To the Editor
The Journal recently published two trials of tocilizumab for coronavirus disease 2019 (Covid-19) (Dec. 10 and Jan. 7 issues).1,2 According to the articles, the primary outcome in both trials was mechanical ventilation or death. However, there are important discrepancies between those articles and the trial protocols and registry information.
In both trials, the original protocols and ClinicalTrials.gov records specify mechanical ventilation as the primary outcome. In the trial by Stone et al., the primary outcome was changed to include death in the final protocol and, almost 2 months after enrollment started, in the ClinicalTrials.gov record. In the trial by Salama et al., the primary outcome was changed in an amended protocol after enrollment had started, but the ClinicalTrials.gov record still specifies mechanical ventilation as the primary outcome. This protocol change is of particular importance, since the results for the “new” primary outcome were significant but the results for the original primary outcome were not reported in the article.
Outcome switching in trials may lead to bias,3,4 and it is not clear why these important protocol changes were not described in the articles.
Andreas Lundh, M.D., Ph.D.
Hvidovre University Hospital, Hvidovre, Denmark
[email protected]
No potential conflict of interest relevant to this letter was reported.
This letter was published on March 3, 2021, at NEJM.org.
-
1. Stone JH, Frigault MJ, Serling-Boyd NJ, et al. Efficacy of tocilizumab in patients hospitalized with Covid-19. N Engl J Med 2020;383:2333-2344.
-
2. Salama C, Han J, Yau L, et al. Tocilizumab in patients hospitalized with Covid-19 pneumonia. N Engl J Med 2021;384:20-30.
-
3. Altman DG, Moher D, Schulz KF. Harms of outcome switching in reports of randomised trials: CONSORT perspective. BMJ 2017;356:j396-j396.
-
4. Chan A-W, Hróbjartsson A, Haahr MT, Gøtzsche PC, Altman DG. Empirical evidence for selective reporting of outcomes in randomized trials: comparison of protocols to published articles. JAMA 2004;291:2457-2465.
Response
Drs. Salama and Mohan reply: Lundh references potential bias introduced by postenrollment modification of the primary outcome with knowledge of the data. Such bias was not introduced in our trial, because the original planned analysis was unchanged and interim efficacy analyses were not performed. Death represents a competing risk for mechanical ventilation; therefore, a composite outcome is needed. Language to clarify the primary outcome was added in version 2 of the protocol to technically align with the original planned analysis, in which death before mechanical ventilation was considered a qualifying event in the evaluation of the primary outcome. Furthermore, version 2 of the protocol was finalized when less than 5% of the patients had been enrolled, while the trial was fully blinded, and well in advance of the internal review of unblinded safety data by the internal monitoring committee. For these reasons, we refute the suggestion that bias may have resulted from updating of the wording describing the primary outcome in the protocol.
In addition, ClinicalTrials.gov updates may lag behind requests for updates. We thank Lundh for bringing this administrative issue to our attention and have requested that the posting be corrected.
Carlos Salama, M.D.
Elmhurst Hospital Center–Icahn School of Medicine at Mount Sinai Hospital, New York, NY
Shalini V. Mohan, M.D.
Genentech, South San Francisco, CA
[email protected]
Since publication of their article, the authors report no further potential conflict of interest.
This letter was published on March 3, 2021, at NEJM.org.
Response
Dr. Stone replies: I appreciate Lundh’s point and agree that ideally a clinical trial is planned in elaborate detail, relatively free of time pressure, and none of the well-conceived plans are altered during its conduct. The reality of Boston in April 2020 dictated different circumstances for the start of our trial. Boston was in the throes of the first pandemic surge. At that time, not only was there no standard of care for Covid-19, but the appropriate outcome measures were also uncertain. After enrollment had begun and we were able to think further about analyzing the results to come, it made sense to expand the primary outcome to include not only the prevention of mechanical ventilation but also the prevention of death. We formally adjusted the primary end point in the protocol 6 weeks into the trial, notifying our institutional review board immediately and updating ClinicalTrials.gov at a later date (before the conclusion of the trial and before the locking and analysis of the data set). Ultimately, this modification of the primary outcome did not alter the primary or secondary conclusions of the trial.
John H. Stone, M.D., M.P.H.
Massachusetts General Hospital, Boston, MA
[email protected]
for the BACC Bay Tocilizumab Trial Investigators
Since publication of his article, the author reports no further potential conflict of interest.
This letter was published on March 3, 2021, at NEJM.org.
March 04, 2021 at 05:01AM
https://ift.tt/3uQCQQg
Tocilizumab in Patients Hospitalized with Covid-19 Pneumonia | NEJM - nejm.org
https://ift.tt/3d5QSDO
No comments:
Post a Comment