CardioExchange Archive

If we are to detect harms early and protect the public’s health while ensuring the availability of effective new therapies, systematic reporting of safety outcomes from all clinical trials must be available. 

In the current Archives of Internal Medicine, Isabelle Pitrou and colleagues (FREE) review the reporting of safety data in randomized, controlled trials published during 2006 in the major journals. Their findings are disappointing but not unexpected. Adverse events were unreported in 11% of 133 articles. Most seriously, data on severe adverse events or adverse-event-prompted study withdrawal were missing in 27% and 47% of articles. 

Editorialist John Ioannidis (subscription required) describes several missteps that lead to “insufficient or misleading” safety reporting: 

the study design ignores or undervalues adverse events
adverse event information is not collected during the trial
reporting of adverse effects is distorted, restricted, or completely withheld

In my opinion, the issue is a question of priority. Word counts and page limits are tight, and investigators (and sponsors) focus on efficacy outcomes. Five or more tables and figures may be devoted to graphical display of efficacy, while a single table summarizes safety data. Because graphs are easier to skim, after a cursory look, the average reader takes away only efficacy data. Even the data that is presented is challenging to interpret — many trials (particularly those sponsored by industry) group adverse events by system, rather than as specific events. 

One potential solution: Journals could require full and complete safety data, to be made available in on-line appendixes when necessary, allowing interested clinicians to weigh all safety information when making treatment decisions. 

Safety outcomes must be prioritized to ensure high-quality patient care. 

Other resources:
The 2007 FDA Amendments Act requires registration of all trials and reporting of primary and principal secondary (including safety) outcomes to http://clinicaltrials.gov — an online source of publicly available, easily searchable information that now includes 70,000 registered trials. With sufficient data quality, ClinicalTrials.gov could be the repository of safety information — ideally progressing to include raw data that allow independent re-analysis and meta-analysis.

So, do you pay attention to the safety outcomes when reading a clinical trial article? Have you ever looked specifically for the safety outcomes and not been able to find them? What did you do about it?