For more on the study generally see these Findings analyses (1 2).
As the authors observed, among the limitations of the featured report are that "UKATT was not designed to investigate the issue of drinking goals, and results of the analysis were not hypothesized". What they are alluding to is the lesser weight attached to findings based on subdividing the participants in a study post hoc – that is, after the data has been collected. In advance of conducting the study, the UKATT team had itemised the propositions they would use the study to test. Whether patients initially aiming versus not aiming for abstinence would fare differently was not among them. In general Post-hoc subsample analyses of this kind are best seen as generating hypotheses for testing in a study specially designed for this purpose. The main problems are that they rob the results of the reassurance of the level playing field created by randomising patients to different treatments, they build on what may be chance variation in the effectiveness of the intervention between different subsamples, test effects not derived from the theory of how the intervention is supposed to work, and (there is no implication that this was a problem in this case) can capitalise on the fact that samples can be sub-sampled in any number of ways until one (perhaps purely by chance) results in a significant finding. As a result, "any conclusion of treatment efficacy (or lack thereof ) or safety based solely on exploratory subgroup analyses are unlikely to be accepted" (Lewis J.A. "Statistical principles for clinical trials (ICH E9): an introductory note on an international guideline." Statistics in Medicine: 1999, 18, p. 1903–1904. http://www3.interscience.wiley.com/journal/63000985/abstract?CRETRY=1&SRETRY=0. These risks are eliminated or reduced by specifying the subsamples in advance at the time the trial is designed but often this is not the case (Al-Marzouki S., Roberts I. "Selective reporting in clinical trials: analysis of trial protocols accepted by The Lancet." The Lancet: 2008, 372, 19 July, p. 201). such analyses are considered suitable for generating ideas for later testing in a study designed for this purpose, where the proposition to be tested is set out in advance along with the criteria for deciding whether it has been supported by the findings. Without this, samples in a study could in theory be sliced up in a multitude of possibly trivial ways (for example, by month of birth, colour of hair), as well as probably more salient dimensions such as age, sex, motivation, or treatment history, until, possibly purely by chance, one sub-division yields a significant finding. In UKATT, such concerns remain, but are lessened by the centrality of the issue (relation of drinking goals to outcomes) to the treatment of alcohol dependence. Though not specified in advance by the study, in this sense, the issue had been specified in advance as a priority concern for the field as a whole.
The fact that average drinks per drinking day differed little (and that at 12 months the difference did not attain statistical significance) between would-be abstainers and the rest of the sample is particularly indicative of that overall drinking outcomes were similar. To calculate this, the researchers assumed 0 drinks on days when patients abstained, meaning that the figures combined both sorts of desired outcomes – abstinence, and reduced drinking on drinking days. This begs the question of whether greater abstinence among would-be abstainers was counterbalanced by heavier drinking on days when they did drink. In UKATT, this was probably not the case; details below.
Though not itemised in the featured report, it is possible to calculate the intensity of drinking on days when patients did drink. At 12 months, those aiming for abstinence drank on average just over 31 UK units They drank on 100-54.82 = 45.18% of days, so on those days drinking averaged 14.12x100/45.18 = 31.25 units after other variables had been taken in to account. This equates to 250g alcohol. on drinking days, those not aiming for abstinence, just over 27 units. They drank on 100-43.49 = 56.51% of days so on those days drinking averaged 15.32x100/56.51 = 27.11 units after other variables had been taken in to account. This equates to 217g alcohol. While patients aiming for abstinence did on average drink slightly more on drinking days, rather than being a consequence of their goal choices, this seems to be a continuation of the pre-treatment difference between the two groups, when would-be abstainers also drank more intensely and to roughly the same degree.
Some of the outcome differences associated with an abstinence objective were statistically significant, but more striking was the similarity in the degree of success achieved, regardless of this initial objective, to the extent that choosing a different criterion for success would have reversed the study's conclusions; explanation below. Though the report sees abstinence-aiming patients as achieving the best results overall, this judgement depends on how one draws the line between success and not success. How the study drew it in some ways favoured the patients who sustained abstinence. Since these were more numerous among those initially aiming for abstinence, these patients too were favoured.
Specifically, the study's criterion required drinkers to be entirely free of difficulties associated with their drinking status, but did not impose the same requirement on abstinent patients; abstinence was considered successful by definition, regardless of how the patient was faring in the rest of their life. Yet in societies where not drinking at all is in statistical and social terms 'abnormal', and leisure and social activities often involve drinking and drinking venues, abstinence is not necessarily an entirely unproblematic choice. This may be particularly the case among some former heavy drinkers whose previous lives and social circles centred on drinking and drinkers, for whom alcohol may have masked or helped manage unpleasant emotions, or who find sustaining abstinence a constant battle. As with controlled drinkers, on balance they will almost always be substantially better off than when drinking heavily, but there may still be some abstinence- (rather than drinking-) related deficits and difficulties in their lives. For relevant contemporary findings, see for example this Norwegian study; for a classical account, We are grateful to Professor Nick Heather of Northumbria University for bringing this article to our attention. see this US study.
Since abstinence was more common among those who aimed for it, this criterion favoured these patients, but still it did so only slightly, a measure of how similar outcomes were, regardless of drinking goal. Were the criterion changed to, at a minimum, appreciable By at least a third. problem reduction, then non-abstinence aiming patients did slightly better to about the same degree that abstinence-aiming patients did according to the study's yardstick; details below.
The study set a high bar for designating a 'successful outcome': either total abstinence, or the total absence of drink-related problems as assessed in the study. By this yardstick, patients who were still drinking but had substantially reduced (yet not to zero) their tally of alcohol-related problems were considered unsuccessful. This type of incomplete but substantial improvement was more common among patients not aiming for abstinence. For example, at the 12-month follow-up, 37% (v. 24% for would-be abstainers) had reduced their problems by at least a third. It meant that slightly more (60% v. 54%) patients not aiming for abstinence achieved at least this degree of improvement, while slightly more (46% v. 40%) abstinence-aiming patients had unambiguously bad outcomes, either not improving at all or getting worse. Though these differences were not statistically significant, by 12 months, neither was the advantage abstinence patients had in successful outcomes as defined by the study. At this time, on the study's criterion, the difference between the groups was 7% (30% v. 23%), compared to 6% in the reverse direction using the less stringent criterion.
A less stringent criterion would also have reversed the conclusion that abstinence-aiming patients generally achieved successful outcomes by doing what they had intended from the start. As 'success' was defined in the study, this was the case. It could equally be argued that while just over a fifth of the would-be abstainers sustained abstinence at 12 months, about the same proportion managed to drink with greatly reduced By at least two-thirds. problems, and still more (about a third) were drinking with substantially reduced By at least a third. problems. Had this been the criterion for success, more abstainers would have been judged to have attained success while drinking than while not drinking.
As the authors commented, even accepting that an abstinence goal was associated with slightly better outcomes, there remains the issue of whether this was because some of the services or therapists were not as supportive of controlled drinking goals and may have undermined patients' confidence in their abilities to achieve these goals.
Several North American experts have recently reviewed studies on goal choice in alcohol dependence treatment, concerned that the dominant abstinence-orientation of their services might unnecessarily restrict access to treatment and limit what counts as success.
1 When reviewers searched studies published from 1977–2005 for the most consistent predictors of successful outcomes in alcohol treatment, choice of abstinence as a goal emerged as one of the top five. Significantly this verdict was based on studies which did not employ abstinence itself as their criterion for success. However, just five such studies were identified. One Though this (http://dx.doi.org/10.1111/j.1360-0443.1993.tb02042.x) was not a treatment study and does not seem to have identified abstinence as a goal. found no link between treatment goals and outcomes. Two others cited in the featured report but not in the review found either no relationship or a reverse one. In the latter, a Swedish study, abstinence objectives were followed by more intense drinking, perhaps because this goal was significantly more likely to have been adopted by physically dependent patients. A further US study found that goal preference at the start of treatment was unrelated to later drinking.
On the basis of this review and further studies it did not include, it seems that while opting for abstinence is commonly associated with better drinking outcomes, this is by no means universal. From the featured report and other studies ( below) it also seems that even when abstinence-aiming patients do end up drinking less, this does not always mean that their lives overall are more satisfactory than patients who did not opt for abstinence. The 'recovery' agenda in addiction treatment emphasising the overall wellbeing of the patient has most closely been associated with abstinence-oriented approaches. However, a focus on the patient's self-experienced quality of life now being advocated for treatment could as justifiably be seen as requiring flexibility in substance use goals.
2 In 2002 a US reviewer asked why studies investigating ways to promote controlled drinking have received less attention than studies of abstinence-oriented treatments. Partly she thought the answer lay in concern that acknowledging the feasibility of controlled drinking for previously dependent drinkers would undermine patients' commitment to abstinence, leading them to try controlled drinking solutions they were unable to sustain.
On the basis of the generally small and methodologically compromised studies which have been done, she concluded that reduced-risk drinking is a viable option for some problem and dependent drinkers, that abstinence and non-abstinence based treatments appear equally effective and safe (though clients with moderate to severe dependence may do better if they commit to abstinence), and that allowing goal choice may be more likely to promote a successful outcome than imposing a goal (in most cases, abstinence). Given no clear advantage for imposing or choosing an abstinence goal, services are more likely to attract and retain a broader range of problem drinkers if they allow choice not just initially but also in the light of experience in treatment. Often patients permitted to choose and work towards controlled drinking find they cannot manage and shift to abstinence as a goal; had this been a requirement of treatment entry, they might never have started treatment and never have committed to abstinence. The reverse transition has also been documented. To avoid premature drop-out one Swedish centre implemented regular three-monthly re-evaluations of patients' drinking goals, and found that as a result more patients started and stayed in treatment. The evaluation of this change found that over two years 44% of patients changed between abstinence and controlled drinking goals, with no apparent detriment to their recovery.
Among the clinical recommendations made in this review are that reduced risk-drinking goals should be specific, quantified, and explicitly agreed so patient and therapist know when they have been adhered to and when not. Goals short of abstinence are contraindicated by certain medical or psychological conditions exacerbated by continued drinking, or when medicines interact dangerously with alcohol, when the patient has repeatedly been unable to sustain reduced-risk drinking, or has a history of severe alcohol withdrawal symptoms. Such patients who nevertheless will not accept an abstinence goal may be persuaded to try this for a trial period, or to contract to revert to an abstinence goal if reduced drinking is not working out.
Concerns that admitting the feasibility of controlled drinking might destabilise patients' recoveries seem to be supported by studies which have found abstinence to be a more stable post-treatment drinking pattern than controlled drinking and less likely to transition to problem drinking (see for example 1 2). It has been argued that abstinence entails the same inability to sustain moderation as severe dependence, while moderation entails a trickier balancing act which repeatedly tests the individual's ability to choose to stop drinking. This understanding is consistent with abstinence being more stable after severe dependence of the kind which leads people to seek formal help and which requires treatment. In one study higher levels of dependence and alcohol intake and seeking formal help with drinking problems were associated with abstinent rather than non-problem drinking modes of recovery. So too was an index of the degree to which patients had saved more spare cash than they spent on drink during the last year of their problem drinking, taken to be indicative of the greater control later required to sustain moderation.
However, such studies do not mean that patients who opt for or, at least initially, achieve non-problem drinking would have done better had the service insisted on an abstinence goal; in this case many may simply not have entered or quickly dropped out of treatment altogether, failing to benefit at all. The degree to which insisting on an abstinence goal would exclude people who could benefit from treatment was particularly clear in a study of heavy drinking gay men in the USA who met diagnostic criteria for alcohol abuse or dependence. Just five out 89 identified abstinence as a drinking goal, and few worked toward abstinence even once they were involved in treatment sessions, yet as a whole the sample substantially benefited from entering treatment, achieving substantial drinking reductions.
3 In 2005 a Canadian review identified an understandable 'play safe' mentality among treatment staff, who are unwilling to advocate non-abstinence gaols for an individual in the absence of any definitive indication from the research of who would be able to sustain controlled drinking. The research does indicate that in general successful non-abstinent outcomes are associated with younger and female patients who are relatively socially integrated and psychologically stable, suffer less severe alcohol dependence, and strongly believe in their ability to moderate their drinking. However, the review observes that neither alone nor in combination are these indicators sufficiently closely associated with successful controlled drinking to be able to securely identify individuals. Nevertheless the author argued that stable controlled drinking following treatment for alcohol dependence is not uncommon, and cite the comments in Alcoholics Anonymous's Big Book that "A certain type of hard drinker" is able to moderate alcohol use. The review argues that allowing goal choice attracts people in to treatment who might then re-evaluate their goals in the light of experience. Since research finds little difference in how well patients do depending on their choice of goal, the balance of clinical advantage lies in widening access by permitting choice. Among the therapies the review found successful in promoting flexible drinking goals were motivational interviewing, behavioural self-control training, and behavioural contracting.
4 In 2007 US reviewers again presented the case for considering alternatives to abstinence as measures of treatment effectiveness for alcohol-dependent patients. They pointed out that most clinical trials have included only participants willing and able to initiate a short period of abstinence prior to treatment, limiting the applicability of the findings to the broader alcohol-dependent population. An abstinence yardstick also creates failures out of the three in four patients who (for example) in seven large multi-site studies were not continuously abstinent for a year after treatment. Outcomes short of abstinence are valuable, they argued, because alcohol-related social and medical damage increases with the heaviness of drinking and can often largely be avoided by low-risk drinking. They recommend that treatment goals be tailored to each patient's readiness to change in accordance with those goals, and may stretch from abstinence through to a reduction in heavy drinking. Continuing drinkers commonly fluctuate in the degree to which their drinking places them at risk, so for these patients in particular clinicians need to re-evaluate goals as treatment progresses, taking in to account fluctuations in the patient's readiness to accept and implement various goals.
The support patients receive in achieving controlled drinking or abstinence goals from staff, family and friends, the optimism they and their associates feel and express about being able to implement these goals, the availability of post-treatment relapse prevention options geared to these goals such as mutual aid groups, and therefore the sustainability of these recovery options, are all likely to be heavily dependent on the local drinking culture and the positions taken by treatment staff and services on the desirability and feasibility of these goals. In turn this environment is likely to affect the extent to which patients adopt controlled drinking or abstinence goals. As has been observed by reviewers commissioned by England's National Treatment Agency for Substance Misuse, moderation as a goal has become far more acceptable in the UK than for example in the USA, which originates the bulk of the research. For UK clinicians, this places a premium on studies conducted in Britain, where results might be quite different to those in the USA.
Before the featured study, there seem to have been four prior British studies, all conducted at NHS hospital inpatient alcohol treatment units. Two were conducted at the same Liverpool unit and another from a different unit in the city. Like the featured report, they all found that choice of goal was meaningful in the sense that successful outcomes generally took a corresponding form. The three Liverpool studies also agreed that overall success rates in eliminating risky drinking were similar whether or not abstinence was chosen. Another study found that opting for abstinence was more likely to be followed by non-problem drinking, but did not report whether lesser degrees of improvement were also more frequent among patients who aimed not to drink at all.
1 One study concerned the minority (37 out of 100) of patients who completed a six-week inpatient treatment programme and remained in contact for the following year. In the final week of treatment patients were expected to be explicit about their goals including choice of drinking pattern. Patients opting for controlled drinking were advised to remain abstinent for at least three months before attempting this, and abstinence was "prescribed" to 10 patients with specific medical contraindications to further drinking. Otherwise the choice was left to the patient. Among these, 15 chose abstinence (they were significantly older and heavier drinkers) and 12 controlled drinking goals.
By the end of the follow-up year, all but one of the patients who staff judged had successfully eliminated high-risk drinking (four abstinence-aimers and five aiming for controlled drinking) had done so by implementing their chosen goals. Outcomes for 11 of the 15 who had opted for abstinence were at best equivocal compared to seven of the 12 would-be controlled drinkers. In general, high proportions of both sets of patients had improved their lives on several dimensions and to roughly the same degree. As the authors commented, broadly the success rates were the same for both sets of patients, but success took different forms consonant with their original intentions.
2 Confirming this assessment, a later study from the same clinic found that taking other factors in to account, choosing abstinence as a goal was unrelated to a composite indicator of successful outcome embracing abstinence/controlled drinking and remaining in contact with the service. Unlike the prior study, this larger sample of 100 patients were not required to have completed the inpatient phase of treatment to be included in the study.
3 A different unit in Liverpool offered inpatient It also offered day care but nearly all the patients in the study had been inpatients. alcohol treatment which embraced controlled drinking as a valid and feasible goal for its heavily dependent Drinking most days and averaging around 200g of alcohol or 25 UK units a day. patients. During about a year after initial treatment, 139 who had remained in regular contact with the service could be classified as virtual abstainers (45 patients), controlled drinkers who nearly always drank within recommended limits (50 patients), or relapsed drinkers (44 patients) who regularly drank to excess (1 2). Controlled drinkers and abstainers generally (81–85%) considered their lives to have improved, though more of the abstainers had sustained this improvement with the aid of further treatment (40% v. 24%) including detoxification (14% v. 8%), psychotropic medication, and/or attending Alcoholics Anonymous meetings (18% v 0%). About half of the abstainers and controlled drinkers had occasionally lapsed.
Once again the patient's drinking goal at discharge Note that this was not necessarily the initial goal at treatment entry. from the inpatient unit was indicative of how they would actually sustain their recovery. Around 70% of the abstainers had intended to abstain but only 34% of the controlled drinkers; among the relapsed patients, 42% had aimed to abstain. From these figures it can be calculated that of the roughly 67 patients who left the unit aiming to be abstinent, nearly half (48%) achieved this and the remainder were almost evenly split between controlled drinkers (25%) and relapsers (27%). Of those not intending to totally avoid drinking, again nearly half (46%) achieved this, about a fifth were abstinent (18%), and just over a third (36%) relapsed according to the study's criterion, though some of these may still have achieved the goals they set themselves. Given this, and also given the slight 9% difference between the relapse rates of would-be abstainers and the rest of the patients, the study is best seen as indicating that in a free choice situation and after a few weeks of treatment, whichever choice patients make is followed by roughly equal degrees of success.
4 Finally, a fourth British study concerned patients admitted to a specialist alcohol unit in Northampton and offered subsequent day care. A year later just under half the 189 (out of 212) patients who could be re-assessed were either virtually abstinent, or drinking in a non-dependent manner without problems and only a handful of episodes of intoxication, the study's criteria for a successful outcome. Which type of outcome was achieved was significantly related to whether the patient had initially chosen abstinence as a treatment goal. Of the 135 who had, 56% were doing well according to these criteria compared to just 14% of the 35 who had opted for controlled drinking. The link with initial goal remained significant after other variables like severity of dependence at intake and social drinking patterns had been taken in to account.
However, in this study half the patients considered to have been unsuccessful had nevertheless substantially reduced By at least a third. their drinking and related problems and severity of dependence. Whether (as in the featured study) these partial successes were more common among patients opting for controlled drinking was not reported. Neither was how many patients who had selected abstinence achieved success that way as opposed to controlled, non-problem drinking.
1 Of the non-UK studies, a Norwegian study illustrated that while an abstinence goal may be (in this case, weakly) related to lower post-treatment drinking, that does not necessarily mean these patients are also happier with their current drinking pattern or have better lives overall – a disjunction echoed in the featured report and in another British study detailed above.
In Norway most interest centres on the outpatient alcohol treatment sample who were allowed to make their own choices of treatment goals. Just a quarter chose abstinence; most of the remainder wanted to curb their drinking, though some had non-specific goals or were not aiming to change. Placing these three choices on a continuum from abstinence at one end to no (specific) change at the other, the further along this continuum patients were, the more they drank in the next 21 months. However, the relationship was weak and entirely absent among an inpatient sample who had been advised to remain abstinent. Even among the outpatients, drinking goal was almost entirely unrelated to a
composite measure
This was a combination of:
• average alcohol consumption throughout the follow-up period;
• degree of satisfaction with their drinking pattern;
• number of readmissions for alcohol treatment during the follow-up period;
• whether stably housed;
• stability of relationship with sexual partner;
• whether in permanent employment.
reflecting the patient's drinking but also their satisfaction with this and the stability of fundamental aspects of their life.
2 The Norwegian study raises the issue of whether patients who according the study's criteria have 'failed' may nevertheless have achieved their own objectives and be satisfied with the outcome. In the latest study to test this proposition, alcohol-dependent women in the USA were recruited in to a trial of abstinence-oriented cognitive–behavioural therapies delivered either individually, or as a couples therapy to the patient with her male partner. Of the 102 women who started treatment, just 57 did so with clear preferences for their future drinking and provided sufficient outcome data to test whether these had been achieved. Of these, 35 preferred Though the nature of the treatment programme meant that all had explicitly agreed to aim for abstinence. abstinence, 22 controlled drinking. Patients who preferred abstinence were on average heavier drinkers, but this potential influence on later drinking was statistically 'evened out' in the analysis. Neither in terms of days abstinent nor days of heavy Over three US drinks in a day equivalent to 42g alcohol or just over five UK units. drinking were these goal preferences related to drinking outcomes, either during the up to six months of treatment or over the following year. However, in the post-treatment period abstinence-preferring patients were significantly more likely to achieve their goals in any given week than were patients who had preferred controlled drinking. The latter might have specified a maximum amount or simply occasional drinking. As far as possible the analysis compared what they drank against their specific individual preferences, though the lack of specificity in some cases may have led to patients being wrongly assumed to have breached their intended drinking limits.
3 In Canada too, the initial drinking goal of alcohol treatment outpatients was generally unrelated to drinking over the following year. In contrast, patients who after four weeks of treatment contracted with their therapists to maintain abstinence were significantly more likely to avoid drinking and heavy drinking than patients who contracted to controlled drinking goals. By the last half the follow-up year, 53% of patients who had contracted for abstinence achieved it or successfully moderated their drinking compared to just 9% who had opted for controlled drinking. The study illustrates that choices made on a more informed basis and in a formal agreement can be more closely tied to outcomes than those made before treatment has started. In this study patients were also asked to set drinking goals after each therapy session. Reporting back at the next session, patients who had set a non-drinking goal were far more likely to have achieved it than patients who aimed to moderate their drinking.
Alcohol treatment guidance published in 2006 by England's Department of Health and National Treatment Agency for Substance Misuse stressed that alcohol misusers should not be excluded from support or treatment "if they decline to change their drinking or choose to pursue a goal of continued, moderated drinking". Known by the acronym MoCAM, the guidance did however recognise that abstinence "will be the preferred goal for many problem drinkers with moderate to severe levels of alcohol dependence, particularly for individuals whose organs have already been severely damaged through alcohol use, and perhaps for those who have previously attempted to moderate their drinking without success. Moderation, or controlled drinking, is often a more acceptable goal for problem drinkers with low to moderate levels of alcohol dependence. Moderation can also be used as a goal with problem drinkers for whom abstinence would usually be advisable, but for whom this goal is not currently acceptable. A reduction in alcohol consumption will be likely to confer benefits and may offer a stepping-stone to abstinence in the future."
The MoCAM guidance was itself informed by a literature review conducted by three of the UKATT investigators so was influenced by their views and by UKATT findings prior to the release of the featured report. This review recommended that moderation as a treatment goal should be reserved for service users with less severe dependence, and suggested some specific severity indicators, but cautioned that "selection of drinking goal is essentially a clinical decision, depending on the unique characteristics and circumstances of the individual", and that refusal to accept an advised goal should not lead to denial of care.