- Home
- Search
- Browse
- Hot topics
- Matrices
- About
- Help
- Contact
Matrix cell
Practitioners; Psychosocial therapies
Key studies on the impact of the practitioner in psychosocial therapies for alcohol dependence. Structured around Carl Rogers’ classic account of the prerequisites of effective psychotherapy.
S Seminal studies
K Key studies
R Reviews
G Guidance
more Search for more studies
Links to other documents. Hover over for notes. Click to highlight passage referred to. Unfold extra text 
S
Fundamentals of effective therapy: genuineness, positive regard and empathy (1957; free source at time of writing). In psychosocial counselling and therapy, no paper has had more influence than Carl Rogers’ formulation of the “necessary and sufficient conditions” for clients to get better, the foundation of arguably all effective substance use counselling. See also commentaries (1 2) on his work. Direct test of his theory listed below. For discussions click here and here and scroll down to highlighted headings.
S
Counsellors’ relationship style affects patients’ relapse rate (1981). US study found a strong link between higher levels of empathy, genuineness, respect, and concreteness exhibited by alcohol clinic counsellors and a reduced risk of their patients relapsing after treatment. For discussion click and scroll down to highlighted heading.
S
Empathy makes the difference (1980). Big differences in therapy content and duration did not affect the progress of US heavy drinkers. What seemed to for at least two years (1983) after treatment was the degree to which their therapists displayed “accurate empathy”. See also this assessment of the impact of empathy in psychotherapy generally (2018). For discussion click and scroll down to highlighted heading.
K
Therapist effects emerge even when subdued by stringent controls (1999). Despite exhaustive selection, training and supervision, some therapists in the landmark US Project MATCH alcohol treatment trial had on average worse outcomes (1998) than their peers, and there was enough variation (1997) in the therapeutic relationship for this to influence engagement and later drinking. Session recordings exposed reasons for variation, including the match between the therapist’s directiveness (2009) and whether the client reacts against direction, subject of a review below. Project MATCH was the “Highlighted study” in cell A2. Sub-study from the same trial below.
K
Reinforcing talk about changing drinking really does seem to promote change (2009). Micro-analysis of tapes of motivational interviewing sessions from the US Project MATCH alcohol treatment trial led to the appealingly simple and plausible conclusions that “What therapists reflect back, they will hear more of”, and that promoting talk about change promotes change itself. However, the study was not designed to establish causality. Other results from the same trial above.
K
Combine authenticity with social skills in motivational interviewing (2005). US study suggests that the quality of seeming ‘genuine’ can suffer if training mandates withholding natural responses, but also that departing from these mandates is risky unless done by a socially skilled therapist. See also an essay (2013) from Drug and Alcohol Findings based on this and other studies, arguing that ‘by the book’ is not always best way to do therapy. For discussion click and scroll down to highlighted heading.
K
Adding strategies targeting change did not improve on non-directive listening (2012). Supplementing ‘Rogerian’ (paper on his theory listed above) non-directive listening with motivational interviewing techniques directed at reducing drinking did not further help (if anything, the reverse) US heavy drinkers cut back, contradicting a similar earlier study (2001). For discussion click and scroll down to highlighted heading and also see “Are these always the important things to do?” in cell A4.
K
Can therapists be too accommodating? (2009). Rarely has counselling been so deeply analysed as in this US study involving mainly alcohol- and cocaine-dependent patients. Expected finding was that some counsellors generate good working relationships with clients which feed through to better outcomes; less expected was that the very ‘best’ relationship builders were not on average the most effective. For related discussion see “Isn’t it just a matter of being nice?” in cell B2.
K
Largest UK alcohol treatment trial finds client–therapist relationship related to post-therapy drinking (2015). Offshoot of the UKATT study (main results highlighted in cell A4) comparing therapy based on motivational interviewing with one focused on reconstructing social networks. Primary issue for this sub-study was whether a better client–therapist working relationship was associated with a greater subsequent chance of remission. Overall it was, but significantly only when the relationship was assessed by client rather than therapist – and when the treatments were considered separately, only for motivational rather than network therapy. For discussion click and scroll down to passage highlighted below.
R
Common core of effective therapy: therapeutic relationships (American Psychological Association, 2018). Includes, but not specific to, substance use. Introduces, cites and synthesises finding from 16 reviews (also analysed for the Effectiveness Bank) of the psychotherapy literature based on the understanding that therapeutic change is generated not only by technical interventions, but by the ways therapists relate to clients, like forming a therapeutic alliance (related review below), being empathic (related study above), and appropriately adjusting to the individual (related review below). Still valuable is an earlier version (2011) of this article which integrated findings on how to adapt therapy to the individual client (work listed below) and on counterproductive behaviours like being confrontational. See also a broader practice-oriented interpretation (2014) of the research from same lead author which draw on these reviews.
R
Therapists who form good therapeutic relationships have better outcomes (American Psychological Association, 2018). One of the (see above) US American Psychological Association task force reviews. Supports the argument that “a good working relationship is an important determinant of treatment success, and that nurturing, maintaining, and as needed, re-establishing such a relationship, are core tasks not just in psychosocial therapies, but in treatment generally”. Earlier an advanced synthesis of research findings (2012; free source at time of writing) from some of the same authors had confirmed that some therapists consistently develop stronger relationships and have better outcomes.
R
Adapt to the client (American Psychological Association, 2011). Includes but is not specific to substance use. US American Psychological Association task force whose overall report is listed above judged that adapting psychotherapy to the client’s reactance/resistance, preferences, culture, and religion/spirituality demonstrably improves effectiveness. Related review below.
R
Some clients like to lead, others to be led (2006). How directive the therapist is during treatment is one of the strongest and most consistent influences on outcomes. There is no ‘right’ degree of directiveness; it all depends on how the client reacts. Related review above.
G
Addiction counselling competencies ([US] Substance Abuse and Mental Health Services Administration, 2008). Includes competencies associated with positive outcomes and the knowledge, skills, and attitudes all substance use counsellors should have. First step is to “Establish a helping relationship with the client characterized by warmth, respect, genuineness, concreteness, and empathy.”
G
What makes a good group therapist? ([US] Substance Abuse and Mental Health Services Administration, 2005). US consensus guidance on the different types of groups, how to organise and lead them, desirable staff attributes, and staff training and supervision.
G
What makes a good case manager? ([US] Substance Abuse and Mental Health Services Administration, 1998). US consensus guidance including the staff skills, knowledge and attitudes needed to fulfil the case management role orchestrating the range of services which may be needed to promote lasting recovery and broader life improvements.
more Search for all relevant Effectiveness Bank analyses or search more specifically at the subject search page. Also see hot topics on treatment staff and matching alcohol treatments to the patient.
Close Matrix Bite 
Links to other documents. Hover over for notes. Click to highlight passage referred to. Unfold extra text
What is this cell about?
Every treatment involves direct or indirect human interaction, but this cell is about treatments in which interaction is intended to be the main active ingredient – ‘psychosocial’, or more colloquially, ‘talking’ therapies. Based on varied understandings of how dependence arises and how it can be overcome or ameliorated, they attempt to change how the patient behaves via their beliefs and attitudes, how they relate to others, and how others relate to them, or directly by ‘shaping’ behaviour through rewards and sanctions.
These differences between therapies have been tested and contested and occupied the lion’s share of research time, but as long as it is a well structured, bona fide treatment which ‘makes sense’ to patient and therapist, the ‘common factors’ shared by supposedly distinct therapies (on which see “Where should I start?” in cell A4) seem more critical to their success.
For patients and researchers, how the therapist relates to the client is the main embodiment of the common factors shared by therapies, and the most salient way they affect engagement and outcomes. In cell B2 we have seen this generally across treatment and in cell B3 respect of medical treatments. Unsurprisingly, the evidence is stronger still for psychosocial therapies, where the structured enactment of the therapist–client relationship is the treatment, forcing attention to it even in studies (listed above) designed to minimise such influences.
In this cell we focus on client–worker relationships, and on whether some practitioners are more successful because they more strongly forge the right kind of relationships. Before moving into that territory, a reminder that therapists and counsellors typically work in organisations which limit or enhance their ability to maximise client progress, an issue explored cell B3 in the context of medical treatments. The same issue will emerge in this cell from a study (described in the supplementary text towards the end of this ‘bite’) which identified significant relationships between abstinence and the characteristics of the treatment organisation, but not (once these characteristics had been taken into account) between abstinence and the client–worker relationship.
Where should I start?
With arguably the most fertile source for practice and research in psychosocial therapy for substance use problems – Carl Rogers’ classic formulation listed above (free source at time of writing) of the “necessary and sufficient conditions” for therapeutic progress: the communication of genuineness; unconditional positive regard – no ‘ifs’ or ‘buts’ qualifying the therapist’s acceptance of the patient; and an accurately empathic understanding of clients in need of help to align their actions, thoughts and self-perceptions. The ‘seminal’ credentials of this paper are indicated by its being reprinted 50 years later (the version listed above), and by the fact that throughout the matrices (including practically every entry in the current cell) you will find these qualities continue to emerge as significant in engaging people with substance use problems in effective treatment.
Introduction to Carl Rogers’ seminal paper, “The necessary and sufficient conditions of therapeutic personality change”
Despite his focus on the universals of relating to clients, Rogers did not dismiss specific techniques like offering interpretations of the roots of the client’s feelings and behaviour, exercises weighing up the pros and cons of change, analysing what triggers unwanted behaviour, and training in social and self-control skills. In his schema, these were not active ingredients in themselves, but also not trivial, because it is partly through such techniques that relational qualities like positive regard are communicated – and communicating these was seen as one of the essentials of effective therapy. From the problem drinkers investigated in a seminal British study highlighted in cell B3 of the Alcohol Treatment Matrix, and from the comments of drug and alcohol treatment clients, we know that these qualities – especially unconditional positive regard – are also what substance use patients seek in a helper, and that they can promote access to and retention in treatment. The study highlighted in the next section also shows they can help reduce the risk of patients relapsing after treatment. Across psychotherapy they have (review listed above) stood the test of time, gaining greater acceptance and prominence rather than becoming overtaken by alternative theories.
An obituary records that Rogers’ theories emerged from an iconoclastic figure who “did not care about appearances, roles, class, credentials, or positions, and [who] doubted every authority including his own,” and from someone (for his time) distinctively devoted to subjecting theory to scientific tests of validity in practice. Produced in 1965 as training aids, you can see this figure in therapeutic action in the (within psychotherapy circles) legendary ‘Gloria’ videos of his encounter with a psychotherapy client, who was also filmed being treated by two other leading therapists of the time. See these accounts (1 2) for more on the videos and their importance in the history of psychotherapy.
In various works Carl Rogers’ insights were further operationalised and translated into practice, among which in 1967 was Charles Truax and his colleague’s influential book, Toward Effective Counseling and Psychotherapy: Training and Practice. Based on work undertaken with Rogers himself, it argued that “genuineness or authenticity is most basic to a relationship”, a quality explored below under the heading, “Being genuine sometimes means breaking the ‘rules’”. According to Truax’s book, having established this foundation, the therapist or counsellor communicates warmth and respect for the client and proceeds to the work of therapy via their “moment to moment empathic grasp of the meaning and significance of the client’s world”.
Truax and his colleague’s understanding of ‘genuineness’ entailed “openness to experience” rather than defensively retreating behind a facade or role, among which for therapists is that of the ‘technical expert’ in their profession. In various guises, openness to experience later emerged as a quality characterising effective practitioners and effective organisations in addiction treatment. That link seems also to extend to the therapists engaged by the UK’s drive launched in 2008 to improve access to mental health care. To find out more about this research unfold the supplementary text towards the end of this ‘bite’ to read about the study, numbered 4 in the list.
Motivational interviewing is the guise in which Carl Rogers’ insights will be most familiar to addiction therapists. A direct inheritor of his person-centred focus, it differs in the therapists’ “intentional and strategic use of questions, reflections, affirmations, and summaries to strengthen the client’s own motivations for change” – though as we’ll see below, these strategies do not necessarily augment outcomes.
Highlighted study
The case for empathic, responsive and socially skilled therapists who build strong relationships hardly seems in need of testing, any more than (as a famous article put it) the case for strapping on a parachute before leaping from a plane. But in fact, though some counsellors and therapists undoubtedly embody these virtues more than others, the evidence (cantered through below) that as a result their patients do better is far from cast-iron solidity. Lack of solidity could be partly due to methodological obstacles. Primary among these is that though this would be the ideal trial for establishing causality, it would be unethical to deliberately and at random allocate vulnerable and troubled people to non-empathic, phony, hostile and incompetent versus better counsellors, just to see what happens. Without this, pinning down whether the therapist’s relational qualities actually cause outcome differences is severely hampered.
Accurate empathy
It is important for psychotherapists to make efforts to understand their clients, and to demonstrate this understanding through responses that address the perceived needs of the client. The empathic therapist’s primary task is to understand experiences rather than words. Empathic therapists do not parrot clients’ words back or reflect only the content of those words; instead, they understand overall goals as well as moment-to-moment experiences.
Therapist responses that accurately respond to and carry forward the meaning in the client’s communication are useful. These responses might: convey understanding of client experience; validate the client’s perspective; try to bring their experience alive using rich, evocative, concrete, connotative language which often has a probing, tentative quality; or attempt to get at what is implicit in clients’ narratives but not yet articulated – guesses grounded in what the client has presented.
From the Effectiveness Bank analysis, Evidence-based psychotherapy relationships: Empathy.
However, you can randomly allocate clients to different therapists, some of whom will happen to be better than others. Go back in time to an era when trials were less tightly controlled, and you might find the wide variation in therapist competence probably seen even today in the ‘real world’ beyond controlled trials. Such a study would effectively (since these qualities tend to go together) randomly allocate patients to counsellors with high versus low levels of empathy, understanding, and warmth. That’s what a seminal US study (listed above) did, and it remains the most convincing test of the effect of these qualities on drinkers seeking treatment in the normal way. As noted in cell B2, the study found a strong link between counsellors’ empathy, genuineness, respect, and ‘concreteness’ (seemingly an amalgam of good communication and accurate empathy 
panel) and a reduced risk of their patients relapsing after treatment. Here we can add that these are the very qualities Carl Rogers expected (section above) to be positively related to patient improvement. If you read our analysis of the study, you will see that it cannot be said on its own to settle the issue of the impact of Rogerian qualities on substance use problems – but with others, it makes a persuasive case (free source at time of writing).
Among these other studies is another seminal paper listed above which randomly allocated clients to therapists. In this case their degree of “accurate empathy” was rated by observers located behind one-way mirrors. The study was led by Bill Miller, originator of motivational interviewing, the main contemporary inheritor of Rogers’s legacy. Conducted in the pre-motivational interviewing era, it was intended as a trial of alcohol therapies differing in intensity and in the breadth of issues addressed, all benchmarked against self-help. ‘No significant difference’ in drinking outcomes was the disappointing conclusion, but the “surprised” researchers stumbled across a “serendipitous finding” which drew their attention in a very different direction, helping to found motivational interviewing. While seemingly fundamental differences in the extent and content of therapy made no difference to drinking, the style in which the therapies were delivered was strongly related to drinking. Across the admittedly small caseloads, the greater the average degree of accurate empathy exhibited by therapists over the eight months of the study, the more likely their clients were to have substantially reduced their drinking by the end. The range was from 100% of the clients of the most empathic therapist improving to 25% of the least empathic – a variation which would have been ironed out in studies with more selected, trained and sifted therapists.
Issues to consider and discuss
Is Rogerian listening really all that’s needed?
Read Carl Rogers’ classic paper and our comments on it in the “Where should I start?” section above, and you will see that he never said empathic and non-judgemental listening was all that was needed to maximise client progress. Nevertheless, at least two studies have tested this interpretation of his model by adding extra elements to a Rogerian approach to see if they improve outcomes – and they came to seemingly opposing conclusions which force us to think more deeply and come to a more nuanced understanding of Carl Rogers’ model.
The extra elements added by the studies were those which transform (free source at time of writing) the non-directive Rogerian approach into directive (but not explicitly directive) motivational interviewing. Though motivational interviewing “rests explicitly in Carl Roger’s approach”, it sharpens it into a goal-directed strategy which “presupposes that the therapist prefers one outcome over another” and tries to get to that outcome by “responding preferentially to language that indicates a desire, ability, reason, need, or commitment from the client” to make the desired changes. In different ways, both the studies below stripped out these directive elements to see whether they really did make a difference. See what you make of them.
First take a look at our analysis of the most recent of the studies. Conducted in the USA, it found that adding the directive elements of motivational interviewing ( figure) to Rogerian reflective listening did not improve drinking outcomes. In greater detail, all the patients received feedback from research staff on their assessment results, which classified their drinking in stages up from low risk to possible physical dependence. For some (the ‘self-change’ patients) this was followed by an instruction to try to change their drinking on their own, after which they were left to their own devices, never having seen a therapist. The remainder were allocated to four therapy sessions of either motivational interviewing or Rogerian reflective listening – effectively, motivational interviewing stripped of its more directive techniques intended to provoke a commitment to reduced drinking. The result was unexpected: improvements all round, but no significant differences on any of the measures of drinking or its consequences, not even between self-change patients and those offered some kind of therapy.
If anything, basic Rogerian reflective listening bettered motivational interviewing
On each measure, if anything Rogerian reflective listening bettered motivational interviewing and more so ‘self-change’. However, these surprising findings need to be put in context. The study’s patients were relatively stable, moderately dependent drinkers. To join the trial they had to be aiming for moderation rather than abstinence, socially stable, and not severely mentally ill or seriously involved with other drugs. Typically they were in their 30s and 40s, employed, well educated, and never before treated for drinking problems. If the study’s version of non-directive listening was all that was needed, that was the case for a caseload much less severe than found in some UK treatment services (1; 2).
About a decade before, a similar sample had been recruited in New Zealand for a similar study (listed above), but its findings were the opposite to those of the US study: frequent heavy drinking was further significantly reduced when motivational interviewing techniques were added to non-directive listening. Without motivational interviewing’s directive elements, around 64% of patients engaged in this pattern of drinking over the six-month follow-up; with these elements, just 43%. Non-directive listening could not be shown to have improved on results from doing nothing except the basics provided to every patient – feedback on the extent of their (excessive) drinking over the past six months, diagnosis of the severity of their drinking, and advice to cut down to within national guidelines; in both sets of patients around two-thirds later continued to drink heavily.
Why the difference between the results of the trials? One suspect is that the New Zealand study’s non-directive listening was very non-directive – to the point that if patients wanted to talk about the weather, that was fine. Here’s how the researchers described this, perhaps to both patients and therapists, seemingly strange ‘therapy’: it “consisted of nonstrategic reflective listening … in which subjects were invited to talk about anything they wanted, not necessarily issues related to drinking. The direction of content throughout the treatment was intentionally left for subjects to determine. [Non-directive listening] was Rogerian counseling cut back, restricting therapist responses to the barest minimum (ie, nondirective reflective listening while maintaining rapport) and retaining the same therapeutic stance irrespective of the content being offered by the subject.”
In everyday language, no matter what the patient said or how they behaved, their distress or their delight, the therapist acted as a friendly, absorbent, but essentially unresponsive sounding board. It might have puzzled and disappointed patients who (unlike in the US study, which used ads to solicit participants) had all gone to a specialist substance use clinic seeking help with what was diagnosed as dependent drinking. Recall the conclusion to which Project MATCH (reports on this study listed above) researchers were driven, that treatment is a “culturally appropriate solution to a socially defined problem”. Could it be that such a stance violated patients’ culturally shaped expectations of what ‘treatment’ should look like? Perhaps too, the limitations placed on non-directive listening stopped therapists communicating empathy and acceptance, one of Carl Rogers’ essentials for effective therapy, and forced them to respond to patients in an unnatural way. If that was the case, the effect would probably have been to undermine perceptions of the therapist as being “genuine” – another prerequisite of effective therapy explored in the following section.
Being genuine sometimes means breaking the rules
Imagine yourself as a psychosocial therapist influenced (as most in the substance use field will directly or indirectly have been) by the Rogerian principles outlined in the “Where should I start?” section and their elaboration into motivational interviewing. You know you are not supposed to insist clients ‘must do’ something, even less to warn of the consequences if they don’t, and still less to express disapproval of their choices, but biting your tongue just doesn’t feel right – doesn’t feel like you are being you. Yet you also know you are supposed to be you – to be “genuine” as Rogers put it, not put on an act. There seems a conflict between these demands, all of which Carl Rogers and/or motivational interviewing’s theorists saw as keys to effective therapy. What should you do?
Avoiding warnings can seem as uncaring and unnatural as suggesting to this pedestrian heading towards a pit that they consider the pros and cons of proceeding, but in the end it is up to them; the natural and caring response is to shout, ‘Stop.’
For guidance, turn to a study listed above of the training of addiction counsellors and clinicians, the implications of which are most easily absorbed from a brief, informal account by Drug and Alcohol Findings. Read at least this.
Are you convinced by our interpretation that (in the context of a caring relationship and a socially skilled therapist) “warning and directive advice which conveys and comes from concern for [the patient’s] welfare and respect for [them] as an equal” can be beneficial, and that withholding such comments can make you feel and sound less than genuine? Was it naive to reassure counsellors that “Everyone knows the difference between warning, advice and concern which conveys and comes from care and respect for one as an equal, and that which comes from and conveys accusation, denigration, and an attempt to exert control.” Is this clear departure from a Rogerian stance and motivational interviewing’s ‘rules’ too risky, easily seeming to the patient to represent a degeneration into negativity and confrontation? If so, we know from a review listed above that the results would often be counterproductive.
At this stage, it might help to remind yourself of the ‘small print’ of the study on which these extrapolations are based: that only when the counsellor was relatively socially skilled did ‘breaking the rules’ in these ways enhance the effect their skills had on client engagement. Like a skilled barber with a cut-throat razor, they were able turn what could have been inadvertently harmful into something which could help get the job done by strengthening the therapeutic relationship.
We should also distinguish the meaning of ‘genuine’ here – being ‘true to yourself’ – from being honest not in rather a deep way about how you represent yourself and your feelings, but in a more surface way about the treatments you offer and their chances of success. A dose of perhaps unjustifiably enthusiastic support can augment (as discussed in cell A3) the effects of both medications and placebos, but may be considered unethical and if found out a risky tactic which could undermine the patient’s faith in the clinician.
If therapists are influential, why don’t more studies register their effects?
Across psychotherapy more effective therapists “generally form better alliances with their patients and have better facilitative interpersonal skills, and provide an emotionally activating relationship”. Frequent reports from patients that a particular therapist, counsellor or keyworker was the catalyst for their recovery suggest that holds too for the substance use field, yet these comments clash with results from trials which often find no significant signs that practitioners differ in effectiveness. Are the researchers missing something “obvious” (free source at the time of writing) to patients and clients
Looking into these difficulties offers a case study in the limitations of conventional scientific methods
and patently observable in normal practice? With random allocation to people trained or selected to be bad versus good therapists ruled out on ethical grounds, proving therapists have an impact on substance use outcomes and establishing the reasons are not straightforward. Looking into these difficulties offers a case study in the limitations of conventional scientific methods, ill-suited in this case for confirming that what seems obviously valid and important, really is. We’ll take a canter through these limitations, always bearing in mind the possibility that when the research does not find an effect, this is because there really is none.
One limitation is that therapist effects are often obscured by the control researchers exercise over trials, including over therapists. Data on these therapist effects is commonly gathered as a by-product of a trial designed to evaluate an intervention, not the interventionists. To test whether a psychosocial intervention can work, it has to be given a good chance to succeed. Usually that means selecting highly competent therapists and/or training them to meet the study’s standard for delivering the intervention. Anything less risks alienating patients to the point that many disengage from therapy and from the study, leaving researchers with a partly untreated, small and probably unrepresentative follow-up sample, exposed to a sub-optimal implementation of the programme. No surprise then that whichever these highly competent therapists patients have been assigned to, on average they improve to roughly the same degree. The surprise is that sometimes therapist effects nevertheless emerge, even in (document listed above) possibly the most highly controlled alcohol treatment trial ever conducted.
Where therapists do appear to have made a difference to client outcomes, a strong candidate for how this happens is via the relationships they establish with clients. At this point, disentangling what caused what becomes a major obstacle to verifying therapist effects: rather than relationships causing therapeutic progress, it could be that patients who are in any event going to do well – or are already doing well when the relationship is assessed – form stronger relationships with their therapists and vice versa, each tending to be more appreciative of the other when their goals are being achieved. In this scenario, a good relationship is merely a by-product of a good prognosis. However, across psychotherapy there is evidence (free source at time of writing) that more is going on – that relationships really do account for part of the variation in how well clients do.
Establishing a link between the client–worker relationship and outcomes may also (if findings on the treatment of depression apply) require an unusually large study and unusually complete monitoring of those relationships, offering sufficient assessments for a reliable average to be computed. It is also critical to take these measures early in therapy before the relationship begins to reflect rather than promote client progress.
Then there are complexities which confound simple analyses. For example, usually researchers test whether the stronger the relationship, the better the outcomes. But what if the association between the two is not linear, but ‘curved’ – relatively weak and very strong relationships both being associated with poorer outcomes? If this is the case, linear analyses might fail to find an association, even if there really was one. If that seems fanciful, look back at the evidence presented in cell B2 there may well be a curve in the association between relationships and outcomes.
A further complexity is that therapeutic relationships matter more in some circumstances and for some people than others, either because they have a greater influence on outcomes, or because there is a greater range than usual in the quality of those relationships – not all tending to be average, but stretching from outcome-eroding poor to outcome-boosting excellent. Then rather than casting doubt on the existence of relationship effects, inconsistent findings are only to be expected. An example emerged from a large and careful study of the US national drug treatment system, which suggested that relationships matter more for previously treated clients and patients than those for whom this is a new experience. Explore the intriguing implications of this study in the corresponding section of the Drug Treatment Matrix by clicking on the eye-opener icon when the new web page opens.
In what was largely the US context, further evidence that who is being treated matters came from a synthesis of research findings (free source at time of writing) which found that the link between therapeutic relationships and outcomes was weaker the greater the proportion of a study’s sample being treated for problems due to their use of illegal drugs – not the case in respect of the proportion treated for problem drinking. Also, the link was weaker the more the sample consisted of ethnic minorities rather than the ‘white’ populations dominant in the countries where the studies were conducted. These two proportions overlapped so much (ie, people being treated for their use of illegal drugs tended to be ‘black’ or otherwise categorised as racial minorities and vice versa) that when one was adjusted for the other was no longer related to a weakening of the link between therapeutic relationships and outcomes.
Another source of inconsistent findings is that therapeutic relationships can matter more for some psychosocial therapies than for others. Just such a finding emerged from the largest ever UK alcohol treatment trial. Main results from UKATT were highlighted in cell A4. Here the focus is on an offshoot analysis listed above of therapeutic relationships among the minority of clients for whom these were assessed and who completed relevant follow-up measures nine months after therapy ended. The puzzle is not that therapeutic relationships were significantly associated with outcomes, but that this held only for a therapy based on motivational interviewing, not one focused on reconstructing social networks. Our “data-informed speculation” is that this was due to greater variation in the influence of the relationship in the network therapy; see if you agree after unfolding and reading our explanation.
Close supplementary text
Client and therapist views of their relationship were assessed after the first attended therapy session, so only clients who attended at least one session could be included. Importantly the analysis adjusted for pre-treatment prognostic factors, helping to eliminate the possibility that the results were not due to any influence exerted by the solidity of the therapeutic relationship, but that instead clients who were in any event going to do well formed better relationships and had better outcomes, without one having a hand in causing the other. Overall this sub-study’s results were consistent with better relationships as experienced by the client (but not by the therapist) promoting moderation in drinking during treatment and commitment to reducing drinking after therapy ended, leading partly via these ‘mechanisms’ to greater moderation in drinking nine months later. But when the treatments were considered separately, this relationship held only for therapy based on motivational interviewing rather than one focused on reconstructing social networks. Had the latter been the only treatment, UKATT would have added to the tally of studies registering no statistically significant links between therapeutic relationships and drinking.
Why this link was lacking in the network therapy can only be the subject of data-informed speculation. The ‘data’ is that on every measure of drinking there was greater variability in the relationship between the client’s perception of the relationship and outcomes among network than among motivational clients. The ‘speculation’ is that this was due to greater variability in the influence of the clients’ relationships with network therapists; variability in outcomes reduces the chances of statistically significant findings, even if on average impacts were similar. In other words, for some clients, how they experienced their working relationship with their network therapist was an important influence on later drinking, for others much less so, while with motivational therapists the influence was more even.
One plausible explanation is that this was because network therapy hinged on (free source at time of writing) enlisting supportive contacts to help moderate drinking. A pilot study had revealed that an appreciable proportion of clients “had difficulty engaging” such a social network. In these circumstances the therapy would have had to focus more on the individual client, helping them develop the skills to recruit a suitable network and possibly too offering support otherwise lacking in their lives. For these clients, how well they worked with the therapist would be more critical than for those whose extended sources of support and ‘treatment team’ were more readily recruited. This cause of variability in the influence of the therapeutic relationship would have been absent in motivational therapy, meaning that the apparent influence of the working relationship with the therapist was more uniform, and statistically significant findings easier to register.
Close supplementary text
Then there is how we assess the client–therapist relationship. The varied measures usually largely reflect the degree to which client and therapist collaborate in the work of therapy – they are ‘task-oriented’. In turn this collaboration is based partly on how they get on simply as human beings encountering each other as they might in ‘real life’ outside the clinic or consulting room. A major component of this so-called ‘real relationship’ is how genuine you feel the other person is being, seemingly important in therapy generally and in substance use treatment in particular (see section above). Another component is seeing and reacting to the other person as they really are. Reflecting the concept’s psychoanalytic roots, this is contrasted to seeing them as an unreal projection of your own unresolved inner conflicts, but it could also be that one’s insight is obstructed by projecting on to them the characteristics of the clients envisaged in therapy manuals, failing to appreciate that for this individual, a departure from those manuals may be called for: 1, listed above; 2, listed above. If components of the ‘real relationship’ were given greater prominence, might this more rounded assessment of the client–therapist relationship emerge more strongly as an influence on outcomes? – as it does in psychotherapy as a whole.
Perhaps we have said enough to demonstrate that relationships are complex and their impacts on treatment outcomes not straightforward. Searching for proof of these impacts using methodologies designed for the simplicity of testing an intervention versus no/alternative intervention, and in studies where relationships are not the main interest, might be likened to probing for signs of the ‘big bang’ origins of the universe with a toy telescope pointed at the ground.
The research has been like probing for signs of the ‘big bang’ with a toy telescope pointed at the ground
You might not see any signs, but that doesn’t mean the ‘big bang’ was a trivial event without consequences. Perhaps this is why a review of research on the therapeutic alliance in the treatment of substance use found client–worker relationships only inconsistently related to substance use outcomes. The same review found relationships were more consistently associated with engagement and retention in treatment, suggesting a less sanguine interpretation of the findings (advanced in cell B2 of the Drug Treatment Matrix) that better relationships make clients want to stick around, but do not make them better – though of course, for some treatments (especially those based on medications), ‘sticking around’ is often a prerequisite for effectiveness.
The supplementary text (click to unfold ) uses modern-day studies to illustrate the difficulties of finding therapist effects, and even more so of pinning down what the better therapists are doing which makes them better. But these studies also show signs of such effects can be found in the links between therapeutic relationships and substance use outcomes, and more so, patients’ engagement with therapy and satisfaction with the therapist. Among the findings, you will come across that same “openness to experience” which became seen in the 1960s (see passage in “Where should I start?” section above) as the foundation of effective therapy and counselling.
Close supplementary text
1 Example one derives from the large US COMBINE trial spotlighted in cell cell A3 because it demonstrated the power of a placebo ‘dummy’ medication. Primarily it was intended to assess the relative impacts on alcohol-dependent patients of different medications allied with different psychosocial support programmes. In the process it gathered unusually good data for assessing therapist effects, because therapy sessions were recorded and rated by observers for (among other things) the degree of empathy demonstrated by therapists. Tapes from 38 therapists and 700 clients revealed (free source at time of writing) that judged by the amounts their clients drank at the end of treatment, the performance of the therapists significantly varied. Some had on average better results, others worse, though with just 11 clinics, no attempt was made to rule out organisational factors – which the second example (below) shows can create the illusion of therapist effects.
Next the researchers tested whether the average degree of empathy which characterised a therapist across their entire caseload accounted for varying outcomes. The answer was negative; some therapists were more empathic than others, but that did not account for why some patients did better. Another analysis instead asked whether the more empathy a therapist showed to an individual client, the less that client drank. This time the answer was positive. However, now we are assessing not the influence of a characteristic of a therapist, but of their interaction with an individual. It means (for example) the possibility could not be excluded that rather than empathy fostering drinking reductions, clients who did what therapists wanted by moderating their drinking elicited greater empathy from the therapist. As an aside, neither the degree of experience of the therapists, nor whether their ethnicity or sex matched those of their client, bore any significant relationship to client drinking – common findings (1; 2, free source at time of writing).
But before we conclude that employing empathic therapists makes no difference, recall that this study was intended to test interventions, not therapists. To level the playing field, before being allowed into the study therapists had been “rigorously screened for their use of empathic listening skills”, were then “extensively and explicitly trained in the interpersonal context” of the intervention, and finally “monitored in their expression of empathy as the trial progressed and … red-lined (ie, stopped from taking clients) if empathy ratings were unacceptably low”. The variation in empathy to be expected between therapists in normal practice had been ironed out to a small wrinkle, and was (or was no longer) a salient variable affecting therapist performance in terms of drinking among their clients.
2 Second example was again a by-product of a large US randomised trial, this time of providing feedback to counsellors on the strength of their relationships with their clients and how well they were doing in reducing their substance use. For reasons explored in our account of the study, the feedback could not be shown to have promoted client–counsellor relationships nor patient progress.
Though it entailed abandoning the level playing field created by randomisation, the researchers searched for other influences which did affect the patients. Their analysis (free source at the time of writing) capitalised on the fact that the study required repeated measurement of how clients saw their relationships with counsellors, enabling researchers to investigate whether clients did better if they had been allocated to counsellors who on average built stronger relationships. Had they conducted a less sophisticated analysis, they might have concluded this was the case, because which clinic a counsellor worked at was related both to the strength of the client–counsellor relationship and to substance use outcomes, easily creating the illusion that it was the counsellors who were the active ingredient.
But the analysis was able to tease apart these influences and assess whether within each clinic, counsellors who typically built stronger relationships had better client outcomes. Essentially, there was no reliable evidence that this was the case. In contrast, when the analysis focused on the clients of an individual counsellor, the better their relationship, the more likely a client was to be abstinent from drink and/or drugs. Perhaps this was a sign that client–counsellor pairings who work well together are most therapeutic, but perhaps instead that clients who are going to do well anyway are more appreciative of their counsellors, and vice versa. The latter interpretation means good relationships would be a by-product, not a cause, of client progress.
There were several reasons why therapist effects might have been obscured or absent in this trial, not least that the therapy was conducted in groups, with all the extra influence exerted by other group members. But it did show that other analyses which failed to account for organisational influences could have falsely allocated those influences to individual counsellors or therapists.
3 A third example of the many we could have selected takes us to Finland and the work of Pekka Saarnio, who has done much to highlight the attributes of effective counsellors (see this study discussed in cell C2). For current purposes, the study focused here had the great advantage that within each of seven clinics, it allocated new clients to the 33 therapists in the study at random. Nearly all their clients were problem drinkers, though most had already been abstinent for a time before the study made its baseline assessments. Of most interest is the relationship between the therapists to whom patients were allocated and their abstinence from drinking six months after treatment started. Once initial abstinence levels had been taken into account, it was zero or near zero – a remarkable finding, because these were simply the clinics’ normal therapists, neither specially selected nor specially trained, and allowed to implement their own preferred programmes. With this freedom to differ, how could they be indistinguishable in effectiveness? However, on one measure they were clearly distinguishable: the degree of satisfaction with them expressed by their clients.
Here we can see a possible reason for the finding in a synthesis of relevant research that the client-worker relationship is more strongly related to engagement and retention in treatment than to substance use outcomes. If the main driver of recovery is the patient, given a half-decent counsellor or therapist – and in the Finnish study, from the clients’ points of view, they were all more than half decent – the patient’s progress will largely reflect their own resources and motivation. Nevertheless, as in everyday life, some therapists will be liked more than others, and patients will be happier to extend their contact with the more likeable personalities. Where we are likely to see an impact on substance use is from therapists who are well below half decent, to the point where the client’s impetus is obstructed. That returns us to the importance (discussed in cell A2) of not doing the wrong thing – from some studies, seemingly more influential than being very good at doing the right things.
4 Stepping outside the substance use field offers a further perspective on therapist effects. The step is to the UK’s drive to improve access to mental health care via the Improving Access to Psychological Therapies (IAPT) programme launched in 2008. It entailed recruiting “Psychological Wellbeing Practitioners” to extend low intensity, cognitive-behavioural support to a greater number of depressed and anxious patients than previously reached. The interventions mounted by the practitioners were in line with guidelines set by the National Institute for Health and Care Excellence and followed set protocols.
According to a study (free source at time of writing) of their impacts, they were a varied set of people, who also significantly differed in the degree to which their patients experienced improved psychological health: patients seen by the top five of the 21 practitioners were over twice as likely to reliably improve on measures of depression and anxiety. What characterised the seemingly most effective practitioners was their “proactivity” in developing their skills though online research, observing others in clinical practice, and actively participating in supervision. Effective practitioners also took pains (so they said) to explain the rationale for the programme to clients, and were confident enough to adapt it to the individual. Their supervisors described the best as open to discussing the difficulties of their work, an openness to learning not encountered among the least effective practitioners. Slipping across to the Drug Treatment Matrix, we can see a parallel here with findings suggesting that practitioners and services characterised by openness to experience, learning and ideas, most effectively engage their clients. For the researchers, their results “challenge the notion that protocol-driven therapies are wholly uncontaminated and unadulterated by the skills of the practitioner delivering the intervention”. In other words, you may tell a therapist what to do, but unlike robots they will do it their own way, and that will affect effectiveness.
Close supplementary text
Search back through your experiences as a patient, client, therapist, counsellor, keyworker, adviser or advised, supporter or supported, and use these experiences as a prism through which to assess the adequacy of the research. If somebody did have a profound effect on you, or you had that effect on someone else, how might this have been demonstrated – how could it have been proved to scientific standards?
Thanks for their comments on this entry to David Skidmore based in England, former probation officer, addiction counsellor and regional manager with the National Treatment Agency for Substance Misuse. Commentators bear no responsibility for the text including the interpretations and any remaining errors.
Close Matrix Bite
Open Matrix Bite guide to this cell 
