To the Editors:
We read with interest the paper by Suissa et al. 1, in which he and his co-authors state that “the effectiveness of inhaled corticosteroids in treating chronic obstructive pulmonary disease remains doubtful.” They review the reported efficacy of inhaled corticosteroids (ICS), noting a lack of pulmonary function improvement but reduction in exacerbation frequency, and a meta-analysis showing reduced mortality 2. However, they conclude that these findings could be biased by methodological problems. One form of bias that they highlight is failure to follow patients after premature withdrawal from randomised clinical trials. Suissa et al. 1 opine that the TORCH (Towards a Revolution in COPD Health) 3 and OPTIMAL 4 studies have provided important evolution in understanding these issues. While both studies used intent-to-treat analysis, at least for the primary end-points, TORCH showed a nonsignificant trend towards improved survival and OPTIMAL did not show significant differences in the proportion of patients experiencing at least one exacerbation over a 12-month period.
Perhaps a more important form of bias highlighted by Suissa et al. 1 is the effect of withdrawal of previously prescribed medications on entry into a randomised clinical trial. In the TORCH study, 48% of patients recruited were previously taking ICS (although Suissa et al. 1 report 51%), whereas in OPTIMAL the proportion was 76% (although Suissa et al. 1 report 77%). Because of the high frequency of patients already taking ICS on entry into these studies, they actually represent two comparisons: 1) continuation of ICS versus withdrawal of ICS, and 2) introduction of ICS versus no change in treatment. We agree that the results of this type of study are equally, if not more, likely to be influenced by the former comparison than the latter. Furthermore, as Suissa et al. 1 point out, a difference due to the former comparison does not necessarily equate to efficacy in the conventional sense, as exemplified by the latter. In their analysis, Suissa et al. 1 show a difference in exacerbation frequency due to withdrawal of ICS but no difference due to the introduction of ICS.
These issues are arguably just as pertinent in the interpretation of the INSPIRE (Investigating New Standards for Prophylaxis in Reducing Exacerbations) study, a recently reported comparison between an ICS/long-acting β-agonist combination and tiotropium 5. The INSPIRE study was remarkable for its run-in phase, during which patients were given prednisolone 30 mg daily for 14 days to “standardize their COPD management before randomization” 5. Rather than standardising chronic obstructive pulmonary disease management, this approach to study design surely represents a marked departure from recommended maintenance therapy for patients in clinically stable state and is, therefore, likely to amplify the phenomenon described by Suissa et al. 1. Furthermore, in the INSPIRE study the proportion of patients previously taking ICS was 50%. We consider, therefore, that the INSPIRE study also falls into the category of studies described by Suissa et al. 1 as potentially biased by methodological problems and its conclusions should be viewed with caution.
In their paper, Suissa et al. 1 suggest that trials of tiotropium might be subject to the same type of bias due to withdrawal of previously prescribed anticholinergic inhaler therapy prior to randomisation, although, to date, there is no scientific evidence to support such an implication. Access to the database of tiotropium clinical trials should enable a meta-analysis of exacerbations in subgroups that had anticholinergic therapy withdrawn at entry, compared with those that were naïve to anticholinergic therapy. Should such an analysis of tiotropium studies show a similar benefit in either group, then it will be reasonable to conclude that the benefits of long-acting anticholinergic therapy are genuinely due to introduction of the therapy rather than its withdrawal. Implications with regard to the efficacy of tiotropium should be reserved until such an analysis is published.
Statement of interest
Statements of interest for all authors can be found at www.erj.ersjournals.com/misc/statements.shtml
- © ERS Journals Ltd
