The Value of Transparency When Reporting Adverse Events

A new study published in the journal PLoS Medicine has found strong evidence that a significant portion of information on adverse events (AEs), gathered from clinical trials, remains unpublished and is incompletely reported. This information should prompt treating physicians and researchers to dig deeper to better understand the side effects of specific treatments.

For the current analysis, the authors combed through 15 medical databases, conducted internet searches, gathered data from conferences, and contacted experts in the field to assimilate information on side effects and adverse effects of treatments—28 methodological interventions met the set inclusion criteria, 9 of which compared the proportion of trials that reported AEs based on published results. The objective was to quantify underreporting of AEs and measure its subsequent impact on systematic reviews of AEs.

Less than half (46%) of the published documents reported AEs, compared with 95% of corresponding unpublished documents. The trend was similar for unmatched studies, where 43% of published studies reported AEs compared with 83% of unpublished studies. The authors concluded that between 43% and 100% (median, 64%) of AEs would be missed if only the published versions of studies were included in their analysis. Additionally, the authors found the following:

• 24 comparisons of named AEs (eg, death, suicide, respiratory AEs), 18 of which named more number of AEs in unpublished documents
• 20 meta-analyses reported odds ratios or risk ratios for AEs, with and without unpublished data
• Unpublished data increased the precision of pooled estimates in 15 out of 20 pooled analyses, without affecting significance or risk

Based on their findings the authors believe that the number and range of AEs are much higher in unpublished data compared with the published version of the same studies. Including unpublished data “can also reduce the imprecision of pooled effect estimates during meta-analysis of adverse events,” they wrote.

The value of a treatment is a combination of its efficacy and its unwanted effects. A drug that improves disease symptoms while simultaneously causing adverse events that lead to additional treatment or hospitalization may not be the most valuable option. Unless all the documented evidence that results from a study is openly published by participating researchers and drug developers, patients and physicians will remain ill-informed when making care decisions. Publication bias and unreported harms data could lead to erroneous judgement by physicians on the benefit to harm ratio of a treatment. This, the authors state, calls for an urgent policy intervention to allow full public access to AE data.

Reference

Golder S, Loke YK, Wright K, Norman G. Reporting of adverse events in published and unpublished studies of health care interventions: a systematic review. PLoS Med. 2016;13(9):e1002127. doi: 10.1371/journal.pmed.1002127.