Talk that works: the rise of cognitive behaviour therapy

The effective ones are as good as drugs

Some people who are troubled seek referral to a counsellor or psychotherapist. “It helps me to get things in perspective,” they say, “It really works.”1 Are such “treatments” specific—that is, do they produce improvement faster than could be expected from natural recovery or from talking through problems with a supportive friend? Judging by advertisements in the local press, a wide range of therapies is now on offer, from plain psychotherapy to exotic varieties like “astrological archetypal counselling.” While people can spend their own money on whatever they wish, just what psychotherapeutic procedures should responsible doctors prescribe for their troubled patients, once they have made the appropriate diagnosis?

Cochrane popularised the use of the randomised placebo controlled trial as the preferred means of deciding which treatments were specifically effective for which disorders, and governments and other funders of health care now require data from a series of such trials before deciding which new drugs they will license and subsidise. Searching the mental health literature, and setting a threshold for effectiveness, our group has recently completed reviewing those disorders in which a specific psychotherapy exceeded that threshold, and so were proved to be more effective than placebo treatment for the same disorder.2 3 The review did not seek to examine every published study, only enough to determine whether evidence of efficacy above the threshold existed. The threshold required that the findings should be independently and repeatedly replicated so that the combined results were superior to the results of four minimum sized trials for that level of effectiveness. Strong treatments could therefore be accepted with a smaller total number of subjects in the active cells (30 would suffice for an effect size4 of 1.5 SD), whereas weaker treatments would need to be supported by trials that included a larger number of subjects (more than 200 if the effect size superiority over placebo was only 0.5 SD).

Treatments for six common disorders met such criteria. In generalised anxiety disorder cognitive therapy following the methods of Beck,5 with the addition of relaxation and graded exposure if required, produced an average effect size of 2.07 in four trials totalling 57 patients.2 In obsessive compulsive disorder exposure to feared situations and response prevention of rituals6 produced an effect size of 1.79 in four trials with 33 patients.2 3 In panic disorder and agoraphobia, cognitive behaviour therapy for the control of anxiety and catastrophic thoughts and exposure to both feared situations and panic symptoms produced an average effect size of 1.26 in five trials with 73 patients.2 7 8 In schizophrenia the addition of behavioural family intervention programmes (education about disorder and treatment, goal setting, and problem solving) produced an effect size at two years of 1.16 over standard drug treatment in three trials with 83 patients.3 9 10 In depression nine studies (170 subjects) using variants of Beck's cognitive therapy generated an average effect size of 1.07.3 11 12 13

The cognitive behavioural approach is not new. Twenty years ago five complex behavioural programmes for alcohol dependence had shown a mean effect size of 0.89 with 102 subjects.3 14 The power of the cognitive behaviour therapies in these six disorders is considerable, certainly equal to the power of the standard drug treatments for depression, anxiety, and schizophrenia. If these psychological treatments had been drug treatments they would have been certified as effective and safe remedies and be an essential part of the pharmacopoeia of every doctor. As they were not developed by profit making companies, and thus are not marketed or promoted, their use often languishes. What then, characterises these cognitive behaviour therapies?

Instead of being concerned with possible causes of illness, cognitive behaviour therapies focus on teaching people how to control their present complaints of disturbed emotions, thoughts, and behaviours. Relaxation and control of hyperventilation are used to combat hyperarousal and panic, cognitive therapy to correct dysfunctional thinking, and graded exposure to feared situations to eliminate avoiding and self defeating behaviours. Structured problem solving teaches people to analyse their problems and to plan and implement the best strategies for change. Most therapists use elements of all four techniques, encouraging patients to take responsibility for managing their own disorder. The details of the treatment programme are discussed with patients before therapy begins, each session is planned, and therapy is usually time limited to achieve a specified outcome. Symptoms, disability, and risk factors are measured regularly throughout the whole therapy. Therapy programmes are usually specified in procedural manuals. The benefits from these techniques are long lasting. As well as reducing symptoms and disablement, they also reduce the general risk factors for neuroses, which in turn reduces the risk of relapse.2

Is there evidence for the effectiveness of psychological treatments in other disorders, albeit at a lower level of proof than that required by the threshold used above? There is support for cognitive behaviour therapy for social phobia, post-traumatic stress disorder, and especially for bulimia,2 15 and there has been extensive discussion about the efficacy of other brief psychotherapies in depression.16 What about the two most commonly used psychological treatments, counselling and long term psychodynamic psychotherapy? Unfortunately, there is no evidence that the results of either therapy are reliably better than placebo treatment,17 and purchasers, providers, and consumers might well decide that they are not treatments that a wise health service should offer.18 Conversely, prudent purchasers, smart providers, and informed consumers might well search out therapy with the cognitive behavioural techniques listed above, especially those supported by best practice protocols and those that include some means of quality control.2