, last version WHO, 2014a,b). Information about changes in proportion of different strength beers in national markets for any given jurisdiction must be corroborated by taxation and other data to the largest degree possible. Since these taxation data are routinely collected by governments, and specifically for alcohol summarized by the WHO, independent corroboration of industry statements seems possible. Already now, monitoring and surveillance of alcohol consumption and related harm is partly based on triangulation and cross-validation of industry and government data with information sharing between the parties (Poznyak et al., 2013). Confounding: while there are limitations to control for confounding in aggregate data (Morgenstern, 2008), some control is possible even at this level and even more when combined with individual-level data (see below), as alcohol exposure has one of the richest country-specific data sets and since sophisticated modelling strategies via pooled cross-sectional time series analyses (Beck and Katz, 1995) or meta-regression (Flaxman et al., 2015) are available. Most public health analyses are based on modelling, and the evaluation of the global 10% ABV reduction should deal with known confounders as achieved by the current Global Burden of Diseases, Injuries and Risk Factors models on risk factor impact (GBD 2013 Risk Factors Collaborators, 2015) or policy recommendations such as minimum pricing per gram of alcohol (Holmes et al., 2014). Modelling should include harm (based on the traditions of modelling alcohol-attributable burden of disease for WHO and the GBD studies—Rehm et al., 2010b). Verification of the ABV of beer products by independent toxicology agencies, based on a published protocol set-up in advance, comprising checks on random samples of beer products chosen from random samples of small-, medium- and large-size off-license retail outlets taken from random samples of jurisdictions from each of the countries of the world that has the highest market share of AB InBev beer sales. This is an action to be implemented by AB InBev. It includes increasing alcohol health literacy by 2025, which we discuss under Goal 4, social norms, since more than guidance labels alone is needed to increase alcohol health literacy. Providing better labelling information, including health warnings, on alcohol containers may increase awareness of the risks and content of products though may not reduce harmful consumption (Wilkinson et al., 2009; Wilkinson and Room, 2009; Kerr and Stockwell, 2012; Pettigrew et al., 2014; Knai et al., 2015; Osiowy et al., 2015). Such labelling has public support, and may play a role in shifting social norms to reduce harmful alcohol use when integrated with other broader social messaging campaigns (Thomas et al., 2014), and when implemented within broader alcohol policies (Louise et al., 2015). To avoid short comings of industry-designed labels (Pettigrew et al., 2015), the content and format of health warning text and labels, and information about the alcoholic content of beverage products, including information on numbers of grams of alcohol, calorie content and the presence of other health-important ingredients, should be set by an independent panel of health communication and labelling experts and epidemiologists. Industry implemented labelling should be evaluated in three ways: Verification of the presence of the label, based on a published protocol set-up in advance, comprising checks on random samples of beer products chosen from random samples of small-, medium- and large-size off-licence retail outlets taken from random samples of jurisdictions from each of the countries of the world that has the highest market share of AB InBev beer sales. Photographic evidence could be collected through smart phones and other hand held digital devices. Verification of the fidelity of the label in terms of format, text and alcohol content and other health-related information as judged against recommendations set by the independent panel (see Pettigrew et al., 2015). Consumer awareness and understanding of the content of labels, based on principles of health literacy (Institute of Medicine, 2004, 2009) should be collected through the cohort surveys of drinkers (see Goal 3). This is an action to be funded by AB InBev, but implemented at the city level, with experimental and control cities planned in six countries, implementing best practices globally by 2025. In the experimental cities, city-specific action plans should be developed to increase the impact of lowering of ABV in beer, and to reduce the harmful use of alcohol. These plans should be based on evidence-based prevention initiatives (for example, including, but not limited to the US Community Preventive Services Task Force Recommendations such as regulation of alcohol outlet density or commercial host liability; http://www.thecommunityguide.org/alcohol/index.html), including scale-up of primary healthcare-based screening and advice and treatment programmes for harmful alcohol use. Ideally, there should be randomized trials with control cities in each country with sufficient sample size both with respect to cities and individuals within cities (e.g., Shakeshaft et al., 2014) to test whether the interventions work. While this seems not possible given that such randomized trials have very high costs and there are not many examples of sufficiently powered community trials in the literature, a different design in combining national aggregate and individual level data could be used to achieve better control (Gmel et al., 2004). One of the main objectives of this part of the evaluation should be to develop a randomized trial based on evidence-based best practices by providing the necessary estimation of effect sizes and their variability between cultures. To evaluate the city actions, we suggest drawing representative samples from the adult population in the respective cities with an oversampling of heavier drinkers. Baseline drinking should be established before the reduction of ABV in beer and other city-based actions in both experimental and control cities, with yearly follow-ups covering the period of the interventions. This would allow an analysis of how the city actions and the industrial interventions to lower ABV affect individual drinking courses. To ensure low drop outs, digital devices to collect data should be used, and self-reports should be corroborated with biological markers in a sub-sample, e.g. by using transdermal alcohol devices (Greenfield et al., 2014). The main objectives of this part of the evaluation are: There is considerable evidence that primary healthcare-based screening and advice and treatment, programmes are effective and cost-effective in reducing harmful use of alcohol (O'Donnell et al., 2013; Rehm et al., 2013; OECD, 2015). The challenge is scale-up (Eccles et al., 2012; Keurhorst et al., 2015), as such programmes are poorly implemented (O'Donnell et al., 2013). A comprehensive package designed for scale-up at city level, embedded in broader community action (Milat et al., 2015), could include improvement in alcohol health literacy (Taggart et al., 2012), context aware m-health applications for self- and co-management (Kaner et al., 2015), and advice and treatment delivered by primary healthcare physicians based on, for example, definitions of heavy and very heavy drinking of the European Medicines Agency (2010) and the presence or the absence of end-organ damage (Rehm et al., 2015). To triangulate the individual cohort data with the aggregate-level data to test if the supply changes in ABV beer resulted in changes of drinking behaviour relevant for public health (Gmel et al., 2004). To identify and empirically demonstrate best practices to reduce harmful consumption of alcohol in the community which could be adopted to be rolled out to other cities and countries, including low income countries, for example in Africa. Evaluation should include: Case study approach to understand factors that affect the scaling process, theoretical-based approaches (e.g. learning theory, Eccles et al., 2012), and identifying mechanisms that lead to improved health outcomes (Shahin et al., 2014). Measures of provider activity to demonstrate whether or not more providers are giving more advice and treatment for case positives according to agreed protocols. Patient surveys to measure alcohol health literacy (see below), alcohol consumption and harmful alcohol use. Referral rates to secondary care to measure the impact on the healthcare system. Ab InBev commits to investing US$ one billion over 10 years to influence health literacy and social norms across its markets through dedicated social marketing campaigns and other programmes, with implementation by both AB InBev and independent others. We argue that a ring-fenced percentage of this money should be used to finance relevant implementation of the other goals, in particular, Goal 3. Health literacy is defined as ‘the degree to which individuals have the capacity to obtain, process, and understand basic health-related decisions’ (Institute of Medicine, 2004), with evidence that inadequate health literacy is associated with impaired health (Institute of Medicine, 2009). Although more than 100 hundred health literacy measures are available (Nguyen et al., 2015), we are not aware of any published measures of adult alcohol health literacy, and would recommend expert development of a short tool to measure needed skills and knowledge. Social norms refer to what most people typically do or approve of and are a communication phenomenon at the individual and collective level (Rimal and Lapinski, 2015). Alcohol-related social norms can be changed in favour of improved health (Hakulinen et al., 2015; Kubacki et al., 2015; Previte et al., 2015). Care has to be taken that social norms influencing programmes do not result in unintended stigma that leads to more harm (see Evans-Polce et al., 2015). We would suggest that the content and format of such campaigns and programmes should be set by an independent panel of alcohol health, alcohol policy and health communication experts. The impact of actions to influence social norms and individual behaviours to reduce harmful alcohol use should be evaluated in two ways: Verification of the fidelity of the programmes to influence social norms in terms of content and format, as judged against recommendations set by the independent panel. Changes in alcohol health literacy, alcohol consumption, alcohol consumption of social networks, perception of others' behaviours and awareness of social norms programmes (Rimal and Lapinski, 2015) should be collected through the cohort surveys of drinkers (see Goal 3). There are academic and other standards in methodology and norms for evaluation, which need to be fulfilled for evaluation to be credible (for example, see HM Treasury, 2011). We suggest some of these minimal standards above. Since AB InBev's Goals 1 and 3 are about population effects of policies, the evaluation should at least match the standard methodology of epidemiological modelling for risk factors, and the policy evaluation as established by the GBD (GBD 2013 Risk Factors Collaborators, 2015), OECD (2015) or the WHO (e.g., WHO CHOICE methodology: Tan-Torres Edejer et al., 2003; http://www.who.int/choice/cost-effectiveness/en/). Sometimes such standards seem to be fulfilled only once societies force industry to do so (e.g., the pharmaceutical industry is forced to comply with regulatory agencies such as European Medicines Agency or US Food and Drug Administration). For preventive activities, even though similar agencies have been proposed (Faggiano et al., 2014), we are far from regulation. As a consequence, a lot of the usefulness of the evaluation and the credibility of results will depend on the independence and the quality of the evaluator, and on the resources available. To assure best quality, evaluation should be done by experts in the field overseen by an independent steering committee, and results should undergo the full scrutiny of the peer review process of high impact journals (HM Treasury, 2011). In addition, the resources reserved for the evaluation should be scaled to programme costs, with common budget estimates ranging between 5 and 20% of programme costs (http://betterevaluation.org/plan/manage_evaluation/determine_resources). As a public health experiment, the AB InBev initiative might be a ‘last’ major public health action of its kind by a global risk company; meaning, if the evaluation methodology is unacceptable to the public health community, the action itself will be considered unacceptable. If this happens, it becomes less likely that another global risk company, including alcohol companies, will attempt public health actions in the foreseeable future. P.A. and J.R.: Drafting of the manuscript. This editorial did not receive any financial support. P.A. received reimbursement of the costs of his attendance to give a presentation on trends in research on health and alcohol to the Global Advisory Council of AB InBev in London, July 2015 (presentation available on request). No potential conflict of interest stated for J.R.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 16 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.