© 2008 American Thoracic Society
Methodologic Shortcomings of the INSPIRE Randomized TrialTo the Editor:In their article, Prof. Wedzicha and colleagues presented results of the INSPIRE study, which included an unexpected 52% reduction in the rate of all-cause mortality in patients with chronic obstructive pulmonary disease (COPD) (1). Patients were given fluticasone/salmeterol, an inhaled corticosteroid/long-acting β-agonist (ICS/LABA), and were compared with others who were given the long-acting anticholinergic (LAAC) tiotropium. Methodologic issues need to be considered when interpreting this apparent mortality reduction. First, the study design imposed discontinuation of inhaled corticosteroids followed by a 2-week run-in of oral prednisolone and salmeterol prior to randomization, which may have affected outcomes (2). Thus, the comparison of patients initiating treatment with ICS/LABA versus LAAC is only possible among patients not on ICS at entry. In ICS users at entry, the comparison contrasts, instead, patients who continue to use ICS (randomized to ICS/LABA) with patients who discontinue ICS (randomized to LAAC). This discontinuation was shown to have radical effects in another study (3), with additional effects likely from the abrupt interruption of the run-in regimen of oral prednisolone and salmeterol. Second, the incomplete follow-up of patients, stopped when patients discontinued the study drug and after which time no deaths were identified, does not permit the fundamental intent-to-treat analysis. The serious bias from incomplete follow-up was shown in ICS trials. The ISEEC trials with similar incomplete follow-up found a significant 27% (P = 0.04) reduction in mortality with ICS compared with placebo (4, 5), whereas the TORCH trial with complete follow-up for a proper intent-to-treat analysis, found a nonsignificant 6% (P = 0.53) increase in mortality with fluticasone compared with placebo (6). Table 1 compares the rate and cumulative incidence of death with ICS/LABA, ICS, and placebo, which were approximated from studies that did and studies that did not have complete follow-up. The large differences are clearly due to a radically higher mortality rate in patients who discontinued assigned treatment, which will introduce bias.
The INSPIRE study design thus violates two major principles of randomized trial methodology. Unfortunately, these design flaws cannot be simply corrected with data analysis. Consequently, the finding of a 52% reduction in mortality with ICS/LABA is subject to important biases and must thus be considered uninterpretable.
Jewish General Hospital FOOTNOTES Conflict of Interest Statement: S.S. has been reimbursed for attending advisory board meetings and conferences and participating as a speaker in scientific meetings financed by various pharmaceutical companies (AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Pfizer, and Sepracor); he has received funding for research grants from AstraZeneca, GlaxoSmithKline, and Boehringer-Ingelheim. REFERENCES
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
