Suboptimal Adverse Event Reporting in Cancer Clinical Trial Publications
In a study reported in the Journal of Clinical Oncology, Sivendran et al assessed the degree to which publication of phase III trials in oncology conformed with CONSORT (Consolidated Standards of Reporting Trials) recommendations for reporting adverse events. They found that adverse event reporting is suboptimal.
Study Details
The study involved review of PubMed, Medline, and Embase to identify randomized controlled phase III trials in metastatic solid malignancies published between January 2009 and December 2011. Publications were assessed for 14 adverse event reporting elements derived from the CONSORT harms extension statement, with a completeness score of 0 to 14 being generated. The elements are shown below, along with the percentage of publications satisfying each.
Median Completeness Score
A total of 175 publications including data on 96,125 patients were included in the analysis. The median completeness score was 8 (range, 3–12). Most publications (96%) reported only adverse events occurring above a threshold rate or severity, 37% did not specify the criteria for selecting which adverse events were reported, and 88% grouped adverse events of varying severity.
Percentage of Studies Satisfying Criteria
Of the articles analyzed, 91% had titles/abstracts stating whether adverse events were addressed and 58% had introductions stating whether adverse events were addressed. For methods, 65% specified whether reported adverse events encompassed all recorded events or a select sample, 93% specified an instrument/scale used to categorize adverse events, 62% specified a surveillance time frame, 14% specified whether early stopping rule was used for toxicity, and 16% specified whether recurrent events in the same patient were counted as separate or single events.
For results, 65% reported reasons for treatment discontinuation, 75% reported if there were deaths related to adverse events, 71% reported whether patients were evaluable for toxicity, 77% reported absolute numbers of adverse events (not percentages alone), 4% did not only report adverse events above a frequency threshold (eg, > 10%), and 12% reported adverse events of varying severity separately (ie, not in combined severity groupings. Sixty-five percent of articles did not use vague descriptors of toxicity in their discussions (eg, “the regimen was generally well tolerated”).
Factors Associated With Completeness Score
In unadjusted analysis including placebo-controlled status of trial, intervention type in trial, funding source, studies of intervention approved for other indication, year of publication, whether trial met primary efficacy endpoint, and journal impact factor, only lack of a specified funding source (mean difference = −1.058, 95% confidence interval [CI] = −1.858 to −0.258) and year of publication (mean difference for more recent year = 0.069, 95% CI = 0.204–0.844) were significantly associated with completeness score. On adjusted analysis, lack of a specified funding source (mean difference = −1.123, 95% CI = −1.983 to −0.262) and year of publication (mean difference = 0.523, 95% CI = 0.196 to 0.849) were also the only factors independently associated with completeness score.
The investigators concluded, “Reporting of adverse events in oncology publications of randomized trials is suboptimal and characterized by substantial selectivity and heterogeneity. The development of oncology-specific standards for adverse event reporting should be established to ensure consistency and provide critical information required for medical decision-making.”
Matthew D. Galsky, MD, of Mount Sinai School of Medicine, is the corresponding author for the Journal of Clinical Oncology article.
For full disclosures of the study authors, visit jco.ascopubs.org.
The content in this post has not been reviewed by the American Society of Clinical Oncology, Inc. (ASCO®) and does not necessarily reflect the ideas and opinions of ASCO®.
