Authors' reply

We agree in part with the points made in these letters. Dr Okhai comments on the different treatment intensity between the conditions and in particular in the ‘control’ condition. The ‘control’ treatment was intended as a surrogate for placebo treatment. It is ethically difficult to have a placebo treatment for anorexia nervosa given the high morbidity of the condition and the lack of any placebo response. Our aim, therefore, was to have a ‘control’ condition similar to treatment as usual that would/could be offered in general adult psychiatry units. It could be argued that this therapy was better than that offered in many such positions in that regular supervision was given by an expert in eating disorders. Furthermore, the patients (2-3 per psychiatrist) were offered treatment for up to a year. We agree that in anorexia nervosa as in other conditions the therapeutic alliance is a key factor in response to therapy. We would argue that the specialist treatments have a specific focus on the therapeutic alliance. Indeed, it is perhaps noteworthy that the results of this study led to a change in the practice of cognitive—analytic therapy on the unit in that it is now preceded by a short course of motivational enhancement therapy to facilitate engagement (Treasure & Ward, 1997).

The number of sessions attended may be a sensitive marker of the therapeutic alliance in anorexia nervosa. For example, in a previous study comparing cognitive—behavioural therapy for anorexia nervosa with dietary management all patients dropped out of the dietary management group early in treatment (Serfaty, 1999).

We agree with Dr Morris that the important ‘take-home message’ is that specialised therapists following a specific therapeutic approach offer the best outcome in anorexia nervosa. This complements the analysis made by Nielsen et al (1998), in which he found that mortality was lower in regions of the country with specialised services. It is, therefore, of concern that such skills are in limited supply.