We thank Dr Palamarthy for the interest shown in our study.1 Three points are made. First, that most of the population studied were hyperventilating, as evidenced by the mean Nijmegen scores. This is incorrect. Our previous work has shown that one-fifth of men and one-third of women with asthma treated in the community in the UK have Nijmegen scores indicative of possible hyperventilation,2 and that those with such high scores had a high probability of responding to breathing therapy.3 In the current study, one of the research questions we addressed was whether patients with asthma with a low Nijmegen questionnaire score responded as well as those with a high score—that is, whether those with possible hyperventilation as well as asthma responded better than those with asthma but without symptoms of hyperventilation. Each randomisation group therefore had 50% of subjects with high Nijmegen scores (⩾23) and 50% with low scores; this is stated in the Methods section and in the statistical analysis section where we state: “We also assessed whether the Nijmegen Questionnaire (a screening tool for symptomatic hyperventilation) score (<23 or ⩾23) or physiological evidence of hyperventilation influenced response to breathing retraining”. As reported in the Results section under the heading “Influence of hyperventilation markers on response to breathing training”, no difference in response to breathing training was found between high and low scorers of the Nijmegen questionnaire, nor between those with low and higher carbon dioxide tensions at baseline. The results imply that this intervention can help many patients with impaired asthma-related health status, regardless of symptomatic or physiological evidence of hyperventilation.

The second point concerns the generalisability of the findings. As detailed in the Consort diagram, 516 subjects out of 3139 invitation letters (outlining the study protocol) responded with interest—a response rate of roughly 1 in 6—and 183 subjects were randomised. Recent work has shown that typical asthma clinical trials recruit a far lower proportion of potentially eligible subjects than ours,4 usually in the order of 2%, and we know of no community-based controlled trials in asthma that have achieved a better recruitment rate. The point on generalisability applies to all randomised controlled trials but we feel that our study, because of the recruitment strategy, is likely to have better external validity than the trials on which current guidelines are based.

Finally, it is noted that the control group receiving asthma education achieved within-group benefits and a significant reduction in exhaled nitric oxide concentration. We agree with these observations, and also with the suggestion that pharmacotherapy and asthma education are vital aspects of asthma management. However, the within-group improvements from baseline and the greater improvements in patient-centred end points noted at 6 months in the breathing therapy group compared with the education group point to the possibility that this intervention may be an effective one for patients with impaired quality of life despite pharmacotherapy, and one that may benefit many patients with asthma. Future studies should investigate whether breathing exercises have additional benefits to effective education.