We thank the authors for their interest in our paper.1 They raise an interesting and pertinent point about identifying patients at greater risk of β-agonist toxicity, a group in whom either the dose of β-agonist should be lowered, or in whom our point about co-prescription of inhaled corticosteroids is even more relevant.2 Unfortunately, we feel that the question about whether these patients could be identified in the context of randomised controlled trials, in which mortality is very rare and important morbid outcomes are relatively uncommon, is that it would be impossible to demonstrate a difference in these rare outcomes stratified by genetic status. This of course amounts to a similar problem as experienced by subgroup analysis of randomised controlled trials, namely a lack of statistical power to detect important differences. Another way of examining whether genotype affects morbid outcomes related to medication use may be to use this as a covariate in case–control studies and examine whether risk is modified. This would require that cases and controls had appropriate genetic material available for analysis and would probably require a prospective study.