Letters to the Editor

Trial risk seems high

The purpose in contacting you is to raise a query regarding a news item published in the January 2002 issue of

ACT

. I wonder whether you could shed some light on itor am I just plain wrong!

CenterWatch Study Qualifies Risk to Clinical Trial Participants (Applied Clinical Trials, January 2002, p. 19) describes the occurrence of 13 deaths attributable to study drug effects amongst 122,000 subjects participating in trials for 130 randomly selected NDAs that had received approval since 1987. It concludes that the incidence of death attributable to investigational treatment is rare.

This seems to me to be anything but rare. These figures, if true, work out at one death for every 9,385 subjects involved in clinical trials! It may be even more alarming when one considers that a proportion of the investigational treatments may have been placebo if the trial was of an inactive comparator crossover design.

I have not seen the metrics published by CenterWatch, and so have not been able to pursue my concerns further. Perhaps you may be able to help me.

Nigel J Crossland, FRQA
Crossland Consultancy
Swettenham Village, Cheshire, UK

CenterWatch responds
Thank you, Nigel, for your feedback. To clarify our findings, CenterWatch analyzed a subset of approved drugs between 1987 and 2001. In all, 126 approved drugs were evaluatedrepresenting nearly one-third of all new molecular entities approved by the FDA since 1987.

We found that each year, on average, 17,200 adverse events (AEs) and 800 serious adverse events (SAEs) are reported to the FDA for approved drugs. Each year, an average of 3.6 deaths attributed to study drug effects are reported to the FDA for approved drugs. This result compares to 0.8 deaths reported to the Office for Human Research Protections (OHRP) for government-funded clinical trials.

We also found that an average of one adverse event per study subject is reported to the FDA, and an average of one out of 30 study subjects on a study drug will have a serious adverse event. Lastly, we found that in reports from industry to the FDA, one out of 10,000 study subjects have died as a result of study drug effects while participating in clinical research studies.

To put this latter statistic into perspective, based on sponsor reports to FDA, more people die from motor vehicle accidents each year than from participating in a clinical trial. The risk of death to a study subject is nearly equal to that of a woman dying during childbirth. Many ethicists and industry professionals consider this an acceptable level of risk relative to the benefits that certain investigational treatments offer.

Our findings articulate the general risks but the data are averages, so they do not convey the high level of variability between clinical studies. Numerous variables affect research risk, including indication area, type of drug, and study duration. Severe, life-threatening illnesses tend to have a higher incidence of adverse reactions and deaths. Studies that are longer in duration tend to report higher levels of adverse events. Sponsors also subjectively define an adverse event. Whereas one sponsor may view a group of related symptoms (fever, cough, and runny nose, for example) as a single AE, another sponsor may require that the site count each symptom as a separate AE.

Our analysis is also based on FDA data from approved NDAs. We have not yet evaluated the many industry-sponsored trials that are terminated before an NDA is filed. As a result, our findings are evolving and we hope to publish revised estimates shortly.To receive a copy of the full article, Quantifying Risk in Clinical Trials, and a detailed description of the methodology, please send me an email at kenneth.getz@centerwatch.com.

Ken Getz
President and Publisher
CenterWatch
Boston, MA