Research Signal – Source date 2021-03-03
In a review of 60 cancer trials, the median incidence of all-grade adverse events labeled drug-related in placebo groups was 49%. That figure raises a useful question: what does a drug-related label tell us when it appears in a placebo group? The answer matters because safety reports compare both what participants experienced and what researchers judged treatment-related.
The review, published in 2021, examined reports from randomized, double-blind, placebo-controlled phase 3 cancer trials of targeted therapy or immunotherapy-related drugs. It found gaps in how those reports presented adverse-event attribution. The numbers can help readers understand the challenge of comparing safety, but they cannot identify what caused any one person’s symptoms.
Counting events and assigning cause are different steps
An adverse event is an unwanted health event recorded during a trial. An all-cause count includes events regardless of whether researchers think the treatment caused them. A drug-related count includes events judged to be related to the treatment.
Those are two different kinds of information. The all-cause count describes what happened among participants; the drug-related count reflects an attribution judgment. A symptom can be recorded without the trial assigning it to the study drug. Conversely, a group-level count of events judged related does not prove the cause of an individual event.
The review reported two measures that are easy to conflate. First, the median incidence of all-grade events labeled drug-related under placebo was 49%, with an interquartile range (IQR) of 39% to 56%. This is the middle value across the reviewed trials’ reported incidences: half of the trial values fell below it and half above it, with the IQR describing the middle half of those values.
Second, the authors calculated an attribution ratio: the number of patients with drug-related adverse events divided by the number with all-cause adverse events. For all-grade events, the median ratio was 88.9% in active-treatment arms (IQR 79.8% to 93%) and 53.9% in placebo arms (IQR 43.4% to 60.9%). The ratio compares patients with adverse events labeled drug-related with patients who had any adverse event; it is not the same as the 49% incidence figure. One describes how often patients in placebo arms had an event labeled drug-related; the other compares patients with such a label against patients with any adverse event in each arm.
Many trial reports left attribution incomplete
Across the 60 trials, which covered 38,174 patients, 40 publications—66.6%—did not provide complete adverse-event attribution reporting. The review defined a complete report as one that included both all-cause and drug-related adverse-event data.
Readers need both counts to interpret a safety comparison. All-cause data show the burden of events recorded in each trial arm. Drug-related data show which events were attributed to treatment. If a report gives only one view, readers have less information for understanding the difference between what participants experienced and what researchers judged treatment-related. The review’s reporting gap does not show how any particular missing detail affected a treatment decision or patient outcome; it identifies a limitation in the published comparisons.
A separate reporting pattern involved higher-severity events
The review also found that journals with the lowest impact factor were significantly associated with deficient reporting of grade 3–4 adverse events (p = 0.02). Grade 3–4 are severity categories used in the trial reports. This was an association between journal impact factor and reporting completeness, not evidence that journal rankings caused the reporting differences.
The finding concerns whether higher-severity events were adequately reported. It does not establish that patients in those publications experienced more severe harm, or that the reported events were caused by treatment. Reporting quality and the frequency or cause of harm are separate questions.
What clearer cancer trial safety reports could change
Consistent reporting of both all-cause and drug-related adverse-event data could help clinicians and patients weigh a treatment’s reported safety profile alongside its potential benefit. That comparison would be more useful if the attribution process also proved dependable across trial arms and publications. The review points to a need for interpretable safety reporting; it does not show that improved reporting has already changed care or outcomes.
Key Sources
- Primary source: Analysis of adverse events attribution and reporting in cancer clinical trials: A systematic review.
DOI or Journal Page · Review or Meta-Analysis · 2021-03-03
Keep exploring