Intended for healthcare professionals

Rapid response to:

Research Methods & Reporting

Recommendations to improve adverse event reporting in clinical trial publications: a joint pharmaceutical industry/journal editor perspective

BMJ 2016; 355 doi: (Published 03 October 2016) Cite this as: BMJ 2016;355:i5078

Rapid Response:

Response to Drs. Proctor and Schumacher

We thank Drs. Proctor and Schumacher for their thoughtful suggestions from November 25 regarding methods to graphically display the risk and time course of adverse events in clinical trials (Figure 1). We agree that there are various methods to compute and display the cumulative incidence and/or risk of experiencing adverse events, several of which consider appropriate competing risks (eg, death or disease progression). In our review of recent reports from oncology trials, we observed a variety of methods used, including Kaplan-Meier estimates of time to event,[1] Kaplan-Meier estimates with censoring for predefined competing risks,[2] and cumulative incidence curves using Fine and Gray’s method for subdistribution of competing risks.[3] As noted by Drs. Proctor and Schumacher, the particular example we presented considers only those who (eventually) experienced the adverse event, which is useful information, but not the only information one would want to see. We elaborate below.

We agree that there will always be room for improvement in the ways to represent key data and, aligned with the mission of Medical Publishing Insights & Practices, we welcome suggestions for such improvements. The examples we selected for our publication were chosen to address the key questions practicing physicians would have for successfully using any new treatment:
• What are the most frequently observed adverse events with this treatment?
• How likely are my patients to experience these adverse events?
• Are there specific timeframes after starting treatment during which the risk of adverse events is highest or additional monitoring and follow up are needed?
While a single representative table or figure may not address all 3 questions above, the aim of our recommendations is to ensure that the data in a manuscript as a whole adequately address these aspects.

As mentioned in our publication, the recommendations and examples provided are intended as a guide for better reporting of adverse events in a clinically relevant and comprehensible manner. Therefore, we thank Drs. Proctor and Schumacher for helping us to continue the dialogue and gain further insights on how to improve representation of clinical trial data (including adverse events) within the published literature.

1. Bonifazi M, et al. Trastuzumab-related cardiotoxicity in early breast cancer: a cohort study. Oncologist. 2013;18(7):795-801.
2. Rugo HS, et al. Meta-analysis of stomatitis in clinical studies of everolimus: incidence and relationship with efficacy. Ann Oncol. 2016;27(3):519-525.
3. Wolff AC, et al. Risk of marrow neoplasms after adjuvant breast cancer therapy: the National Comprehensive Cancer Network Experience. J Clin Oncol. 2015;33(4):340-348.

Competing interests: Competing interests have not changed since publication of the original article.

07 December 2016
Jesse A. Berlin
Vice President, Global Epidemiology
Neil Lineberry, Bernadette Mansi, Susan Glasser, Michael Berkwits, Christian Klem, Ananya Bhattacharya, Leslie Citrome, Robert Enck, John Fletcher, Daniel Haller, Tai-Tsang Chen, Christine Laine
Johnson & Johnson
1125 Trenton-Harbourton Road, PO Box 200, Mail Stop TE3-15, Titusville, NJ 08560, USA