KEY STUDY Project Project MATCH: unseen ...

KEY STUDY

Project MATCH: unseen colossus

The largest alcohol treatment trial ever, Project MATCH’s value lies more in its unanticipated

findings than in what it set out so painstakingly to prove – less in matching treatment technologies to

patient variables, more in the human touch and doing whatever you do well. by Mike Ashton

From Drug and Alcohol Findings issue 1. Editor, Drug and Alcohol Findings.
Subscriptions and more information
www.drugandalcoholfindings.org or e-mail The author owes a considerable debt to the members of the advisory

panel and others who contributed their expertise and experience Ackowledgements, p 19.

[email protected]

In one respect the title of this key study is Research design and findings MATCH’s therapies correspond to
an exaggeration. Project MATCH’s approaches commonly practised in the UK.6
colossal (or, as others have put it, Project MATCH tested three psychosocial Their key features are outlined below. Details
“Titanic”1, 2) status is beyond doubt, meas- therapies on 1726 problem drinkers, nearly are in the manuals, themselves a (if not the)
ured by the conviction that the like of it will all diagnosed as dependent. Though dispa- major output of the project ( Doing it the
never be seen again. Yet it has not overshad- rate in personal characteristics and in the se- MATCH way: the manuals, p. 20).
owed practice debates to a degree commen- verity of their drinking, they were a relatively
surate with its bulk;3 ‘underseen’ rather than ‘pure’ alcohol problem sample; various forms Twelve-Step facilitation therapy was
‘unseen’. Awareness of its implications has of comorbidity were excluded. newly developed by MATCH but based on
been impeded by the volume of the research the familiar tenets of Alcoholics Anonymous
output and the complexity of the findings. The therapies were selected to be distinct (AA). Over 12 weekly sessions clients were
from each other and for their potential to encouraged to accept that they suffered from
Some may also have preferred to look the reveal matching effects. All were delivered the disease of alcoholism, to begin working
other way. Designed as the definitive test – on a one-to-one basis in non-residential set- through AA’s twelve steps, and, most impor-
and anticipated confirmation – of a quarter tings, according to manuals developed by the tantly, outside the sessions to become
century’s most promising theories about MATCH team, with abstinence from alco- engaged in Alcoholics Anonymous.
which types of drinkers do best in which hol as the goal. The project went to extraor-
types of treatments (‘matching’), MATCH dinary lengths to ensure that variation in out- Also delivered over 12 weekly sessions,
seemed to justify the demoralising conclu- comes could not be put down to variation in cognitive-behavioural therapy sees prob-
sion that ‘It just doesn’t matter what you do’. the quality or extent of treatment. lem drinking as a learnt if maladaptive re-
“It Ain’t Necessarily So” was how the Jour- sponse to life’s problems. The therapy
nal of Studies on Alcohol headlined its editorial CAPSULE (adapted from an existing guide7) aims to re-
on MATCH.4 The shock of solid negative programme those responses by teaching
findings colliding with strongly supported Project MATCH is the largest scientifically coping skills and alternative strategies for
expectations led to a splattering effect as theo- rigorous alcohol treatment trial ever seen. handling high-risk situations.
retical and empirical effort was expended in
the search for explanations. It tested whether outcomes could be im- Delivered in four sessions but still over
proved by matching clients to one-to-one 12 weeks, motivational enhancement
Though not constructed as a test of interventions based on motivational inter- therapy was adapted from motivational in-
whether treatment ‘works’ (there was no un- viewing, AA’s twelve steps, and cognitive- terviewing.8 This brief intervention aims to
treated control group), affirmation that it did behavioural therapy. generate motivation for and commitment to
was the considerable crumb of comfort res- change, operationalised as progressing the
cued from the $28 million project. It led to Matching effects were few and modest; client through the cycle of change ( Cycling
questions about what approaches “designed motivational therapy was best for angry cli- to recovery, p. 16). If possible, the client’s part-
to differ dramatically in philosophy and pro- ents, twelve step for those highly dependent ner was included in the first two sessions.
cedures”5 might have shared which led to or with pro-drinking social networks.
such similar outcomes. The tentative leads Clients were randomly assigned to these
on this issue may be the study’s most impor- Even with ‘difficult’ clients, the briefer mo- therapies at treatment sites in two types of
tant legacy. At their outer limits, these tend tivational therapy generally performed as well settings: in the outpatient arm, the MATCH
to deconstruct alcohol ‘treatment’ into more as the more intensive interventions. intervention was a standalone treatment; in
mundane human virtues – the desire to get the aftercare arm, clients were referred to
better, supportive human relationships, Matching might yet be seen with a wider MATCH after at least a week’s inpatient or
competence, friendly persistence and the range of treatments, settings, clients and vari- intensive day hospital treatment. Intake as-
provision of a culturally endorsed lifeline for ables. MATCH’s grounding in a medical sessments were followed by outcome evalu-
a culturally defined problem. model of alcohol dependence circumscribed ations every three months in the first year
its reach in these respects. after treatment. For the outpatients only, there
These thoughts emerge from a process was also a three-year follow-up. Re-contact
which takes us through the study’s design All the treatments seemed effective with a rates and checks on self-reports of drinking
and findings, its methodological strengths range of clients; the client’s readiness to change lend considerable confidence to the results.
and limitations, and finally its implications had a major positive impact on outcomes.
for practice in the UK. First, some advice: Not a lot of matching
MATCH is a multi-layered mega-study, Practice implications include making moti-
neither effortlessly nor quickly digested, but vational therapy the therapeutic starting Uncontaminated by prior intensive treat-
very filling – a feast rather than a snack. Set point, clearly structuring interventions, and ment and followed up over three years, the
aside some time to digest and enjoy. engaging clients in mutual aid networks. outpatient arm of the study provides the pur-
est and longest-term test of matching. It is
Generalisability of the findings to the UK is also the one most relevant to UK practice:
limited by the US context and by the extraor-
dinary measures taken to safeguard the
integrity of the research and the treatments.

1999 ISSUE 1 DRUG AND ALCOHOL 15

KEY STUDY

here, as in the USA, the pressure is on to honest and fair ... a creative parent advantage for twelve step over cogni-
achieve results without the expense of resi- and clinically meaningful way to tive-behavioural therapy: a modest 8% fewer
dential care. How these patients fared at three communicate the trial’s implications ... drinking days and two units less when drink-
years is our prime focus, with other results insightful interpretations ... excellent ing. On these measures, the motivational in-
drawn on for elaboration or confirmation. tervention fell between the other two, so
Thomas Babor close to both that the differences might have
After three years 85% of the 952 outpa- occurred by chance.12
tients completed tests probing 33 predicted Principal Investigator, MATCH Coordinating Center,
matches. Two reached statistical significance. commenting on this article in draft As well as scotching expectations of poorer
With so many opportunities for matching ef- results after the briefer therapy, MATCH
fects to emerge, these two may have done so sumed less when they did drink. Twelve step furnished the first sound demonstration that
by purely by chance – just as enough throws therapy seemed to neutralise pro-drinking the lay wisdom of the twelve steps can do as
of a dice will eventually produce consecu- social influences partly by fostering AA par- well as clinically developed therapies. The
tive sixes. However, statistical counterbal- ticipation, a ready-made anti-drinking social caveat is that the non-twelve step therapists
ances and the fact that both matches ‘make network. In contrast, three years after moti- could not encourage AA attendance, itself
sense’ increase confidence in their validity. vational therapy the influence of a pro-drink- associated with positive outcomes. Without
ing social circle seemed to reassert itself. this restriction the other treatments might
The first match was that clients high in What of clients with relatively anti-drinking have outperformed the twelve step option.
anger did best after motivational therapy; they social networks? They did better after moti-
had drunk on fewer days in the previous three vational than twelve step therapy.9 Focusing on improvements in the outpa-
months (33% compared to 24%) and con- tient arm at three years almost certainly un-
sumed less when they did drink. This find- The aftercare arm of the study – whose derstates the full sample’s progress; despite
ing was both persistent and in line with the clients had previously undergone intensive more severe initial problems, after one year
theory being tested. More mysterious was the treatment – contributed one further match. 15% more clients in the aftercare arm had
reverse finding that low-anger clients did least After one year clients highly dependent on achieved abstinence. Allocation to arms of
well in motivational therapy. alcohol drank less after twelve step than af- the study was not random, meaning that bet-
ter cognitive-behavioural therapy; they even ter outcomes in the aftercare arm cannot be
The other match emerged only at three fared better than low dependence twelve step securely attributed to the preceding inten-
years but was the largest and perhaps most clients. For low dependence clients, cogni- sive treatments – but it seems a fair bet that
interesting. Clients with social circles highly tive-behavioural therapy was the better. these account for at least part of the effect.
conducive to drinking did best after twelve
step therapy. They drank on 16% fewer days In summary, matching effects were few, Taking both arms together, at one year
than after motivational therapy and con- modest, and none generalised across both (the latest they were both sampled) drinking
arms of the study. In the outpatient arm just days had decreased fourfold to under six per
Cycling to recovery two were long-term: the relative efficacy of month and an average intake of 26 units on a
motivational therapy for highly angry clients, drinking day had dropped to five. Patients
Prochaska and DiClemente’s cycle of change and of twelve step therapy for clients with also improved in many other areas including
model is fundamental to much treatment pro-drinking social networks. depression, alcohol-related problems, drug
practice here and in the USA. It provides a use, and liver function.
common language for communicating about Other seemingly well-founded theories
clients and a rationale for tailoring treatments failed the MATCH test. Most predicted that Client’s readiness key to change
to the client’s readiness to respond. This de- those with more severe problems would do
scription is adapted from MATCH’s manual less well in the briefer motivational inter- Even if there seems little to be gained from
on motivational enhancement therapy ( vention. But this seemed just as suitable for matching client characteristics to treatments,
Doing it the MATCH way: the manuals, p. 20). heavy as for less heavy drinkers; at three years, some characteristics may promote recovery,
clients highly vulnerable to dependence were whatever the treatment. Little can be done
People who are not considering change in doing better after motivational therapy. Cli- about a client’s age, gender, social status and
their problem behaviour are described aspre- ents with greater psychological problems did so on, but other characteristics might poten-
contemplators. In thecontemplation stage as well in any of the treatments. However, tially be fostered by treatment providers.
individuals begin to acknowledge they have motivational therapy was no more effective Again our focus is on the (more UK-relevant)
a problem and to consider the feasibility and for unmotivated clients, supposedly its forte. outpatient arm of the study
costs of changing their behaviour. As they The anti-social personalities with whom cog-
progress, they move on to the determina- nitive-behavioural therapy should have ex- Topping the list of client characteristics
tion stage, where the decision is made to take celled did as well in the other modalities. linked to treatment success – and even more
action and change. Once individuals begin to important than the initial severity of their
modify the problem behaviour, they enter the Treatment works alcohol problems – was their readiness to
action stage, which normally continues for With the matching results a disappointment, change their behaviour, reflecting what the
three to six months. After successfully nego- the study’s sponsors emphasised the overall client brings to the process before treatment
tiating the action stage, they move to main- impact of the treatments, whoever the clients.10 has begun.13 Over three years down the line
tenance or sustained change. If these efforts Lack of an untreated comparison group un- this still had a profound impact on abstinence
fail, a relapse occurs and the individual be- dermined their case, but the improvements and restraint when drinking. The importance
gins another cycle were impressive. Again, our focus is on the of this factor is supported by the rapidity of
enduring results after the standalone treat- change; practically all the improvement in
Lasting exit Relapse ments. Three years later, almost 30% of the drinking was evident by the first week of
Maintenance former alcoholics had not touched a drop in treatment ( chart, p. 19).14
the preceding three months. The remainder
Action Pre- drank on only a third of the days, typically By three years, how pro-drinking the cli-
CYCLE OF CHANGE contem- Enter consuming 11 UK units a day compared to ent’s social circle had been on intake to treat-
plation 19 before treatment.11 ment no longer made any overall15 difference
to outcomes. Still modestly predictive was
Determination Contemplation Set against these pre–post treatment gains, how confident the client had felt about tack-
differences between treatments were trivial. ling their drink problem, an association also
Sophisticated computations teased out an ap- found among aftercare clients.

Some severe problem clusters were ex-
cluded, but it should still give pause for

16 DRUG AND ALCOHOL ISSUE 1 1999

KEY STUDY

Core documents aggression and masculinity and high in nur- client and an intervention that leads to
turance did best. In contrast, twelve step change. These are the issues explored below.
MATCH’s major outcomes and implications therapy benefited from high aggression and
have been reported in four core papers. low nurturance. Above the relatively high Clients: only the best?
These are drawn on throughout the text and floor set by MATCH, greater educational The project homed in on alcohol problems,
usually not specifically referenced. achievement or experience conferred no fur- excluding other problems which may have
ther benefits; in twelve step therapy these confused the results. Drug users were not
Project MATCH Research Group. “Match- were negatively related to outcome. Even in excluded but drug dependants25 and recent
ing alcoholism treatments to client heteroge- twelve step therapy it did not matter whether injectors were. Also partly or wholly excluded
neity: Project MATCH posttreatment drink- the therapist was a recovering alcoholic. With were: the under-18s; the psychotic; the po-
ing outcomes.” Journal of Studies on Alcohol: one exception, gender also was irrelevant; tentially violent; the socially isolated or
1997, 58, p. 7–29. One-year outcomes for among outpatients, and only in twelve step homeless; and those currently under crimi-
both arms of the study relating to the most therapy, women did better with women. nal justice supervision. There were so few
promising of the matching variables. highly disturbed outpatients that potential
Perhaps these crude categories failed to matches might have been missed.
Project MATCH Research Group “Project capture the client-therapist affinities needed
MATCH secondary a priori hypotheses.” to generate emotional bonding and the feel- So the study can only afford limited clues
Addiction: 1997, 92 (12), p. 1671–1698. ing of a shared enterprise – the ‘therapeutic about how to handle the most disturbed and
One-year outcomes for both arms of the alliance’. For therapists and clients these feel-
study relating to an alternative less well es- ings were tapped via a questionnaire. Among many criticisms of
tablished set of matching variables, plus a outpatients a firmer alliance (as experienced MATCH’s design are guilty of
summary of all one-year results. by either side after their second meeting) was
modestly but consistently associated with being ‘wise after the event’
Project MATCH Research Group. “Match- treatment participation and better out-
ing alcoholism treatments to client heteroge- comes.19 In turn, entering treatment ready Nick Heather
neity: Project MATCH three-year drinking to change led to a better therapeutic alliance.20
outcomes.” Alcoholism: Clinical and Experi- violent drinkers and those ordered into treat-
mental Research: 1998, 22 (6), p. 1300– These findings chime with UK research ment by the courts. Since these are UK policy
1311. Three-year outcomes for the outpatient showing that a therapist’s initial “therapeu- priorities, there is an argument for querying
arm of the study. tic commitment” to an alcohol client is MATCH’s relevance ( Violence is the issue,
strongly related to whether that client en- p. 18). However, nearly half the outpatients
Project MATCH Research Group. “Match- gages in treatment. This commitment was were using illicit drugs, half the total sample
ing patients with alcohol disorders to treat- experienced by the client as the worker be- had a prior psychiatric diagnosis, and half
ments: clinical implications from Project ing accepting and warm – not surprising, as were unemployed. Unless apparent only at
MATCH.” Journal of Mental Health: 1998, liking the client seemed a key component.21 the excluded extremes, the range of prob-
7(6), p. 589–602. Summarises findings and lems was sufficient for matches to emerge.
gives the MATCH Group’s assessment of Methodological strengths and
their practice implications. limitations Also excluded were patients who rejected
entry into a study which entailed randomised
thought that the worse their alcohol depend- Much of MATCH’s significance resides in treatment. Matching effects may have failed
ence and social functioning at intake, the better its ‘collateral’ findings: treatment seemed to to emerge partly because clients with strong
outpatients did at three years. The strongest exert an impressive impact; the client’s readi- treatment preferences would have excluded
link was between alcohol dependence and ness for change strongly predicted out- themselves. Two very common matching
abstinence, suggesting that the most depend- comes;22 and the bond between therapist and methods – client preference and therapist
ent tended to cope by not drinking. Turning client had a more consistent impact than intuition/assessment – were not tested.
to the study as a whole (aftercare as well as either the match between treatments and
outpatient), at one year a raft of unpromis- clients, or between broad therapist and client Measurement: therapeutic in itself?
ing symptoms including ‘sociopathy’ and characteristics. But these findings emerged Repeated test batteries mostly administered
psychiatric severity as well as dependence and from a study not designed to test them. How face-to-face were essential to the study’s de-
poor social functioning, also had either no much faith can we have in them, and did the sign, but it’s more than plausible that the
impact or inconsistent impacts on outcomes. design somehow obscure the matching ef- therapeutic benefits of eight hours of initial
fects it sought? testing and five three-monthly follow-up in-
Matching therapist to therapy terviews partially submerged differences in
Though matching clients to treatments Matches may have failed to emerge be- the impacts of the formal treatments.26
seemed relatively unimportant, there was cause the study sacrificed external in favour Indeed, patients sometimes confused their
some evidence for another kind of matching of internal validity ( Glossary).23 Internal va- research contacts with their therapists.27
– matching therapists to treatments. Detailed lidity is the extent to which the design ad-
findings are as yet unpublished, but clues equately tests predictions about the interven- Treatments: too good, or the same
about which therapists operate best within tion’s impact. The higher it is, the more under the skin?
each of the treatments are available from a confidence we can have that the results are The most frequent ‘criticism’ of MATCH is
summary paper16 and from MATCH re- not due to something else. But tight design that its treatments were so good that improve-
searchers’ comments last May in Leeds.17, 18 – such as restricting patient intake and thera- ments in clients left no room for matching.
Though provocative, methodological limi- peutic discretion – risks divorcing a study This argument was dismissed by the
tations reduce confidence in the findings. Pa- from the typically more messy world out- MATCH team and seems misconceived. If
tients were not randomly allocated to thera- side. This is the issue of external validity – a treatment is rubbing a client up the wrong
pists and the study’s quality controls probably generalisability to other clients and settings. way, the greater the fidelity and persistence
helped confine the therapist effect to one or of its delivery, the worse would be the results.
two ‘outliers’ with unusually poor results; In the understandable search for internal
outside MATCH, variable therapist compe- validity,24 in at least three ways MATCH may More convincing is the argument that fea-
tence is likely to exert a greater influence. have obscured the effects it sought: client se- tures shared by the treatments led to similar
lection; measurement; and treatment. Alter- outcomes. All were 12-week, one-to-one
In tune with the approach’s non-confron- natively, perhaps its search for matches was
tational style, motivational therapists low in too crude or misdirected, failing to capture
what it is about the interaction between a

1999 ISSUE 1 DRUG AND ALCOHOL 17

KEY STUDY

counselling interventions. At a deeper level, perhaps with all to emerge, explanations which leave
all provided a clear and credible programme mental health work, there is a sense MATCH’s fundamental assumptions un-
for recovery, delivered with stringent qual- in which ‘It’s not what you do, it’s touched and with them the possibility that a
ity control and measures to maximise treat- later study would find those elusive matches.
ment compliance. Compatible with this view the way that you do it’ But there is a deeper level of doubt over
is the finding that therapies thought to work MATCH’s approach, one probed by our own
through distinct psychological systems actu- Mike Ward and other expert commentators, and by two
ally worked in similar ways. Cognitive- of the study’s experienced therapist-research-
behavioural training did not specially improve UK, controlled drinking and harm reduction ers; it relates to MATCH’s vision of the
social skills or psychopathology. Motivational have far greater currency than in the US. nature of addiction.
therapy’s non-confrontational style did not
particularly strengthen the client-therapist More generally, the research protocol cur- MATCH coordinator Thomas Babor has
bond.28 Such findings have been reported tailed the leeway for clients to modify their characterised its matching tests as based on a
elsewhere.29 therapeutic programme – a serious restric- “technological, medical model”.34, 35 Under-
tion when the essence of a therapy is client lying most was the assumption that fixed fea-
However, the scope for other approaches empowerment, the case for motivational in- tures of the client’s personality and initial al-
to prove more suitable for some clients than terviewing.32 MATCH’s motivational cohol problem could be keyed into treatment
for others remains wide. Matching might therapy still did remarkably well, but allow- technologies, much as a key fits a lock and
have been more evident if family therapy, ing greater client leeway might have thrown enables change – an opening door. But what
pharmacotherapy, psychodynamic therapies, up clearer findings about which clients were if the lock was in a dynamic relationship with
group approaches, self or mutual help had best placed to exploit this freedom. its environment, changing shape depending
also been in the frame, or if the MATCH on the colour of the door, the time of day,
treatments had been varied in intensity. Ex- Too thin to match? and the weather? No key could be made con-
clusion of group approaches – the “stock in Among MATCH researchers a favoured ex- sistently to fit – or not until we understood
trade” of alcohol treatment30 – is seen as par- planation for the lack of matching is that the these interactions; is this what happened in
ticularly unfortunate. theories available to be tested were too sim- MATCH?
plistic. MATCH generally tried to match
Britain is different treatments to uni-dimensional, standard cli- The study’s focus on psychological defi-
ent attributes, but real-world therapists make cits36 internal to the client fits with a concept
Two factors which might have obscured multi-dimensional assessments combining of dependence as a disease ‘inside’ the pa-
matching are specially relevant to the UK. objective and intuited client attributes. Multi- tient. In this vision, the severity of the dis-
First, MATCH therapies were all delivered dimensional (or ‘thick’) matching might have ease would be a natural predictor of its re-
at hospitals, but UK policy emphasises com- proved more effective. However, it is doubt- sistance to treatment. However, this was not
munity settings such as primary care, the pro- ful whether current research methods could the case in MATCH nor in a very differ-
bation office, or the local voluntary alcohol capture such complexity;33 adding in combi- ently designed major US study of alcoholics
service.31 What works there with the type of nations of variables and those less suscepti- followed up eight years after seeking treat-
clients seen in those settings is not necessar- ble to measurement might threaten the ment in everyday conditions.37 In this ‘natu-
ily the same as what works at US hospitals. replicability and applicability of the findings. ralistic’ study the impact of treatment was
Secondly, though MATCH’s manuals were overshadowed by the patients’ long-term
designed to allow a flexible response to cli- Looking the wrong way? access to social and community resources.
ent needs, clients could neither choose their These included AA, but also more signifi-
initial drinking goal nor modify it as therapy The previous section explored the ‘techni- cantly the respect, understanding and sup-
progressed – it was always abstinence. In the cal’ reasons why matching may have failed port experienced from family and friends.
Stability of relationships and of employment
Violence is the issue are also important predictors of successful
Mike Ward argues that MATCH is largely beside the point. outcomes.38 MATCH itself found that, what-
ever the treatment, participation in AA’s anti-
Reading MATCH in the context of community tal health services, “The first thing is to tackle drinking social structures was associated with
care in the UK in the late 1990s is like looking the link between substance misuse and mental less drinking.
at yesterday’s high-tech gadget – interesting, health”.67
but things have moved on. Such results fit an alternative view of ‘de-
Why this concern? Substance misuse may pendence’ as a functional way of relating to
‘Social care’ incorporates three main ele- have been the real cause of most recent homi- one’s environment – a vision which redirects
ments: change; maintenance; and control or cides committed by the mentally ill in the UK.68 attention away from the client’s personality
protection. For the last few decades alcohol Similarly, a US study of recently discharged and towards the varying difficulties they ex-
treatment has concentrated on the first: which mental patients found “co-occurring substance perience in changing this relationship.39 By
treatment is most likely to foster change? abuse disorder to be a key factor in violence”.69 definition, these difficulties lie not inside the
MATCH was born of that focus. But today’s client but between them and those who af-
debate about mental health and community The new challenge for substance misuse fect and are affected by their drinking, rela-
care is driven by fear of violence by people dis- services is how to control people and protect tionships which themselves will be embed-
charged from mental health services. The pres- society. Of course, effective treatment protects ded in the wider context of that society’s
sure is on simply to control people, and alco- society from substance-related crime – but only social structures.40
hol and drug services are not exempt. among those presenting for treatment. For the
dually diagnosed, the major need is to work Treatment or faith?
In top policy making circles substance mis- with those who don’t volunteer for treatment.
use is now recognised as central to the per- Arguably, which treatments work best for ‘vol- Returning to (and stretching) the lock (pa-
ceived ‘failure of community care’. The Director unteer’ clients is now less important than de- tient) and key (treatment) analogy, what if it
of the Centre for Mental Health Service Devel- veloping strategies to address these concerns. all depended on whether the lock was ready
opment recently argued that to improve men- to be opened, and anything which looked like
by Mike Ward Acknowledgments its idea of a suitable key would spring it?
Then no matter what the key’s shape, as long
as it looked sturdy, polished and above all
key-like, it would ‘work’.

18 DRUG AND ALCOHOL ISSUE 1 1999

KEY STUDY

100% OUTPATIENT ARM Rapid and plation phase of the cycle of change, all be- For the UK, arguably the least pertinent
dramatic fore treatment had started.42 In many studies match involved psychiatric severity. There
% days abstinent reduction in – MATCH is one – such differences might was some evidence that relatively problem-
Intake drinking in the wrongly have been attributed to treatment. free clients did better after twelve step therapy
50% first week of Another study found that pharmacological – but only in the short-term and with re-
treatment cast supports to abstinence were overshadowed spect to abstinence rather than controlled
CBT doubt on by the client’s initial readiness to take action drinking. Given twelve step’s dominance in
MET whether over their alcohol problem.43 The centrality the USA and the greater salience of control-
TSF treatment was of readiness to change would also explain led drinking in the UK, the transportability
0% the active why no matching effects duplicated across of this tentative finding must be open to
Baseline Treatment weeks 1-12 ingredient in MATCH’s aftercare and outpatient arms; question.
Months 1-3 change. these populations (one mainly abstinent, the
other trying to be) were at different stages of MATCH’s ‘failure’ (if that’s what it was)
Source change so different factors were important. to find compelling matches was one with the
reference 14 distinctly positive implication that, within the
Practice implications study’s limits in terms of clients, treatments
Is this why the MATCH treatments – and attributes tested, treatment providers
competently delivered and meeting US cul- Much of MATCH’s significance, prompted need not bother too much about triaging new
tural expectations of what ‘therapy’ should by what it did not find and by its ‘collateral’ patients and can get on with treating them.
look like – indiscriminately evinced such findings, might lie in its longer-term contri-
rapid and dramatic results, well before treat- bution to understanding addiction. Emerg- Treatment works ...
ment would ‘bite’? For the most successful ing from MATCH ought to be a research Or, more accurately, MATCH suggested it
clients, what was there from the start – and it programme designed, as MATCH was not, can work very well when provided on an in-
was the best predictor of outcomes – was to probe these more fundamental issues. But dividual basis using motivational interview-
readiness to change. what of the here and now? ing, cognitive-behavioural or twelve step ap-
proaches delivered with thorough therapist
Ruminating on these phenomena took Matching (of this kind) unimportant screening and training, stringent quality con-
two of MATCH’s respected researchers very After their one-year results MATCH authors trol and persistent anti-drop out measures.
far from the project’s implicit view of treat- were confident only of the near futility of ‘Anything goes’ is certainly not the MATCH
ment as a technical fix to a medical or psy- matching their treatments to the client at- message.45
chological disorder.41 William Miller and tributes they tested. Later (with the largest
Thomas Babor argued that the active ingre- Moreover, within wide (but not extreme)
dient was the client’s decision to put their life its key failing was to assume limits, these treatments work almost regard-
in order and the resources available to them there was something different about less of the initial severity of the client’s alco-
to do this. Miller believed treatment merely substance misusers to which a magic hol, social or psychological problems. Such
gave people ‘permission to recover’ and pro- approaches can consider themselves vindi-
vided some of those resources. Babor specu- treatment could be matched cated for a substantial range of problem
lated that offering any culturally accepted drinkers and should be available within a dis-
route to recovery might work the same magic Bob Purser trict’s mainstream alcohol services.
as ‘treatment’ or ‘therapy’ in Western socie-
ties. In some cultures, faith healers and witch match yet found emerging at three years) they ... but this cannot be assumed
doctors also give ‘clients’ the belief that they were more upbeat. Despite their modest size, Perhaps the most salutary lesson to take from
can get better and the confidence to go ahead “the matches ... are reasonable considerations MATCH is its clear demonstration of the
and do it – effectively, do it themselves. ... to take into account when planning a treat- danger of untested assumptions. The only
ment program” if the therapies are “deliv- way to be sure that current treatments or in-
Supporting such views, a recent study re- ered with fidelity by trained therapists”. novations actually do deliver results is to
ported substantially better psychological monitor the bottom-line – outcomes46 – and
functioning when clients seeking treatment Among the matches most relevant to UK the treatment processes that lead up to them.
were in the action as opposed to the contem- practice was the finding that motivational Drop out rates pre and post assessment, dur-
therapy coped best with highly angry cli- ing treatment, and in aftercare, can pinpoint
Acknowledgments ents; its non-confrontational style may be less where clients are lost to the process. Investi-
likely to ignite short fuses. Also relevant is gation of why this is happening should lead
We are grateful to Griffith Edwards, editor of Addiction, the fact that in the long term clients im- to the testing of remedial measures, such as
for permission to draw on an a draft paper summarising the mersed in pro-drinking social networks those documented in MATCH manuals.
findings of Project MATCH and to MATCH’s Thomas did best in the therapy which focused on en-
Babor for forwarding that and other papers. couraging participation in anti-drinking net- Even some small UK alcohol agencies also
works. Especially where these are lacking, manage to assess longer term outcomes. The
This report draws on the views of several experts, each of building social networks supportive of the jolt of high relapse rates may be unpleasant,
whom was asked to focus on a particular aspect but were client and of their drinking goals may be an but should spur improvements. Commis-
also invited to comment more broadly. Though they have important buffer against relapse.44 In sioners could be made aware if these im-
enriched it, they bear no responsibility for the final text. MATCH this was achieved via AA; in the provements translate into cost savings down
UK, other networks might be acceptable to the line. They are likely to be especially im-
thanks the following for generously giving its a broader range of clients. pressed if (as will often be the case) savings
readers the benefit of their experience and expertise. accrue to the services for which their author-
Methodological strengths and limitations Twelve step therapy seemed to have par- ity also foots the bill (reduced hospital ad-
ticular affinity for those highly dependent missions, GP visits, prescriptions etc).
Professor Nick Heather, Director, Centre for Alcohol on alcohol, but again the findings were in-
and Drug Studies, Newcastle City Health NHS Trust; consistent and ambiguous. Without prior in- Make motivational therapy the
Principal Investigator (Research Coordination), UK Alcohol tensive treatment, interventions based on starting point
Treatment Trial (UKATT). cognitive or motivational approaches worked One of MATCH’s most important findings
Practice implications Bob Purser, Director, Aquarius, just as well. was that motivational therapy was at least as
Birmingham. Aquarius is a multi-site community-based suitable for heavier and more vulnerable
alcohol agency which bases its practice on evaluations of
alternative programmes. Barbara Elliott, then Director of
ACCEPT Services. ACCEPT is a day centre in London offer-
ing abstinence and controlled drinking service. Mike Ward,
Purchasing Manager (Alcohol & Drugs), Surrey County
Council. Mike has extensive experience managing drug
and alcohol services and recently investigated the relation-
ship between substance use, homicide and mental illness.
Comments on the first draft Thomas Babor, Allan
Zweben and other members of the Project MATCH
Research Group. Barbara Elliott. Nick Heather. Bob
Purser. Libby Ranzetta, Services Officer, Greater London
Alcohol Advisory Service (GLAAS). The
Editorial Board.

1999 ISSUE 1 DRUG AND ALCOHOL 19

KEY STUDY

drinkers as the two more intensive therapies. pact of these measures – MATCH was de- Doing it the MATCH way:
Within the limits of problem severity tapped by signed to eliminate rather than test variabil- the manuals
MATCH, motivational interviewing seems ity in quality – but ‘quality counts’ is widely
a safe and cost-effective starting point47 for seen as the project’s key lesson,51, 52 particu- Twelve step facilitation therapy
one-to-one professional outpatient therapy. larly as US alcoholism treatment generally manual: a clinical research guide
It may be less suitable if rapid in-treatment under-performs compared to MATCH.53 for therapists treating individuals
reduction in drinking is important, but even with alcohol abuse and depend-
this finding could be an artifact of MATCH’s To this issue MATCH and its sponsors ence. Nowinski J., Baker S., Carroll
four sessions being strung out over 12 weeks. devote most space in their assessment of the K. Project MATCH Monograph
project’s practice implications; matching is Series, Volume 1. 1995.
This conclusion cannot (yet) be applied relegated to single paragraph.54 The implica-
to the homeless, isolated, illiterate, psychotic, tion is that service commissioners need to Motivational enhancement
violent, suicidal, criminally involved, drug concentrate not so much on purchasing spe- therapy manual: a clinical research
dependent or young drinkers partly or wholly cific therapies, as on purchasing and encour- guide for therapists treating indi-
excluded by MATCH. Since these are among aging quality staff inputs. MATCH itself viduals with alcohol abuse and
the UK’s treatment priorities, it would be spotlighted therapist training, manual-guided dependence. Miller W.R., Zweben
folly to dismantle more intensive treatments therapy and measures to encourage client A., DiClemente C. et al. Project
without a similarly rigorous test of brief compliance as the key quality variables; each MATCH Monograph Series,
therapies for at least these populations. All is dealt with below under its own heading. Volume 2. 1995.
the more so since MATCH’s ‘aftercare’ cli- Such strategies are not unknown to UK agen-
ents started off worse but in the end fared cies,55 but the MATCH $millions and its in- Cognitive-behavioral coping
significantly better than those without the ternationally respected collaborators brought skills therapy manual: a clinical re-
benefit of prior intensive treatment.48 them to unprecedented heights. search guide for therapists treat-
ing individuals with alcohol abuse
Mutual aid prevents relapse Therapists as important as treatment and dependence. Kadden R.,
One interpretation of MATCH is that AA Carroll K., Donovan D., et al.
proved itself at least as good as professional MATCH’s therapists were qualified, expe- Project MATCH Monograph
therapies; dismantle the treatment industry rienced in and committed to the relevant Series, Volume 3. 1995.
and give the money to mutual aid groups was therapy, trained over three days, sifted
through two videotaped test clients, super- Improving compliance with alco-
MATCH will increase the vised weekly by staff of leading US centres, holism treatment. Carroll K. ed.
pressure to ask more questions about and monitored by random videotaping to Project MATCH Monograph
what helps people and what does not correct therapeutic drift. Commending such Series, Volume 6. 1997.
procedures, the MATCH team suggest that
Barbara Elliott “therapist preparation [may be] at least as Published by the US Department of Health
important a factor as treatment modality”.56 and Human Services and available from:
the controversial implication some drew Distribution Center, National Institute on
from the study.49 MATCH countered that One objective of therapist selection and Alcohol Abuse and Alcoholism, PO Box 10686,
its therapy was not AA but a professional (and training is to maximise the ability to forge a Rockville, Maryland 20849-0686, USA.
expensive) programme intended to foster AA therapeutic alliance with clients. MATCH Fax 00 1 202 842 0418 http://www.niaaa.nih.gov
participation: “A conclusion that AA attend- confirmed that this bond consistently con-
ance can simply be substituted for treatment is tributed to good outcomes. Matching the If Babor and Miller’s suspicions are cor-
therefore unwarranted”.50 therapist’s personal attributes (empathy, ag- rect ( Treatment or faith? above), these mes-
gression, etc) to the type of treatment also sages will have powerfully communicated the
But could AA-type volunteers have done leads to better results. For the core motiva- ‘permission to recover’ which gave clients the
just as well as MATCH’s twelve-step thera- tional approaches, non-judgemental, warm confidence to action their readiness to
pists? We know that ex-alcoholic therapists and empathic individuals, committed to the change, leading to rapid improvements on
did as well as those without a history of al- therapy and to the clients, are likely to achieve treatment entry. What MATCH did other
coholism and that, above the relatively high the best outcomes. services can aspire to, building on the head
floor set by MATCH, experience and edu- start given by the project’s manuals.
cation, and the accepted attributes of a good The best therapists will probably be able
therapist, seemed if anything a hindrance to to switch emotional tone depending on the Make it hard to stay away
encouraging twelve step participation. What- therapy and the client’s readiness to change. What was it that kept MATCH clients com-
ever the treatment, such participation was as- The best services will be as thorough as ing back for on average 70%59 of their sched-
sociated with better results. Together these MATCH in screening who will treat their uled sessions? MATCH’s unusually low
findings suggest that committed (but trained clients and in ensuring they stay on song. drop-out and high client compliance were
and monitored) volunteers could protect a minimally related to the severity of the cli-
service’s investment in mainstream therapy Structure and codify the treatment ent’s problems. There is no justification here
by promoting engagement in mutual aid net- for focusing on the least damaged referrals
works. Agencies which fail to take out this Clear, credible programmes offering a cul- on the grounds that they are most likely to
insurance risk seeing the benefits of their pri- turally appropriate route out of alcohol prob- benefit from treatment. Rather, MATCH’s
mary treatment rapidly negated. lems, persuasively communicated by com- philosophy (supported by its findings60) is to
mitted therapists, may provide the initial see compliance not as a feature of the client,
Quality counts impetus to successful treatment. All this is but as emerging from the partnership be-
MATCH clients were engaged with selected, easier to achieve on the basis of a well struc- tween client and treatment.61
trained therapists in clear, well structured, tured manual. MATCH’s manuals provided
quality-controlled programmes from which the focus for training and the yardstick against Several of MATCH’s compliance strate-
neither was allowed easily to drift. No de- which to monitor therapeutic drift. They also gies amount to fostering this sense of part-
finitive statements can be made on the im- codified the therapy’s rationale and structure nership. Among them are the cost-less vir-
and mandated therapists to communicate tues of a pleasant and respectful atmosphere,
these to clients at the first session. Clients clear communication of staff and client roles
will also have gathered that they were being and responsibilities, communicating empa-
enrolled in a prestigious study testing ‘gold thy and concern for the individual’s welfare,
standard’ treatments.57, 58 and pride of participation – in MATCH’s

20 DRUG AND ALCOHOL ISSUE 1 1999

KEY STUDY

case, in a national study;62 elsewhere, per- codified its tactics in yet another manual and why; have and give confidence that to-
haps the feeling that the client is integral to a ( Doing it the MATCH way: the manuals). gether you can make things better; assume
venture of which they as well the staff can nothing. Perhaps above all – timing: catch
feel proud. Postscript: is virtue out of date? people at the cusp of change, or somehow get
them there, and the rest of the journey may
Other anti-drop out measures did require MATCH is a multi-million dollar project of be bumpy and long, but it will be downhill.
resourcing: overcoming practical barriers to a scale few will see in a lifetime. Its results
attendance (eg, childcare, transportation); too are ‘multi’ – multiple, multi-layered, There is, though, the nagging feeling, here
appointment reminders; and rapid response multi-faceted. But the bottom lines seem and in the USA, that such old-fashioned vir-
to missed sessions through letters and tel- taken from a how-to-live-your-life manual tues are giving way to others characterised
ephone calls. MATCH was not designed to of the old school. Be thorough; there is no by cost-driven cuts, short-termism, and so-
test such measures, but another study has substitute for quality; judge not; be welcom- cietal self-defence from deviant minorities.65
reported dramatic improvements in client re- ing and warm, but be persistent; like people In so far as this process has advanced less far
tention from modest but systematic efforts and they will like you; cooperation gets the in Britain than in the USA (where it has been
along these lines. By reducing re-admission job done; help yourself by helping others; driven by medical insurance companies),
rates, such measures could more than pay know what you should be doing, do it, and MATCH may have greater impact here than
for themselves.63, 64 Helpfully, MATCH has do it well; explain clearly what you are up to in its homeland.66

1 Drummond D.C. “Treatment research in the wake of 21 Cartwright A., Hyams G., Spratley T. “Is the inter- and the tranquilliser buspirone.
Project MATCH.” Addiction: 1999, 94(1), p. 39–42. viewer’s therapeutic commitment an important factor in
determining whether alcoholic clients engage in treatment?” 44 Humphreys K., op cit.
2 Glaser F.B. “The unsinkable Project MATCH.” Addiction: Addiction Research: 1996, 4(3), p. 215–230.
1999, 94(1), p. 34–36. 45 Orford J., op cit.
22 Outpatients only.
3 Glaser F.B., op cit. 46 Glaser F.B., op cit.
23 Heather N. “Some common methodological criticisms
4 Schuckit M.A. “Editor’s corner: it ain’t necessarily so.” of Project MATCH: are they justified?” Addiction: 1999, 47 Cisler R., Holder H., Longabaugh R., et al. “Actual and
Journal of Studies on Alcohol: 1997, 58, p. 5–7. 94(1), p. 36–39. estimated replication costs for alcohol treatment modalities:
case study from Project MATCH.” Journal of Studies on
5 Project MATCH Research Group. “Summary of Project 24 Given the optimism over matching in the ’80s, it was Alcohol: 1998, 59, p. 503–512.
MATCH.” Addiction: 1999, 94(1), p. 31–34. essential to design a trial that would provide as unambigu-
ous answers as possible regarding the effectiveness of 48 Gordis E., Fuller R. “Project MATCH.” Addiction: 1999,
6 Edwards G., Marshall E.J., Cook C.C.H. The treatment of client-treatment matching. Previous research had been in- 94(1), p. 57–59.
drinking problems: a guide for helping professions. 3rd conclusive largely because such safeguards had not been
edition. Cambridge University Press, 1997. taken; it was important not to repeat these faults. The in- 49 Schaler J.A. “Selling water by the river: the Project
vestigators would rightly have attracted criticism if this ex- MATCH cover-up.” Psychnews International: 1996, 1 (5).
7 Monti P., Abrams D., Kadden R., et al. Treating alcohol pensive study had not allowed unequivocal conclusions to
dependence: a coping skills training guide. New York: be drawn. 50 Project MATCH Research Group. Journal of Mental
Guilford Press, 1989. Health, op cit.
25 Except cannabis and tobacco.
8 Miller W.R., Rollnick S. Motivational interviewing. New 51 NIAAA press release, 17 December, 1996.
York: Guilford Press, 1989. 26 For example, by forcing clients to reflect on their
problems, their progress in overcoming them, and providing 52 Alcohol Concern. “Project MATCH – what treatments
9 One other outpatient match faded into insignificance at least a not unsympathetic ear. for what clients.” Acquire: summer 1997, p. 1–3.
within nine months of treatment ending. Up to then clients
relatively free of psychological problems drank on fewer 27 Meet the MATCH makers conference, op cit. 53 Dawson D.A. “Correlates of past-year status among
days after twelve step than after cognitive-behavioural treated and untreated persons with former alcohol
therapy. 28 Richard Longabaugh speaking at Meet the MATCH dependence: United States, 1992.” Alcoholism: Clinical and
makers conference, op cit. Experimental Research: 1996, 20(4), p. 771–779. Within
10 NIAAA press release, 17 December, 1996, quoting Dr five years of being treated for alcohol dependence, 70% of
Fuller of the NIAAA: “These findings are good news for 29 Finney J.W., Noyes C.A., Coutts A.I., et al. “Evaluating former patients had been dependent or had abused alcohol
treatment providers and for patients who can have con- substance abuse treatment process models: I. Changes on in the past year and 11% were abstinent.
fidence that any one of these treatments, if well-delivered, proximal outcome variables during 12-step and cognitive-
represents the state of the art in behavioral treatments.” behavioral treatment.” Journal of Studies in Alcohol: 1998, 54 Project MATCH Research Group. Journal of Mental
59, p. 371–380. Health, op cit.
11 Derived from the median drinks per drinking day figures
in: Project MATCH Research Group. “Matching alcoholism 30 Glaser F.B., op cit. 55 An example is Aquarius’s counsellor training, assessment
treatments to client heterogeneity: Project MATCH three- and supervision programme based on its own and Alcohol
year drinking outcomes.” Alcoholism: Clinical and 31 Department of Health and Social Security. Pattern and Concern’s manuals.
Experimental Research: 1998, 22(6), p. 1300–1311. range of services for problem drinkers. 1979.
56 Project MATCH Research Group. Journal of Mental
12 Motivational therapy scored between the other two but 32 Miller W.R. “Motivational Interviewing with problem Health, op cit.
was not significantly worse than twelve step therapy nor drinkers.” Behavioural Psychotherapy: 1983, p. 147–172.
better than cognitive behavioural. However, total abstinence 57 Zweben A., Barrett D, Carty K. eds. Strategies for facili-
in the preceding three months was more common among 33 Finney J.W. “Some treatment implications of Project tating protocol compliance in alcoholism treatment research.
twelve step clients (36% v. 24% and 27%). MATCH.” Addiction: 1999, 94(1), p. 42–45. Project MATCH Monograph Series 7. NIAAA, 1998.

13 No such association was seen (at one year) in aftercare 34 Meet the MATCH makers conference, op cit. 58 Mattson M.E., Del Boca F.K. “Compliance with treat-
clients, presumably because most had already ‘made the ment and follow-up protocols in Project MATCH: predictors
change’ to abstinence. 35 Orford J. “Future research directions: a commentary on and relationship to outcome.” Alcoholism: Clinical and
Project MATCH.” Addiction: 1999, 94(1), p. 62–66. Experimental Research: 1998, 22(6), p. 1328–1339.
14 Project MATCH Research Group. “Matching alcoholism
treatments to client heterogeneity: treatment main effects 36 Orford J., op cit, in which the author refers to Thomas 59 Allen Zweben, personal communication, 1999, referring
and matching effects on drinking during treatment.” Journal Babor characterising the MATCH model as based on “the to both the aftercare and outpatient arms of the trial. The
of Studies on Alcohol: 1998, 59, p. 631–639. correction of individual client ‘liabilities’” (‘liabilities’ is the author adds that “The breakdown is as follow: MET 0.80,
only direct quote). CBT 0.68, and TSF 0.63”.
15 As opposed to the matching effect noted earlier (twelve
step therapy especially benefiting those with pro-drinking 37 Humphreys K., Moos R.H., Cohen C. “Social and 60 Connors G.J. et al, op cit, p. 592. In the outpatient arm
social circles). community resources and long-term recovery from treated of the study therapeutic alliance was positively related to
and untreated alcoholism.” Journal of Studies on Alcohol: client attendance.
16 Project MATCH Research Group. “Matching patients 1997, 58, p. 231–238
with alcohol disorders to treatments: clinical implications 61 Zweben A., et al, 1998, op cit, p. 11–12.
from Project MATCH.” Journal of Mental Health: 1998, 38 Costello R.M. “Alcoholism treatment effectiveness:
7(6), p. 589–602. slicing the outcome variance pie.” In: Edwards G., Grant 62 Mattson M.E. et al, op cit, p. 1337.
M. eds. Alcoholism treatment in transition. Croom Helm,
17 Meet the MATCH makers conference. Organised by 1980. 63 Lash S.J., Blosser S.L. “Increasing adherence to
Leeds Addiction Unit and Society for the Study of Addiction, substance abuse aftercare group therapy.” Journal of
14–15 May 1998, Leeds. The current author attended and 39 Orford J., op cit. Substance Abuse Treatment: 1999, 16(1), p. 55–60.
took notes which are drawn on in the following account.
40 Lindström L. “Life is short, the Art long.” Addiction: 64 Mattson M.E. et al, op cit, p. 1336.
18 Ranzetta L. The findings of Project MATCH. Report of a 1999, 94(1), p. 45–47.
conference held on 14–15 May 1998, Leeds. Greater 65 Frances R.J., Miller S.I., eds. Clinical textbook of
London Association of Alcohol Service, 1998. 41 Miller W., Babor T. Meet the MATCH makers addictive disorders. Guilford Press, 1998. See especially
conference, op cit. the editors’ preface and Alterman et al’s assessment of the
19 Connors G.J., Carroll K.M., DiClemente C.C., et al. state of US substance abuse treatment in chapter 19.
“The therapeutic alliance and its relationship to alcoholism 42 Hile M.G., Adkins R.E. “The impact of substance
treatment participation and outcome.” Journal of Consulting abusers’ readiness to change on psychological and 66 Glaser F.B., op cit.
and Clinical Psychology: 1997, 65(4), p. 588–598. Com- behavioral functioning.” Addictive Behaviors: 1998, 23(3),
ments that almost certainly the ‘alliance’ effect was artificially p. 365–370. 67 Edward Peck quoted in Community Care: 9 July 1998,
constrained by the study’s design and data analysis. p.10.
43 Hernandez-Avila C.A., Burleson J.A., Kranzler H.R.
20 Ranzetta L., op cit. “Stage of change as a predictor of abstinence among 68 Ward M., Applin C. The unlearned lesson – the role of
alcohol-dependent subjects in pharmacotherapy trials.” alcohol and drug misuse in inquiries into homicides by
Substance Abuse: 1998, 19(2), p. 81–91. Data were derived mentally ill people. 1998.
from a placebo trial of the antidepressant fluoxetine (Prozac)
69 Steadman H.J., Mulvey E.P., Monahan J., et al. “Violence
by people discharged from acute psychiatric inpatient
facilities and by others in the same neighborhoods.”
Archives of General Psychiatry: 1998, 55, p. 393–401.

1999 ISSUE 1 DRUG AND ALCOHOL 21