One of our hot topics – important issues which sometimes generate heated debate. In the absence of more or less inescapable impediments to heavy drinking like ramping up the price of cheap alcohol, widespread screening and brief advice have been the great hope for drink-related public health improvements. Patchy effectiveness and poor implementation have led that ambition to be questioned.
The advent of brief interventions represented a radical realignment away from aiming for abstinence among a (relatively) few ‘alcoholics’, to reducing harm and preventing more serious problems among the bulk of non-dependent heavy drinkers (1 2). Instead of narrow and intensive, the strategy was (and remains) to spread thin and wide, deploying easily-learnt interventions delivered in a few minutes by non-specialist staff.
After introducing the topic we examine:
Should the SIPS trials mean the end of ‘brief interventions’?; the results of the project mean to guide UK policy. Do the findings mean we should abandon what has come to be defined as a ‘brief intervention’ and revert to an unsophisticated warning about excessive drinking?
Expected health gains and cost-savings rest on unreal foundation; the fundamental issue – whether in real-world circumstances, brief interventions really do reduce consumption sufficiently to improve health.
The struggle for widespread implementation; even if they are effective, how much evidence is there that screening/brief intervention programmes can be implemented widely and well enough to improve health across an entire population?
UK policy (nearly) abandons universal screening and opts for targeting; how UK policy and practice has responded to these findings, focusing on the decision not to attempt universal screening and instead to incentivise or require a more targeted approach.
The targets were no longer to be drinkers forced to or who chose to seek help, but the far greater number whose sub-critical consumption generated no impetus for intervention. They were to be identified by biochemical tests, a few screening questions, or clinical signs, while coming into contact with services for other reasons. The resulting package is variously known as ‘screening and brief intervention’ or ‘identification and brief advice’, a package which could be replicated so widely that even if only a minority responded to a small degree, the result would be a worthwhile improvement in health across a population of drinkers. Individuals would benefit by being diverted from yet more risky drinking, but unlike treatment, the population was the main target, not the individual.
Among these drinkers there was no platform of existing serious harm from which to justifiably insist on abstinence, and no joint-enterprise between a patient acknowledging the need for help and the clinician offering that help. After its popularisation in studies among concerned drinkers or those already in treatment, motivational interviewing, with its focus on non-confrontationally generating motivation to change, came to be seen as an appropriate style for the interventions. Simpler advice-giving and didactic approaches are also common, and cognitive-behavioural principles were an early enhancement to these more atheoretical interventions.
Especially in the absence of population-level measures like increased price and diminished availability which materially obstruct continued heavy drinking, screening and brief intervention offers a relatively inexpensive strategy to generate widespread voluntary change among at-risk drinkers – usually defined as exceeding national drinking guidelines or scoring as risking alcohol-related harm on screening tests. From the start, primary care was seen as the key delivery vehicle, since it reached entire populations, not just those with identified serious illness. Partly because this setting has been a focus for the research, it also has the strongest research record. As the key setting in public health terms, primary care is the focus of this hot topic.
From their origins in research in the 1970s and 1980s, screening and brief intervention have come to form major planks in national public health and alcohol strategies, and their implementation has been promoted through national programmes backed by funding, training and implementation targets, now in the UK transitioning to the embedding of this work in routine medical practice. But just as their policy and practice significance has reached a peak, doubts have been building over whether their initial promise will be realised. The doubts broadly fall into two categories: whether real-world screening/brief intervention really does reduce consumption sufficiently to improve health; and whether these programmes can be implemented widely and well enough to improve health across an entire population.
Of these concerns, the most fundamental is whether brief interventions really do generate appreciable benefits
Of these issues, the most fundamental is whether in routine practice brief interventions generate reduced drinking leading to improved health. If this was considered proven, it might be worth continuing the struggle to find ways to implement the widespread screening needed to find the patients who would benefit, and to persuade practitioners to deliver the interventions. Even if that effort was only partially successful, patchy implementation does not necessarily mean a programme is worthless. In a health care sector which (like primary care) repeatedly sees the same patients, low rates of intervention can cumulate over the years to a programme which reaches a high proportion of patients. Intervening with a small minority of patients each year may still generate public health gains – but only if there are appreciable and lasting positive effects in the first place. On this count, opinions differ, but the evidence might justifiably be considered too weak to warrant the attempt to induce wider implementation.
Contrasting with the faith placed in brief interventions in national UK policy is the downbeat verdict of UK and US researchers who have themselves researched brief interventions. In 2017 they summed up (free source at time of writing) the evidence as they saw it: “After more than three decades of study in primary care, it now seems unlikely that brief interventions alone confer any population level benefit, and their ultimate public health impact will derive from working in concert with other effective alcohol policy measures.” What prompted this conclusion was the lack of convincing evidence that in real-world circumstances, brief interventions do reduce alcohol-related ill-health, coupled with the difficulty of persuading GPs to focus on not-very-heavy drinking when patients often have multiple lifestyle risk factors, and they and the doctors may be more concerned with here-and-now problems rather than the risk drinking will cause future harm.
The citation offered by the researchers for their prediction that impacts “will derive from working in concert with other effective alcohol policy” was a paper by Professor Nick Heather, whose work forms a thread through from the first trial in Dundee in 1985 of a brief intervention in general medical practice ( below) to the SIPS trials, whose results released in 2012 were to help determine an important element of government policy below.
Though appreciating the difficulties, in 2006 Professor Heather optimistically referred to the “steadily gathering momentum” of an “international movement dedicated to reducing alcohol-related harm by achieving the widespread, routine and enduring implementation of screening and brief intervention”. In contrast, by 2012, the year the unexpectedly negative SIPS findings emerged, he was arguing that “Widespread dissemination of [screening and brief intervention] without the implementation of alcohol control measures … would be unlikely on its own to result in public health benefits”.
He identified four requirements for such benefits to be generated but judged only one had been satisfied – evidence that brief intervention “reduces consumption to low-risk levels in some of those who receive it” – and even that has arguably not been demonstrated in real-world circumstances below. Meeting the remaining three requirements was, he wrote, “currently unlikely, either because they are difficult to achieve or because there is no evidence to support them”. A major gap was that “public health potential … is unlikely to be realized without the widespread deployment of universal screening,” something no national health care system had yet been able to achieve. The (for alcohol harm reduction) ideal scenario of drinking being asked about at every contact with a health professional, followed if indicated by help or advice to cut down, “might not be tolerated by the general public, not to mention the health professionals asked to deliver it, and might therefore be an electoral liability to any political party supporting it”.
Scepticism is apparent too among the doctors who would have to take on or champion this work. With sufficient incentives and sanctions, medical staff can be persuaded to screen and advise at very high rates, but this comes at a cost – not just in resources, but in what could have been done with the same time if practitioners and patients had been freer to decide their priorities. In 2012 an editorial in the Lancet medical journal caustically observed that “lecturing” patients about their lifestyles takes up time in the average 12-minute GP consultation which could have been used to more fully address why the patient attended in the first place, or some other condition or lifestyle threat – uses which might more cost-effectively improve health than a diversion to drinking. Perhaps, the Lancet hinted, brief interventions are themselves a ‘diversion’ from politically risky but more effective public health strategies, like eliminating cheap alcohol by setting a high minimum per unit price.
To a degree these doubts are inherent in the nature of brief interventions. Such minimal interventions, not very different from the normal practice against which they are compared, and conducted with unmotivated patients, could not be expected to produce large, easy-to-detect effects on drinking, and even more so on health outcomes only marginally affected (if at all) by small drinking reductions. The point about them is that they are undertaken ‘opportunistically’ during contacts made for other reasons and when clinicians and patients have other, more immediate priorities, inherent barriers to implementation. Demonstrating effectiveness in a widely implemented programme was always going to be difficult, and ‘not proven’ a likely verdict.
After optimism fuelled by some success in controlled trials, a rethink has been happening, which can be understood as the emergence of these inherent limitations as the evidence base expands into more real-world trials. Not least in impelling this rethink were the results of what was seen as a definitive test of relatively real-world screening and brief intervention in England, results government was waiting on to inform its stance on incentivising this work in primary care – the SIPS trials.
For the future of alcohol screening and brief intervention in Britain, studies do not get more important than those highlighted in an Effectiveness Bank bulletin issued in May 2012. It summarised findings released by the SIPS project, funded by the UK Department of Health in 2006 to evaluate different screening and brief intervention methods in England. These preliminary findings were followed by formal journal publications from the primary care, probation and emergency department arms of the study. In each case the results gave little reason to extend or elaborate intervention beyond an unsophisticated warning about excessive drinking to what has commonly come to be defined as a ‘brief intervention’.
“Thank you for taking part in this project. Your screening test result shows that you’re drinking alcohol above safe levels, which may be harmful to you. This leaflet describes the recommended levels for sensible drinking and the consequences for excessive drinking. Take time to read the leaflet. There are contact details on the back should you need further help or advice.”
Unexpectedly, across all three settings and whatever the intervention, a year later the proportion of risky drinkers had fallen by about the same amount. Most basic was a 30-second warning that the patient was drinking “above safe levels, which may be harmful to you”, plus an instruction to read the alcohol information booklet they were handed panel. For Britain’s National Institute for Health and Care Excellence, this would not count as a brief intervention at all, and the researchers intended it to be a relatively inactive ‘control’ condition against which the longer interventions could shine. But instead this terse warning captured the limelight; despite its simplicity and brevity, there was no evidence that supplementing it with more costly individualised brief interventions based on sophisticated counselling techniques and scientific understandings generated further benefits.
Another important finding was that implementation often required specialist support and patient throughput was low, suggesting difficulties in reaching a large proportion of the population. Incentivised with per-patient payments, most primary care practices managed to implement the interventions, but 10 of 24 needed help from research staff and nine did not recruit the targeted 31 patients over the 15 months of the trial. The average practice identified just two risky drinkers per month.
‘Do just the minimum,’ is the message austerity-hit commissioners might receive
Experts have stressed that the findings do not mean handing over an alcohol advice leaflet is all it takes. Screening plus the script of SIPS’s ‘control’ intervention incorporated assessment, strong feedback on that assessment, an implicit call to action to stop “excessive” drinking above “safe”, “recommended”, and “sensible” levels, and a reminder in the form of the leaflet – potentially effective ingredients. Interventions came after patients had been quizzed by researchers about their drinking and related problems and their readiness to do something about these, possibly thought-provoking interventions in themselves. Also, while what the interventions were intended to be is clear, what was actually done is not. Most fundamentally, analyses of this kind cannot establish that interventions are equivalent, only that the study has not shown they differ. There may actually be differences not picked up by the study which a further study could yet establish (1 2). That remains a possibility, but a combined analysis of screening test scores in the primary care and emergency department trials, and across all three SIPS trials, found that the most likely explanation was indeed that at the six-month follow-up the longer interventions were no more effective than the terse warning.
In the end these considerations seem to have mattered little. UK policy and expert opinion continue to advocate the more extended interventions the SIPS trials failed to vindicate, while others were convinced the results had shown these worthless, and that for brief interventions, very small, and very basic, really is beautiful. ‘Do just the minimum,’ is the message austerity-hit commissioners might receive, encouraged by the “Less is more” take on the study from the Department of Health’s Director of Health and Wellbeing. Other UK studies have also failed to find longer and more elaborate interventions improve outcomes; unfold supplementary text.
There is, however, some support (1 2) for extending not so much the duration of a single-session brief intervention, but delivering the intervention in at least two stages separated by a period of time, such as a few weeks, and the utility of longer interventions cannot yet be written off.
Among the possibilities left open by SIPS, one was particularly damaging. Rather than being equally effective, perhaps all the interventions – including the brief warning – were equally ineffective. Without a no-intervention comparator, there was no way of knowing whether the interventions played any part in the outcomes. If in the circumstances of the trials, advising drinkers was simply ineffective, it would explain why piling on yet more advice in the longer interventions made no difference.
This is not, however, an officially accepted interpretation. Underlying current policy in England is NHS England’s expectation that alcohol screening and brief intervention will save the health service money because less drinking will mean fewer alcohol-related illnesses and injuries, in turn leading to fewer hospital admissions. The result is an estimated net annual saving of £27 per briefly-intervened patient, and more if other health service costs are taken into account.
That estimate was based on a simulation model which has now been published. Analysts estimated that over the next 30 years, screening and advising newly registering primary care patients would cumulate to a saving of £215 million in alcohol-related provision such as hospital admissions, yet have cost the same services just £95 million. Patients who received the brief interventions would in total gain 32,000 extra years of life adjusted for quality (QUALYs). It meant that via these programmes the health service would improve health and at the same time save money – seemingly an unmissable bargain. Also simulated was screening every patient the next time they see their GP; it would cost more, but much more would be gained.
The new-registration screening strategy modelled by the analysis is directly relevant to current policy in England ( below), but even over ten years it would not reach most of the adult population, one reason why a leading researcher has said the resultant drinking reductions “may not be detectable on population-level measures”. The model’s assumptions of 100% implementation are likely to be wildly optimistic.
Even then, benefits and cost-savings for each individual patient might be worth trying to extend implementation for. However, both sides of the ‘save £27 per patient’ estimate are highly uncertain. To get close to inducing GP practices to screen every patient at registration or at the next visit, and then to counsel all those screening positive, might take substantial incentive payments and costly checks that the work has been done to an acceptable standard, neither seemingly costed into the calculations. Also not costed in was the value of what might have been done with the same time. The model’s assumption that without systematic screening and intervention, nothing effective would be done about drinking, maximised the calculated benefits, but is also unrealistic, especially for the heavier drinkers who incur most costs.
However, the most fundamental question is whether the benefits of brief interventions in terms of reduced drinking leading to improved health would materialise in routine practice. The analysis itself shows how critical these assumptions are. When intervention effectiveness was assumed to be about half that in the ‘best-estimate’ calculations, and drinking reductions sustained for three rather than seven years, the new-registration strategy would cost twice the health-cost savings it generated. Next we look for this ‘missing link’ – evidence of real-world effectiveness.
Screening and brief intervention programmes could only create health gains and cost-savings if they really did change drinking in ways which improved health. In respect of this key element in the calculations, the evidence relied on in the simulation model was not robust enough to be confident of the calculated benefits. It derived from an amalgamation of the results of primary care brief intervention trials which attempted to answer the crucial question of whether effects would transfer from tightly controlled research studies to routine practice. With the cachet of the Cochrane Collaboration behind it, this analysis became an influential foundation for cost-effectiveness calculations and policy.
Its answer to the question was encouraging: “Thus not only do brief interventions appear to be effective at reducing alcohol consumption in primary care patients, but this body of published work also seems to be relevant to the real world of clinical practice.” This conclusion rested largely on the finding that impacts in the more real-world trials did not significantly differ from those of trials further divorced from routine practice. However, though some were more real-world than others, few if any of the trials can be considered to have truly been conducted in real-world conditions. Notably (more in these background notes), ‘real-worldness’ was assessed only for the brief intervention phases of the trials. Before this came the selection of sites and of patients at those sites willing to participate, and the screening process which brief intervention relied on to select its recipients and to provide a basis for the conversation. Once patients were in the trials, more whittling down of the samples usually did or may have happened, further reducing confidence in the applicability of the findings to patients overall.
An example is the British trial assessed as most relevant to routine practice. It recruited only a quarter of the practices it approached (many said they had no time) and just over 1 in 10 contributed data to the analysis. With it seems at least 18 months and in some cases nearly three years in the study, nurses from the 49 practices in the trial screened just 498 patients. Even on the assumption of only a year to undertake this work, it meant the average practice screened fewer than one patient a month. Over the entire trial, fewer than three risky drinkers were identified per practice. With loss to follow-up, the 12-month results relied on just 78 of the roughly 131 patients approached by the nurses,
We simply do not know whether screening and brief intervention improves health
forming what must have been a tiny proportion of all the adult patients seen during the trial. Results from such a highly selected sample cannot be assumed to be representative of what would happen in a widely implemented national or local programme – and if they were, they were not encouraging below.
The most convincing demonstration that in the British primary care context brief interventions can modestly reduce drinking remains a study published in 1988, and again the sample was so highly selected that the results cannot be assumed indicative of what would happen in a widespread implementation. Of 4454 patients whose responses to a survey indicated their drinking had been excessive or caused them concern, the study ended with 748 who supplied data for the 12-month follow-up. They had been identified as risky drinkers not by the primary care practices, but by the research team. Perhaps significantly, the intervention occurred in two stages a month apart and was compared to nothing systematic at all – the control group received advice only if they asked for it or if blood tests indicated liver damage.
Based on simulation exercises, a later review suggested that primary care screening and brief intervention programmes are effective and cost-effective health improvers, but some of these simulations must have faced the same limitations in the ‘real-worldness’ of their source studies as the Cochrane analysis. When rather than extrapolating these from drinking reductions, the reviewers focused on evaluations which had actually measured health gains, they acknowledged that the results “do not allow any firm conclusions to be drawn”. Though we can expect these to flow from long-term reductions in drinking, on the critical question of whether health gains actually materialise, the answer is, we simply do not know.
It had been conducted in Dundee in 1985, led by the prominent brief intervention researcher ( above) who over two decades later was involved in the SIPS trials ( above), to which its results were remarkably similar. Whether screening had been followed by a warning from the doctor much like the SIPS warning ( above), or a brief intervention featuring assessment feedback, a self-help booklet, and further consultations, drinking reductions did not significantly differ. Unlike SIPS, another set of patients had been allocated to no advice on drinking at all, and they too did no worse than the other patients. For the researchers, “The results … provide little support for the hypothesis that the DRAMS scheme [the brief intervention] is superior to simple advice and to no intervention in helping problem drinkers seen in general practice to reduce alcohol consumption.” More than the 104 patients who completed initial assessments might have been needed for any small gains to register as statistically significant, but it seems this explanation is unlikely to apply to the SIPS trials above.
Predating SIPS by a few years, another British trial might also have given pause for thought. It was the one assessed by the Cochrane review ( above) as the UK trial most relevant to routine practice. After suffering from low recruitment to the trial and low rates of screening and brief intervention, it found no statistically significant evidence that a 5–10-minute brief intervention by primary care nurses was more effective than usual and, we can assume, much briefer unstructured advice, despite costing nearly £29 more per patient. Again, it could not definitively be said that the brief intervention had contributed nothing, just that there was no reliable evidence that it had. On some measures a larger and/or more fully followed up sample might have produced a statistically significant difference. For example, screening test scores fell modestly after brief intervention but not after usual advice, and at the six-month follow-up an average reduction in UK units of alcohol per week of 7.2 among brief-intervention patients contrasted with just 1.5 after usual advice. However, by 12 months – the intended follow-up point for the study – the results were virtually identical: a drop of about 7 UK units a week.
The two trials were relatively real-world, since both screening and intervention had to be done by routine primary care clinicians, unaided by research staff. For similar reasons, the SIPS trials themselves were billed as “pragmatic”, relatively real-world trials. On the basis of all three trials, the implications could range from abandoning post-screening intervention to reversing the drive for systematised brief interventions and letting GPs or nurses issue a simple health warning in their own words – or continuing the search for what constitutes a better way to address risky drinking.
One way to improve the fit between screening and brief interventions and the reality of general practice is to defocus from alcohol and address the multiple lifestyle risks often presented by patients, an option being pursued in Wales below. Whether this would work was tested by a Welsh trial of a brief intervention during which patients and GPs or practice nurses chose which risk-behaviours to focus on. Patients at risk due to their diet, lack of exercise, smoking, or drinking were asked to join the trial after screening, in this case by research staff.
A report published in 2013 compared outcomes in respect of these same risk-behaviours three months and a year later for patients in practices randomly allocated to be trained in the intervention versus those not trained. When the criterion was clinically relevant change on at least one of the behaviours, there was no statistically significant difference; at the 12-month follow-up, proportions improved were virtually identical, regardless of the training offered the practices. On drinking in particular, this was also the case, except that at 12 months the trend was for greater reductions in drinking among patients at untrained practices. There were some significant positive effects on exercise and diet, but of doubtful clinical importance. When health was directly assessed, none of the measures of lifestyle-related health significantly favoured trained practices.
In the Netherlands, it was worse – not just no effect, but a negative one. These emerged from a trial said to
The intervention increased the odds that patients would continue with hazardous or harmful drinking
“reflect the effects of such a programme when conducted in a naturalistic setting” – in other words, what happens when a close to real-world attempt is made to train and support GPs to identify and briefly intervene with risky drinkers.
Of 77 primary care practices, 40 had been randomly allocated to be offered extensive training and support to implement this programme. With this support, two years later 36% of their risky-drinking patients had reduced their consumption to a low-risk level. But in the remaining practices the corresponding figure was 47% – a statistically significant difference in favour of not trying to train and support GPs to offer screening and brief interventions: “Therefore, we concluded that the intervention did, in fact, increase the odds that patients would continue with hazardous or harmful drinking.”
Equivalence of outcomes would have been more understandable. All the patients, including those at non-supported practices, were mailed personal feedback and advice on their drinking. Given poor implementation of the training programme, this might have been enough to precipitate whatever changes in drinking were going to happen. But instead a negative effect emerged which within the context of the trial, could not be dismissed as a chance occurrence. One explanation might be that the generally incomplete training left doctors feeling less confident than before of their abilities to deliver alcohol-related care below.
Whatever the efficacy of brief interventions, they can only directly have an impact if experienced by the patient. Studies reviewed below show that without material or reputational and possibly career-affecting sanctions/incentives, implementation drives based on educating, persuading and supporting practitioners have reached just a minority of the intended patients. Strong sanctions and incentives can generate the desired activity, but may be costly and tempt services and practitioners to short-change quality and ‘game’ the system.
An example of the limitations of persuasion and support comes from the Netherlands. It started with an invitation to 2758 general practices to join the study which netted just 77 practices, an early sign of a lack of interest. Of the 40 randomly allocated to be offered training and support to implement alcohol screening and brief intervention, half did not complete what was considered a minimal programme, involving all the practice’s GPs attending a training session. The bottom line was that screening and intervention rates did not significantly improve in comparison to practices not offered support and training. A year after being offered support, GPs at the practices reported screening only 1 in 8 patients (fewer than before training and support) and counselled 1 in 20. Medical records told a similar story.
It was a surprising result: “Despite this relatively intensive programme aimed at provider, organization and patient, we did not find significant differences in change between the intervention and control groups … The degree of participation in the training sessions and visits was frugal.” On at least one measure, the training had been counterproductive; it left the doctors feeling less confident of delivering alcohol-related care than before they had been trained – possibly, speculated the researchers, because most GPs missed the sessions intended to bolster confidence. Perhaps for this reason, patients at practices offered training and support more often continued to drink at a risky level above.
The Netherlands was one of the European nations in the five-nation ODHIN trial (England was another) of how to increase the rate of screening and advice for risky drinking in primary care. At its most basic, some of the randomly allocated practices were merely given information on national safer drinking recommendations and asked to screen all adult patients. The rate of screening and of screened patients given advice both fell (in the case of screening, to just 3.5% of patients), resulting in a near 60% drop in the proportion of patients offered advice about their drinking. Supplementing this with a modest level of training and support was also followed by a near 30% drop in the rate of advice-giving down to just 1% of all patients, a fall due to a drop in the screening rate to 5.5% – not as bad as in the base case, but not better to a statistically significant degree. Other intervention-promoting tactics further increased advice-giving ( below), but not to a significant degree. However, there were significant impacts when the analysis assessed whether among all the combinations of intervention-promoters, certain elements seemed to make a difference. In this analysis, having training/support in the mix modestly seemed to help prevent deterioration in screening and advice-giving rates, but overall the results offered little evidence that on its own or allied with incentive payments, training and support can underpin widespread implementation.
Another trial established what in the US context it believed was an upper limit on the willingness of primary care doctors to conduct brief alcohol interventions (and then to refer all these patients to the research team) without this being mandated or there being direct financial or career consequences. The doctors were to undertake this work within a liberally timed one-hour health check with patients already identified by research staff as risky drinkers. Mechanisms to increase the brief-intervention rate included training, feedback, prompts to intervene attached to risky-drinking patients’ notes, and for each referral, thank-you notes, chocolates, and a contribution to the doctor’s continuing medical education fund. Still, just 39% of risky-drinking patients were referred to the researchers after presumably being offered a brief intervention by the doctors. More may have been advised without being referred, but even at the most prolific centre the referral rate was 51%.
Among the experiences which led Professor Nick Heather to doubt ( above) whether routinely widespread screening in general practice was feasible was Sweden’s “massive effort” to persuade medical practitioners to screen for risky drinking though a national education and information campaign. Despite this effort, in a population survey just 14% of adults who had seen a physician in the last 12 months recalled having been asked about their drinking, while in another survey, a fifth who had visited a health service in the past 12 months recalled talking about drinking, equivalent to 13% of the total adult population. If these conversations were the result of systematic screening, that was either imperfect or the results were often not used as a guide to intervention, because abstainers and moderate drinkers were as likely to have been spoken to as hazardous drinkers. Only when drinking reached “excessive” levels were conversations significantly more likely.
Some indication of the impacts was that out of a total sample of 3185, of whom 1988 had visited a health service in the past 12 months, just 24 the 950 hazardous/excessive drinkers said that as a result of talking there about their drinking they had cut down. Perhaps part of the reason was that few conversations included advice on how to cut down or probed the patient’s willingness to change. Even among hazardous and excessive drinkers, nearly half recalled the conversation as lasting less than a minute, associated in the study with a greatly reduced likelihood of patients saying it had prompted reductions. “Clearly, there is a gap between how brief intervention research is conducted and real-world practice as observed here,” said the researchers, a further indication that the results of trials cannot be extrapolated to what would happen in a widespread routine implementation.
Commenting on these findings, a team including researchers from the US Veterans Affairs system said they indicated the need for “accountability measures” to promote implementation – not just educating and persuading, but finding ways to hold staff and managers to account for doing this work. At the same time, they had to admit that in respect of screening, such measures had to date increased numbers at the cost of quality below.
Another northern European experience also reinforced Professor Heather’s doubts. It derived from Finland, where “despite the substantial effort to institutionalize” screening and brief intervention in the health system and public acceptance of this work, “the extent to which people are being asked about their alcohol consumption, and the extent to which heavy drinkers receive advice about it, is still lower than would be required to produce a population-level reduction of excessive drinking or alcohol-related harm”.
The study which prompted these comments had found that in the past 12 months a third of Finnish adults who had been in contact with a health service recalled being asked about their drinking. Even among heavy drinkers (similar to the excessive drinkers in the Swedish study described above), only about 4 in 10 had been asked about their drinking; of these, just half said the questions had been followed by advice. “Our results show that practice in health-care settings is slow to follow evidence-based recommendations,” said the researchers. As in Sweden, efforts to promote implementation had been based on education, training and support for professionals in primary and occupational care.
Experience in Finland and Sweden is particularly significant, because these countries top the league of the proportion of hazardous drinkers documented as having been advised about their drinking by a health care professional. If in those nations this is still too little to noticeably affect public health, then it falls even further short elsewhere.
Sanctions and incentives strong enough to work against the grain of primary care practice and induce widespread screening and intervention entail a risk on a different dimension – not quantity, but quality. The risk is that quality will be so poor that patients do not benefit, and even that the numbers needed to qualify for rewards and avoid sanctions will be recorded in the absence of meaningful intervention. Strong sanctions and incentives can generate what looks like the desired activity, but also tempt services and practitioners to short-change quality and ‘game’ the system.
The US Veterans Affairs (‘VA’) medical service for former military personnel offers what looks like a large-scale example of these processes. Its management and performance systems were able to create world-leading implementation rates, including an over 90% screening rate. Despite this achievement, evaluations found many risky drinkers were screened but not identified, and that impacts on drinking were minor or non-existent. Results from this early phase of the national VA system offered no encouragement to its continuation, but may change as the system beds in and is developed. Unfold supplementary text for more on these studies.
Elsewhere in the USA, ‘booster’ training and incorporating screening questions in simple-to-implement procedures which matter to the service (in this case, for their quality accreditation) has resulted in over 80% of patients being screened. Documented brief intervention rates too were elevated to over 60% by making it easy to record an intervention in a way which included the record in quality reviews of the clinician’s performance. Even in the unpromising environment of an emergency department, near universal screening and acceptable intervention rates can be achieved if the need to screen/intervene for each patient is visible to staff and made easy to meet, and staff supervision includes monitoring performance and correcting underperformance.
Already described above are the at best modest impacts of training and support in the five-nation ODHIN trial of how to increase the rate of screening and advice for risky drinking in primary care. One of the options tried in 2013 was supplementing basic information for GPs on national safer drinking recommendations plus a request to screen all adult patients, with per-patient payments for alcohol screening and advice. In England these payments were 6 Euros per screening and 25 per patient advised, up to a ceiling of 2200 Euros per practice. Relative to basic information, across the five nations payments significantly doubled the screening rate, feeding through to a non-significant doubling in the proportion of patients seen at the practices who were offered advice on their drinking. But still just 1 in 8 attendees were screened leading to about 2% being advised.
When the analysis assessed whether among all the combinations of intervention-promoters, certain elements seemed to make a difference, payments emerged as the most important component, one which allied with training and support nearly quadrupled the screening rate leading to a near-significant 2.3 times greater proportion of patients being advised. However, even this combination would across the entire sample have raised the proportion of risky drinkers who received advice from an estimated 3% before to only about 9% afterwards. Without knowing if there were any impacts on the patients, the most the researchers could say was that jurisdictions “could consider” promoting brief alcohol advice in primary care through training and guidance, financial and performance management arrangements, and strategic leadership.
The Veterans Affairs experience shows that incentives can divert recorded clinical practice in desired directions, but without professional commitment, the figures may be manipulated to record phantom interventions. This is believed to have happened at some primary care practices paid for intervening in England, and in Scotland among health staff working in non-primary care settings. Their accounts suggested that targets created “perverse incentives to maximise reporting of [alcohol brief intervention] delivery”.
More anecdotally, among general practices incentivised by payments, quality deficits have been observed in England (1 2) and in Scotland, where some practitioners were said to have adopted a relatively “perfunctory” approach to advising patients. A close, on-the-ground look behind the statistics conducted in the London borough of Haringey provided a revealing account of just how wide the theory-practice gap can get: unfold the supplementary text to take a look.
As elsewhere, trials conducted in Britain have found that without strong incentives, implementation rates are very poor. Since GPs generally failed to implement, attention turned to practice nurses, but they too screened just a handful of patients. This finding emerged from a British study which offered the nurses various training and support options (1 2). Out of 270 nurses approached, 212 agreed to use the screening/intervention programme for three months, but just 128 implemented it, screening 5541 patients and intervening with 1333. The most expensive implementation procedure (training plus continued support) resulted in the greatest number of interventions and was also the least costly per patient who received a brief intervention. Nevertheless, nurses offered this support typically screened just four patients a month and intervened with one every two months.
The biggest shortfall was in the screening rate; just 2% of patients seen by the nurses were screened. In contrast, of the 28% found to be at risk, an intervention was delivered to 64%. The screening shortfall was partly because universal screening was not attempted. Instead, most of the nurses who implemented the programme “did so opportunistically, that is when they had enough time to undertake the extra screening and intervention activity. Programme implementation also tended to occur in specific contexts such as new patient registrations, well person checks or in chronic disease monitoring clinics.” In other words, screening occurred usually only when the nurses had the time (typically five minutes) not just for this, but also for any ensuing intervention, or when it formed a natural component of broader health checks.
This study was one of several UK implementation trials with similar implications described in our background notes on a Dutch trial which found no impacts on drinking or related problems from a primary care brief intervention. In ‘debriefing’ sessions, the doctors expressed considerable discomfort with the intervention, fearing that doctor-patient rapport would be damaged by introducing drinking ‘artificially’ when the patient was attending for some other reason and without a naturally emerging clinical prompt. Despite the likelihood that the GPs who volunteered for the study were highly motivated, almost universally they said they would not carry on screening.
Early hopes for a public health impact from alcohol screening and brief intervention rested on the ideal of universal screening of adult primary care patients, an ambition still cautiously being recommended for Britain in 2012. But by at least 2004, research of the kind described above had convinced policymakers that universality lacked evidence of value and effectiveness.
Now the ambition in England and Scotland has been scaled back to screening new patients, during health checks, and/or ‘targeting’ screening only at patients thought to be at risk, while Wales has opted for discretionary and less structured interventions for lifestyle risks in general (see sections for each country below). Unless the targeting criteria perfectly identify risky drinkers (in which case screening is unnecessary), targeted screening is bound to dilute the hoped-for public health benefits of a universal programme. However, truly universal screening is hard to get close to. If targeting – favoured by primary care staff – encourages more complete implementation, it could find just as many or more risky drinkers.
The definitive rejection of universal screening (targeted screening was not considered) came in a judgement made in 2010 and last renewed in 2017 by the UK National Screening Committee, a body which advises on national screening campaigns for illnesses such as bowel and breast cancer. After commissioning a review of the evidence, they decided against universal screening for “alcohol misuse” in adults on the grounds that there was no screening test which met their criteria, and “a lack of evidence that a population screening programme would improve morbidity and mortality or would reduce social harm”. Studies to date had, they pointed out, generally tested brief interventions, not whether randomly allocating people to screening versus no screening led to reduced harm.
Their finding that no universal screening method has been proven reliable has been questioned, but there is no question about the lack of evidence for screening’s impact on population health. Even then, the committee could have recommended screening on the grounds that it allowed patients to make an informed choice about their drinking, knowing the risks revealed by the screening test – but they would have to have accepted that there was a test validated against a “gold standard” such as a biochemical marker of unhealthy drinking.
The route from screening nearly everyone to today’s less ambitious plans was punctuated by heated arguments over whether it was appropriate or feasible to ask primary care staff to question patients about their drinking, when this was not why they came to the surgery and there was no apparent reason to raise the issue. Controversy peaked when in 2003 a review in the British Medical Journal concluded that on average 1000 patients have to be screened to gain just two or three no longer drinking to excess. It was not necessarily that brief advice was ineffective, but that so few patients got to the point of receiving it. Critics hit back, but British studies ( above) confirmed that very low rates of screening and intervention were the norm.
In England this issue too was addressed by the SIPS project, whose other results have been described above. In primary care it found (1 2) that though a targeted method (screening newly registering patients or those whose complaints suggested excessive drinking) started with fewer eligible patients (1274 v. 1717), it ended up netting more patients who screened positive (461 v. 439) because it was more likely to reserve screening for patients who actually were risky drinkers. However, over a quarter of the patients who did not meet the targeting criteria turned out to score as risky drinkers; in a targeted programme, their drinking risks remaining unaddressed.
On the assumption that a universal strategy truly would be universally implemented, not missing risky drinkers may be the decisive consideration. Sweden offers an example of how it might work. Despite screening questions being bundled with research assessments, at nine practices high-coverage screening was achieved by making this routine on reception to the surgery. Passing the results to the clinical staff seeing the patients meant they raised the issue of drinking twice as often as at other practices left to identify risky drinking from the patient’s presentation and symptoms.
The main way to encourage wider implementation for priority public health issues in primary care is the Quality and Outcomes Framework, whose substantial financial incentives make it a major UK-wide driver of primary care practice. In 2012 the UK alcohol strategy said government was awaiting the results of the SIPS trials before deciding whether to incorporate alcohol screening and brief intervention into the framework. The results offered no support ( above) to brief intervention as usually defined, and the framework continued to embrace smoking but not drinking. How much of a difference including alcohol might have made can be gauged from the impact of what in 2011 was included in the framework, but only in respect of serious mental illness – payments for primary care practices which screen these patients for risky drinking.
Compared to the year 2000, following introduction of the incentives alcohol screening among people with schizophrenia and other psychoses increased dramatically in primary health care across the UK – an 839% rise. Among the remainder of the primary care caseload, the increase was just 62%. In the incentives era their drinking was recorded for 78% of the patients diagnosed as psychotic seen at the general practices in the study, virtually all in such a way that the record qualified the practice for the incentive payment. Per 1000 patients per year, the proportion for whom drinking was recorded had been about the same for psychotic and non-psychotic patients, but for the latter jumped after 2011 to reach 728 out of 1000 while it lagged at 184 among other patients. However, for just 588 (or 5%) of the 11,585 psychotic patients in the study had the practice recorded the use of a screening test.
Nevertheless, in its more limited targeted form, screening plus brief intervention remains an important policy strand in the UK. In England, directors of public health were expected to include it among attempts to address the population-wide determinants of ill health, in line with guidance from the National Institute for Health and Clinical Excellence on prevention and treatment of drinking problems and associated quality standards and guidance for commissioners. These documents’ insistence that commissioners and managers of NHS-commissioned services “must” ensure staff have enough time and resources to carry out screening and brief intervention work effectively seems a tall order, given the consistent appeal in the SIPS trials to workload pressures as a reason for incomplete implementation and the need for specialist support, despite the services having volunteered to join the studies. However, in primary care contractual requirements and extra payments have been used to reinforce the professional duty to take note of clinical guidance; how is the subject of the next sections.
Instead of being incentivised under in the Quality and Outcomes Framework, until April 2015 a scheme was available for practices to opt in to which qualified them for payments for screening and brief intervention as “enhanced” services over and above core contractual requirements. From April 2015 this work became absorbed into the core national primary care contract, requiring every contracted practice in England to screen all newly registering patients. No specific payment is attached; the main external pressure to implement are contract-compliance audits by commissioners.
Under the contract, screening entails asking the three or four questions of the FAST or AUDIT-C questionnaires, followed for positive-screen patients by the full ten questions of the AUDIT questionnaire. Based on the risk level revealed by the last step, patients should be offered brief advice, more extended counselling, or referral to specialist services – a stepping up contrary to findings in the SIPS trial that even for higher risk drinkers, brief advice or more extended counselling offer no extra benefits compared to a basic warning and leaflet. Each step of the process is to be recorded by GPs, records intended to be audited by local commissioners to ensure the required actions were completed. In theory this would enable the identification of practices which either fail to screen or record an abnormally high proportion of patients as not needing further assessment or advice.
Similar work has been incorporated in the NHS Health Check for older adults, intended to be repeated for each patient every five years. Uptake of the check has sometimes been below expectations, but each year it engages substantial minorities of patients (1 2). Local studies show that though it is meant to be included, drinking is far from universally addressed, but when it is, some patients say they have cut down; unfold supplementary text for details.
Studies conducted during the period (2008 to 2015) of the earlier enhanced services scheme suggest a new system was needed, though it remains to be seen whether the new arrangements will prove better. Under the pre-2015 national scheme, practices had to be offered the chance to be paid per newly registered patient screened for risky drinking, and were expected to follow this up as needed with advice and/or referral. Payments started at a modest £2.33 per screened patient with no further payments for follow-on care. Local areas could also offer different (usually more generous) incentive schemes.
Results of household surveys conducted in England in 2014 revealed that though about 60% of both risky drinkers and smokers had in the past 12 months seen their GPs, just 6.5% of risky drinkers recalled being invited to discuss their drinking. In contrast, smoking was addressed among 50.4% of smokers, nearly eight times as frequently. Risky drinkers were actually more likely to be advised about their smoking than about their drinking. The more serious the drinking risk, the greater the likelihood of the issue being raised; for each point on the AUDIT screening test above the risky drinking threshold, patients were 17% more likely to recall a brief intervention.
For the researchers, the main reason for the greater coverage of smoking interventions was likely to have been their incentivising through the substantial sums payable under the Quality and Outcomes Framework applicable to all practices, rather than the smaller payments in relation to drinking for practices which opted-in to an enhanced-service scheme: “[T]his study adds to the evidence suggesting that more substantial incentives are likely to be associated with greater delivery of brief intervention.”
Concerns that the small payments and limited remit of the national enhanced service scheme would not generate a high proportion of patients advised about their drinking were confirmed by a study of 16 general practices in northern England in 2011 to 2013. Practices signed up to the national scheme did screen just under half their newly registering patients, but these amounted to just 4% of all patients. At scheme practices, just under 1 in 10 of all adult patients were recorded as having received a brief alcohol intervention. All these proportions were higher in practices signed up to the national incentive scheme, but still well below those expected from a universal system, when a quarter of patients might qualify for brief intervention. Some practices rarely followed screening with recorded advice, others advised more than would be expected. In interviews GPs explained that they prioritised the more lucrative Quality and Outcomes Framework payments and had little faith that a brief intervention would work with patients who did not already want to change their drinking; some also felt that formalising this activity was incompatible with holistic, patient-centred care.
Even with incentive payments available, formal screening can be rare. A study investigated patients who had registered in 2007, 2008, or 2009 at a sample of UK general practices, and stayed at their new practice for at least a year. Practice records showed drinking had been addressed for a creditable 76% of the 382,609 patients. The bad news was that despite the possibility of enhanced-service incentive payments for use of the tests, just 9% of their records documented use of a validated screening test. Perhaps partly for that reason, recorded consumption was much lower than expected from face-to-face population surveys. Compared to survey findings, nearly twice as many patients had been recorded by their GPs as not drinking at all, and only 2% of men and 1% of women were recorded as higher-risk drinkers, compared to the expected 8% and 7% respectively.
The greater the incentive, the greater the temptation to record interventions which have not in fact happened or not to the specified standard, an issue which preoccupied attendees at a conference on alcohol screening and advice in England in 2015. On this count, the records on smoking interventions is not reassuring. When in England substantial payments for brief interventions for smoking were incorporated in the Quality and Outcomes Framework, a gap opened up between the interventions recorded by primary care practices and those recalled by patients. More patients recalled being advised, but not as many as there should have been according to the GPs’ records. One explanation is that GPs were more often recording advice which was not in fact given, or not given in such a way that it registered with patients.
Beyond primary care, screening – and in this case, universal screening – and brief advice are from 2017/18 being incentivised for inpatients by supplementary payments under the CQUIN (Commissioning for Quality and Innovation) system; unfold supplementary text for details.
With the emphasis on pragmatism rather than universality, Scotland’s national screening and brief intervention programme is based on guidelines for primary care which advocate screening only when the clinician has reason to suspect risky drinking, and which say any resulting intervention “should, whenever possible, relate to the patient’s presenting problem”. In 2009 its national alcohol policy prioritised screening and brief intervention, backed by a health service target for 2008/09–2010/11 to deliver 149,449 brief interventions in primary care, emergency departments and antenatal care, supported by dedicated funding. The target was exceeded and similar targets were set for the following years and again exceeded, topping 94,000 in both 2011/12 and 2012/13 and reaching 104,356 in 2013/14.
Though the numbers seem large, set in the context of what was in any event an estimated 111,200 primary care consultations for alcohol misuse during 2006/07, the initial target of around 50,000 a year across all three settings does not seem ambitious.
After 2013/14, numbers fell to 99,252 in 2014/15, still substantially exceeding targets in the last year before the programme was absorbed into NHS plans agreed between local health boards and the Scottish government, intended to “strengthen the continued aim of embedding [brief alcohol interventions] into core NHS business, ie that [brief alcohol interventions] are part of the day-to-day practice of health professionals and others, not an add-on to their role”.
In the first year (2015/16) of the new system, 97,245 alcohol brief interventions were recorded in Scotland, continuing the trend down since 2013/14, a trend most noticeable in the three priority settings where the programme started in 2008/09 – primary care, emergency departments and antenatal clinics. In those settings what was 84,444 brief interventions in 2013/14 had fallen by nearly a fifth to 67,861 in 2015/16, almost certainly due to a waning in primary care, which accounts for by far the largest number. Still, targets were again substantially exceeded.
The targets relate to meeting corresponding standards of service. For primary care the recommended brief intervention approach leans heavily on motivational interviewing (1 2). As in England, Scottish guidance ignores the apparent lesson of the SIPS trials that a basic warning and leaflet is as effective as longer and more sophisticated counselling above.
In 2011 an evaluation of Scotland’s programme found that “healthcare staff see the delivery of [brief alcohol interventions] as a worthwhile activity for NHS staff”. But of the three settings, only primary care practices really accepted the challenge: head-count financial incentives, the ability to seamlessly advise after screening, and more of a feeling that this was an appropriate activity, lifted their performance way above emergency departments and antenatal clinics. But even in GPs’ practices, it seems most risky drinking attendees were not screened, and the quality of the work was unclear. The barriers identified in an international review remained evident, particularly in antenatal and emergency care settings, which accounted for relatively small numbers of interventions. Competing priorities, not enough time, concerns over relationships with patients, feelings that this was not what you should be doing, all hampered implementation.
Nevertheless, a simulation model estimated that the Scottish national programme had made a small contribution to the decline in alcohol-related harm in Scotland, even if the brief interventions had been successful in only 15% of cases. To reach this estimate the report made some bold assumptions about the drinking and mortality reductions to be expected from the programme, based as other estimates have been on the Cochrane review which questionably concluded that trial findings would translate into routine practice above.
Scottish local health boards decide how to meet their brief intervention targets. Interviews with 13 GPs shortly after the programme was absorbed into local NHS plans make it clear that paying for this activity is an important incentive. All the GPs also said that the biggest barrier to an effective brief intervention was lack of time in a consultation lasting barely more than 10 minutes, which primarily has to address the complaints which brought the patient to the surgery. Time is in turn linked to money, because sufficient money can help pay for extra time.
A companion study also funded by Scottish Health Action on Alcohol Problems found clear evidence that financial incentives in primary care affected delivery. If screening was incentivised but not intervention, then screenings were abundant, yet led to few patients receiving a brief intervention. In the same case-study area, when the system changed to target delivering the intervention, an unusually high 41% of recorded screenings were followed by an intervention. Interviewees said money was not the only factor, but it was a key one.
Sharing the ambition elsewhere in the UK to embed this work in routine practice, Wales mounted a national campaign under the banner, “Have a word!”, an attempt to win hearts and minds through exhortation, information and training. Launched in May 2012, by August 2016, 13,308 individuals from a wide range of organisations had been trained, but after that the campaign ended as Public Health Wales decided to deliver brief alcohol interventions within an approach to multiple lifestyle risks such as smoking and inactivity. These interventions are to be embedded in practice through the Making Every Contact Count programme.
In England as well as Wales, the programme aims to extend the delivery of public health advice by minimal-cost training of non-specialist staff from a wide range of organisations in the basic skills of health promotion and disease prevention. Brief lifestyle behaviour change counselling is the chosen intervention model, a form of brief intervention. The focus is not on primary care, but as described in the Yorkshire and Humber region in England, on non-professionally qualified staff. However, the prospects for a multi-lifestyle intervention of this kind affecting drinking seem at the moment poor above.
In theory brief interventions have tremendous public health potential; consistently realising that potential in routine practice is today’s challenge, one yet to be convincingly overcome. See all our relevant analyses by running this hot topic search, or go to the first row (highlighted by the link) of the Alcohol Treatment Matrix to home in on the most important seminal and key studies, reviews and guidance.
Thanks for their comments on this entry in draft to Professor Nick Heather of Northumbria University in England, and Professor Richard Saitz of the Boston University School of Public Health in the USA. Commentators bear no responsibility for the text including the interpretations and any remaining errors.
Last revised 11 October 2017. First uploaded 01 May 2010
STUDY 2016 Improving the delivery of brief interventions for heavy drinking in primary health care: outcome results of the Optimizing Delivery of Health Care Intervention (ODHIN) five-country cluster randomized factorial trial