Spreading alcohol brief interventions from health care to non-health care settings: Is it justified?

The development and evaluation of opportunistic alcohol brief interventions (ABIs) began in Scotland in the 1980s in general hospital wards (Chick, Lloyd, & Crombie, 1985) and primary health care (PHC: Heather, Campion, Neville, & MacCabe, 1987). From these beginnings, ABIs have spread in several senses of the word: geographically to most parts of the developed world and some parts of the developing world; from randomised controlled trials to attempts at widespread dissemination in practice (Babor et al., 2007; Nilsen & Holmquist, 2010); from face-to-face encounters to electronic forms of intervention (Cunningham, Khadjesari, Bewick, & Riper, 2010); from hospital wards and general medical practice to a variety of medical (Crawford et al., 2014; Gentilello et al., 1999; Gilinski, Swanson, & Power, 2011; Hulse & Tait, 2003; Smith, Hodgson, Bridgeman, & Shepherd, 2003) and other health care (Dhital et al., 2015; McAuley, Goodall, Ogden, Shepherd, & Cruikshank, 2011) settings; and from health care (HC) to a variety of non-health care (NHC) settings judged to be suitable for ABI implementation. It is the last of these that is the focus of this special section of Drugs: Education, Prevention and Policy.

As noted by Thom and colleagues (2016) in this issue, recent years have seen a considerable effort, backed by national guidance and advisory bodies (National Institute for Health and Care Excellence, 2010) and government (Scottish Government, 2013), to persuade a wide range of professional groups beyond HC to incorporate identification of hazardous and harmful drinkers and the offer of ABI for those so identified into routine practice. In this situation, it would clearly be advisable to learn as much as possible from 30 years experience of developing, evaluating and implementing ABI in HC settings and decide whether similar principles apply. At the same time, it is essential to recognise differences between HC and NHC settings that might impinge on the effectiveness of ABI and its acceptability to practitioners and their clients.

Such similarities and differences are clearly recognised and usefully described in the papers making up this special section (Herring et al., 2016; Sondhi, Birch, Lynch, Holloway, & Newbury-Birch, 2016; Thom, Herring, & Bayley, 2016). The classic triad of role legitimacy, role adequacy and role support, originally described by Shaw, Sprately, Cartwright, and Harwin (1978) nearly 40 years ago, together with the related concept of therapeutic commitment, seems as important in NHC as in HC settings for analysing difficulties in persuading practitioners to undertake alcohol work. In particular, while role legitimacy for ABI activity seems to have increased over the years among medical practitioners (Wilson et al., 2011), there is clearly a long way to go before it is seen as fully relevant to their work by many NHC professionals. So too, a lesson learned in trying to implement ABI in HC settings, that educating individual practitioners about ABI is ineffective without accompanying organisational changes (Babor & Higgins-Biddle, 2000), applies just as forcefully in NHC sectors. Familiar obstacles to implementation encountered in research in HC settings – lack of time, workload pressures, anxieties about offending clients, and a difficulty in getting past a preoccupation with “alcoholism”– emerge here too.

All three papers are concerned with training issues and report disappointing findings regarding the translation of training into practice. However, these findings provide valuable information about issues concerning what should count as training in the various settings discussed, what kinds of worker should be targeted in training, and how training impacts on practice. The need for more research on training on ABI in NHC settings is clearly indicated.

One important difference between HC and NHC settings concerns ethical issues. The point is made that, in for example PHC, it is taken for granted that norms regarding confidentiality and consent will be observed but that this may not be the case in NHC settings (Herring et al., 2016: p.xx). Without careful attention to ethical issues of this kind, there is a danger that relationships between professional workers and their clients may be damaged in NHC settings. Above all, the three papers here bear witness to the need to make sensitive adjustments to methods and procedures for screening and intervention in order to adapt to the specific contextual factors at play in each of the settings in which it is proposed to implement ABI.

But however valuable the insights and lessons learned from the research reported in this special section, it is assumed that spreading ABI from HC to NHC settings is a valid and beneficial policy. At the same time, it is recognised in these papers that the evidence for the effectiveness of ABI in NHC sectors generally is thin. The prior question to be asked in this editorial is whether or not this proposed extension of ABI implementation is a justified use of scarce resources.

How strong is the evidence for ABI in health care settings?

A straightforward and frequently-encountered view of the desirability of spreading ABI to NHC settings assumes that, since ABI in HC has been established to be effective, and despite the difficulties of dissemination, there is no reason why the same principles and methods of intervention should not now be applied to NHC settings; since ABI works in one kind of setting, why should it not work in the other? Unfortunately, this simple motivation for wider ABI implementation overlooks a number of difficulties. First, research failing to report a benefit of ABI in HC settings is frequently encountered in the literature. Given the small effect sizes to be expected from ABI, the report of null findings and failures to replicate previous findings should come as no surprise; nevertheless, null findings are often difficult to explain, lead sometimes to confused interpretation and are discouraging to researchers and policy-makers alike (Heather, 2014a).

It is often concluded from systematic reviews and meta-analyses that evidence in favour of the effectiveness of ABI is strongest in PHC (O’Donnell et al., 2013) but is inconsistent or limited in other medical settings, even those where enthusiasm for its potential effectiveness is perhaps strongest, like Accident and Emergency Departments (D’Onofrio & Degutis, 2002; Havard, Shakeshaft, & Sanson-Fisher, 2008) and general hospital wards (McQueen, Howe, Ballinger, & Godwin, 2015). Although implementation in these two major settings for ABI is arguably justified as evidence-based, this is not on the basis of an unequivocal or “robust” evidence-base.

In other medical settings, the evidence is even weaker, mainly because not enough research has been done clearly to determine effectiveness. A recent trial of ABI illustrates the point that its introduction in a novel setting may appear highly plausible on theoretical grounds but proves unwarranted when tested by careful research. Crawford and colleagues (2014) set out to test the effectiveness of ABI among patients of sexual health clinics (The Sexual Health and Excessive Alcohol: Randomised trial [SHEAR]). Reasons for being optimistic about the prospects for ABI in this setting are that surveys have repeatedly demonstrated that a high proportion of those attending sexual health clinics are drinking above recommended levels and that people who drink excessively are more likely to be diagnosed with a sexually transmitted infection. There was also evidence from previous research that ABI in sexual health clinics did reduce alcohol consumption (Lane, Proude, Conigrave, de Boer, & Haber, 2008). Yet the results of this large, rigorously conducted, pragmatic, randomised controlled trial failed to show clinically significant reductions in alcohol consumption among excessive drinkers or provide evidence of a cost-effective use of resources. There was a suggestion of a slight effect of ABI on consumption in the data but not enough for the investigators to recommend routine introduction of ABI in sexual health clinics. Now, it is possible to find explanations for why the expected benefit of ABI did not appear in this setting; it could be, for example, that the intervention was too brief (only 2–3 minutes) or that, rather than “universal” intervention among all patients, a more targeted approach among those drinking more heavily and/or with a history of sexually-acquired infections would be more likely to succeed. Whatever the merits of these suggestions, however, the main conclusion is that the effectiveness of ABI in this medical setting was not supported by high-quality research and therefore, in the view of the investigators, could not be recommended for implementation in routine practice. Should the same standards of evidence be applied to the introduction of ABI in NHC settings?

Efficacy or effectiveness?

Even in those settings where ABI can be shown to be efficacious in optimal research conditions, there is increasing doubt whether the evidence points to its effectiveness in naturalistic research conditions. The important distinction between efficacy and effectiveness research has attracted some attention recently in the field of ABI studies (Heather, 2014b; Saitz, 2014a). Suffice it to say here that both types of research are essential to support the introduction of ABI routinely in specific settings and that one should not proceed directly from tightly-controlled efficacy trials to widespread implementation without the intervening step of demonstrating that ABI works in the real-world of everyday practice. At the same time going straight to effectiveness trials without first establishing efficacy makes null findings from effectiveness trials uninterpretable.

Without evidence of effectiveness in real-world conditions, it can reasonably be argued that widespread implementation of ABI risks wasting precious resources that might be better expended in other areas of HC. Such an argument has indeed been mounted by Richard Saitz (2014b) who occupies the valuable role in the field of ABI research of championing the cause of rigorous evidence of effectiveness before widespread implementation of ABI should be considered. To anticipate a conclusion developed later in the editorial, research on ABI in NHC settings is very far from satisfying the level of scientific evidence of effectiveness necessary for practical implementation that Saitz demands.

Does ABI work at all?

As the foregoing suggests, there is now a serious debate over whether ABI can be said to be effective at all. On the other hand, there is also a view that to ask whether ABI as a whole is effective is to ask the wrong question. From this point of view, the mistake is to regard ABI as a single kind of intervention, or even two or three kinds, rather than an umbrella term for a whole family of interventions united by common principles and a few essential properties (Heather, 1995). The current crisis in evidence for ABI is reminiscent of the situation with regard to treatment proper in the late 1970s when the question was seriously addressed whether treatment for alcohol problems was effective (Edwards & Grant, 1980). The most pertinent response to this question derived from the work of Paul (1967) who insisted, with regard to the effects of psychotherapy, that “The question towards which all outcome research should ultimately be directed is the following: What treatment, by whom, is most effective for this individual with that specific problem, and under which set of circumstances?” (p.111). Correspondingly, the critical question for evaluating the effectiveness of ABI might be: What kind of brief intervention, delivered in what form, by what kind of professional, is most effective in reducing alcohol consumption and/or problems in what kind of excessive drinker, in what kind of setting and circumstances? Answers to such a question would be more complicated than those usually reported in research papers but arguably more realistic and practically useful. This perspective on effectiveness is obviously relevant to the wide variety of NHC settings in which ABI might be delivered.

The range of non-health care settings and the evidence to support implementing ABI among them

There is in theory a wide range of possibilities for the delivery of ABI: any setting where hazardous and harmful drinkers might be found, where professional workers in those settings might be persuaded to offer ABI and where their clients might benefit from it. The chief limitation seems to be the ingenuity of researchers and policy-makers in suggesting novel settings for ABI. The potential advantages of intervening in these settings are, first, that it may be targeted at population groups not easily reached in HC settings, especially in PHC, and, secondly, that it may be aimed at types of alcohol-related harm beyond health harms, e.g. family disruption, violence and other types of crime, workplace absenteeism, accidents and job performance, and financial difficulties. Prevention of harm to individuals may be extended more obviously in NHC settings to prevention of alcohol-related harm to families and communities.

The NHC setting where there has been most work is educational establishments, particularly universities and other higher education institutions. There has been a very large amount of research on the effects of ABI among university students, especially in the USA, with generally favourable conclusions from reviews (Carey, Scott-Sheldon, Carey, & DeMartini, 2007). As this is such a well-established area of research and implementation of ABI, nothing more will be said about it here except to note that it is an example of an application of ABI outside HC with firmly established credentials.

At another extreme are NHC settings for ABI that are entirely novel. These include housing settings, the topic of a paper in this special section (Herring et al., 2016) and job-seekers agencies (Freyer-Adam et al., 2014). Between these novel settings and well-studied educational establishments are a collection of settings that vary in the amount of attention that has been paid to them. Two prominent examples are the workplace and the social services sector and research here has been usefully reviewed by Schulte and colleagues (2014). ABI programmes in both settings are adversely affected by low participation and high drop-out rates but there are at least some RCTs in the workplace, mainly in occupational HC settings, although very-little research in social services. Further, in social services many clients with alcohol problems have complex needs in other areas and present varied and challenging problems for the delivery of ABI (Fitzgerald, Molloy, MacDonald, & McCambridge, 2015). The authors’ main conclusions were that, compared to HC, the reviewed settings are far more heterogeneous in terms of client groups, external conditions, and the involvement of other substance use disorders. Future research should attempt to describe these differences systematically and consider their implications for the deliverability, acceptance, and potential effectiveness of ABI for different target groups, organizational frameworks, and professionals.

Arguably the most potentially fruitful non-HC setting for ABI implementation is the criminal justice system (CJS: Graham, Parkes, McAuley, & Doi, 2012) – or, rather, settings plural, since the CJS includes a range of opportunities for ABI, e.g., police custody, magistrate court, probation service and the prison system itself (Sondhi et al., 2016). A high proportion of individuals in all these settings (> 50%) have been shown to have an alcohol use disorder or be hazardous drinkers and heavy drinking is a factor in a wide range of crimes but particularly in those involving violence (Flatley, Kershaw, Smith, Chaplin, & Moon, 2010). CJS settings for young offenders can be considered a separate set of opportunities for ABI implementation and here too the prevalence of alcohol use disorders is very high (Newbury-Birch et al., 2015).

International research on the effectiveness of ABI in CJS settings has recently been reviewed (Newbury-Birch et al., 2016), with the conclusion that there is very-little evidence of effectiveness at any stage of the CJS, owing mainly lack of follow-up data. There has been one high-quality cluster randomised trial of ABI in probation services in the UK (Newbury-Birch et al., 2014). This provided no evidence for the effectiveness of brief advice or brief lifestyle counselling in reducing alcohol consumption among offenders on probation, though there was a slight but statistically significant effect on reoffending rates.

Is evidence the only consideration for implementation policy?

There would be wide agreement among researchers that policy-makers often fail to base their decisions on research evidence and, indeed, that this is a ground for serious complaint. But, while few policy-makers would suggest that policy should ignore evidence entirely, is it the only consideration on which policy should be based? The work of Andrew Tannahill (2008) is highly relevant here. Tannahll addresses the extent to which health improvement action, including policies and activities on the ground, should be based on evidence and what the relative places should be of evidence, theory and ethics in health improvement decision-making. The area of heath promotion, although clearly not identical, is sufficiently close to that of ABI to be of relevance to present concerns.

Tannahill first argues that no evidence-base is without bias and that types of action for which evidence is strongest are not necessarily the most important for achieving population health gain. More basically, there will never be enough evidence for the effectiveness of everything we might wish to enact. Thus, if before implementing an intervention we insist that the evidence in its favour must be unequivocal, we might have to wait forever; decisions must always be made on the basis of an uncertain body of evidence. The degree of uncertainly we are prepared to accept will depend on circumstances, mainly on the urgency of the problem we wish to address. As a simple illustration of this point, during the recent ebola epidemic the standard of evidence taken to justify deployment of a newly-developed vaccine against the virus was lower than would be the case had the threat to human life posed by the epidemic been less grave (Henao-Restropo et al., 2015). The urgency of decisions about implementing ABI in novel NHC settings is clearly not of the same order but one can argue, given the high prevalence of alcohol-related harm in our society, that it has some urgency about it.

Based on these arguments and others, Tannahill proposes that we need a broader base for health improvement decision-making than evidence alone and that we should therefore think of evidence-informed health promotion action, not evidence-based. This, he says, “appropriately reflects the complexity and uncertainty of the real world of health improvement action—complexity and uncertainty that need to be recognized and accommodated rather than swept under the convenient carpet of oversimplification” (p.384). Tannahill then advances a decision-making framework for health promotion, public health and health improvement based on a triangle of ethical principles, theory and evidence. Tannahill’s ideas are much more elaborate and nuanced than it has been possible to convey in this short summary and the interested reader is encouraged to consult the original.

An example to illustrate Tannahill’s advice to decision-makers comes from a recent review of the effectiveness of emergency department-based brief interventions for individuals with substance-related problems, including alcohol problems, by the European Monitoring Centre for Drugs and Drug Addiction (2016). Although concerned with brief intervention in a medical setting and including drugs besides alcohol, the conclusions of this review are relevant to the implementation of ABI in non-health settings. The evidential conclusion of the review is that, while there are potential benefits of brief interventions in emergency departments, a definitive statement about effectiveness cannot be made for a number of reasons (q.v.). However, the authors go on to state that the feasibility of brief interventions delivered by emergency department personnel, the absence of reported adverse effects and the potential cost-effectiveness all suggest that brief interventions could be considered as integral to the training of emergency department HC staff.

This recommendation could be seen to follow from Tannahill’s suggested triangulation of theory, ethics and evidence: the theory comes from the observation that brief interventions offer “an important ‘window of opportunity’ in which to engage with people with substance use problems who might otherwise never receive any form of assessment, referral or intervention” (p.1) and the fact that brief interventions “can be delivered in a variety of settings, by a range of workers (after training) and in a short time frame” (p.1), all three factors combining to keep costs low. We might add that there are several theoretical perspectives that predict the effectiveness of ABI (Bandura, 1997).

The ethical part of the triangulation first invokes the basic principle of “do no harm” but also what has been termed “the extended precautionary principle” (Foxcroft, 2006, pp.14–18). Whereas the original precautionary principle might be expressed as, “Prohibiting an activity where there is scientific uncertainty of potential harm from the activity is justified”, the extended principle becomes, “Supporting an activity where there is scientific uncertainty of potential benefit from the activity may be justified” (Foxcroft, 2006, p.15). The relevance of the extended precautionary principle to the project of spreading ABI to NHC settings is obvious.

Conclusions

The question posed by this editorial is whether the spread of routine implementation of ABI from HC to NHC settings is justified. In other words, what are the circumstances under which policy-makers and researchers should attempt to implement ABIs in NHC settings on a routine basis and be sure that such implementation would not be a waste of precious resources that could more beneficially be deployed elsewhere? We reviewed arguments that this should not be solely a matter of research evidence. First, theory should be taken into account. The theory in this case is that there are good reasons to believe hazardous and harmful drinkers can benefit from ABI, that ABI is clearly effective in PHC, and that the introduction of ABI is an effective and cost-effective use of resources on a priori grounds. Secondly, ethical issues should be considered. Besides the over-riding principle of doing no harm, the main ethical principle is that, in the face of scientific uncertainty, the potential benefits from ABI should be supported (the extended precautionary principle).

These arguments might be a response to scientists like Richard Saitz who insist that there should be clear evidence of effectiveness from high-quality research in specific settings and real world conditions before the roll-out of ABI can be considered (Saitz, 2014b). They might too call into question the recommendation of the investigators on the SHEAR project (Crawford et al., 2014) that ABI should not be routinely implemented in sexual health clinics on the basis of their findings (see above). These are difficult matters of judgment and this editorial does not take a definite stance on them; rather, the issues on both sides have been raised for the reader’s consideration and in the interests of informed debate.

But what is not, in my view, a matter for debate is that, with the exception of educational establishments, evidence for the effectiveness of ABI in any non-HC setting is currently insufficient to justify the widespread implementation of ABI. Even accepting the merits of Tannahill’s arguments on the place of theory and ethics in evidence-informed policy, and accepting too the validity of the extended precautionary principle, evidence to support routine implementation in the workplace, social services and the CJS simply does not, at present, stack up. There have been well-designed and -conducted RCTs in the CJS and the workplace but these have provided no support for the effectiveness of ABI in these settings. In other settings, particularly those in the social services sector, there is no relevant evidence whatever. When more evidence becomes available, the arguments eloquently expressed by Tannahill (2008) and Foxcroft (2006) may come into play. In the current state of evidence, however, the cliché of the elephant in the room is hard to resist: everyone in the ABI field of research knows that there is no evidence to support routine implementation in NHC settings but, encouraged perhaps by pressure from government bodies and prestigious national guidance authorities, many proceed as though there were.

This negative conclusion on ABI implementation does not detract from the usefulness of the findings reported in the papers of this special section. However, this usefulness should be seen in terms of developing models of ABI for testing in research, including methods of training as well as screening and intervention itself. We have seen that the critical question for evaluating the effectiveness of ABI is what kind of brief intervention, delivered in what form, by what kind of professional, is most effective in reducing alcohol consumption and/or problems in what kind of excessive drinker, in what kind of setting and circumstances. It was also concluded from the three papers in this special section, and from much other research, that there is a need to adapt ABI to the contextual factors associated with each specific setting. Hence the need to carefully develop, pilot and refine special forms of ABI before they are tested in research. Furthermore, in future research on ABI in non-health settings, attention must be paid to the efficacy-effectiveness distinction and to the requirement of establishing efficacy in optimal research conditions before effectiveness is tested in real-world conditions (Heather, 2014b). Without prior demonstrations of efficacy, it is impossible to know, when real-world research produces null findings, whether ABI cannot work in principle in that setting (i.e. is inefficacious) or merely does not work in the particular conditions of testing, in which case further work is needed to discover the conditions under which it will work in practice. It also follows that “political pressures for quick results from premature effectiveness trials should be resisted” (Heather, 2014b, p.1).

Admittedly, funding for research just recommended may be hard to come by. Owing to the dominance of biomedical research funding and the paucity of funding for alcohol research generally, research on ABI in NHC settings may have a low priority among funders. Be that it as it may, if we wish to succeed in reducing alcohol-related harm via the widespread implementation of ABI in NHC settings, there is no alternative to the cautious and thorough approach to research evaluation advocated here.

Declaration of interest

The author reports no conflict of interest.