Vioxx, NIH crises lead to questions about drug safety and IRBs’ role

But are IRBs responsible for long-term safety of FDA-approved drugs?

The revelation that Merck’s popular painkiller rofecoxib (Vioxx), as well as other drugs in the COX-2 inhibitor class, pose an increased risk of heart attack and stroke in patients who take them has led many in the research community to question whether the instiutional review boards at the sites where the drug was tested failed in their mission. Disclosure by investigators and FDA regulators that they knew of clinical data indicating a statistically significant number of adverse cardiovascular events among study subjects has only added to the controversy.

If IRBs are charged with protecting the interests of human subjects enrolled in clinical research, many wondered, how could a drug with such glaring risks be approved?

The debate has been further heightened by allegations by an internal investigator at the National Institutes of Health (NIH) that senior officials there conspired to cover up significant research misconduct that allegedly occurred during a clinical trial of the AIDS drug nevirapine, which was conducted on pregnant HIV-infected women in Uganda under sponsorship from the National Institute of Allergy and Infectious Diseases. The NIH whistle-blower, Jonathan Fishbein, currently is scheduled to testify before Congress about allegations that the researchers failed to adhere to the study protocol, mishandled study drugs, poorly maintained patient records, and failed to report many adverse events, including deaths. According to one internal NIH audit report, thousands of adverse events in the study were never documented.1

Such revelations lead many in the public and clinical research arenas looking for a single place to assign blame, and, unfortunately, that scapegoat has become the IRB, says Paul Goebel, CIP, vice president of Chesapeake Research Review Inc., a private biomedical and social science research support firm based in Columbia, MD, that provides a central IRB and scientific ethical review consulting for different institutions.

"Many seem to expect the IRB to be able to prevent abuse, misconduct, or an adverse event from occurring," he says. "Failing that, they expect the IRBs to know when these shortcomings occur and take prompt action."

But such expectations misconstrue the mission, structure, and functioning of IRBs as they have been traditionally established, Goebel points out.

Under the original federal regulations and guidance that established IRBs, they were charged with ensuring that a study protocol met federal guidelines and then periodically monitoring the study’s progress to ensure that investigators do what they said they were going to do and follow the approved protocol.

"They were originally designed for an initial review of a project, and then a how is it going?’ look after a period of time, usually annually," Goebel explains. "Most IRBs have now recognized that more frequent review is necessary for high-risk studies. For example, the IRB for the first artificial heart study performed a review after each installation before permission was granted to the researcher to install the next one. The investigator soon moved his research to another institution, in part because of the close IRB oversight of the study. The IRB was asking questions and demanding assurances that were viewed by some as excessive at that time, but would be considered normal and necessary today."

Informed decisions possible?

Currently, IRBs are simply unable to perform the added oversight functions that the public increasingly seems to be asking them to assume, agrees Michael Goodyear, MD, assistant professor in the department of medicine at Dalhousie University in Halifax, Nova Scotia. Goodyear currently is serving on a task force to recommend improvements in clinical research oversight to Health Canada.

"The public has far too much confidence in the system and a very incomplete idea of what IRBs are all about," he says. "In fact, IRBs are usually woefully underresourced. The IRB’s primary role is the conduct of a study locally. On the other hand, there is nothing that stops IRBs talking to each other or to other bodies if they are concerned. I strongly encourage that."

Even so, he adds, IRBs and research ethics boards (their Canadian equivalent) are not meant to be "content experts, but principle experts."

However frequently they monitor progress or communicate with other boards, IRBs still are not structured to evaluate complex safety data, such as the occurrence of cardiovascular events in the COX-2 trials, adds Goebel.

First, they usually only get information about adverse effects occurring in subjects at their site — which may be only one out of several nationwide or internationally, he says. And, they often are only told that an effect occurred, not whether it was expected or unexpected, and whether other factors, such as subject compliance, were involved, Goebel adds.

"The IRBs do not get enough information about an adverse effect to enable them to make an informed decision. The IRBs are told that an event occurred. They do not have access to the complete study data. And, many IRBs do not have statisticians as members," Goebel continues. "Whether or not to continue a study often involves a statistical evaluation of the raw data. It is very rare for an IRB to stop a study. Most adverse effects do not present a black-and-white, clear-cut situation upon which to stop a study. The recent Vioxx and Celebrex heart problems only surfaced after sophisticated statistical evaluations of hundreds or thousands of uses."

Monitoring of this kind is and should be the purview of an independent data safety monitoring board (DSMB) established by the study sponsor, he notes.

DSMBs, the existence of which can be mandated by an IRB, function to evaluate the raw data collected by all of the study sites to determine whether any significant effects are discernible, Goebel says.

"In my view, this is the right place for such decisions to be made. NIH requires a safety monitoring plan to be in place for some studies," he says. "This framework could be extended to all studies conducted in humans for which the research could result in substantial or permanent harm to the subjects."

Research watchdogs also should not let sponsors, study monitors, and principal investigators off the hook when it comes to the occurrence of adverse events or disclosure of potential misconduct, Goebel continues.

Under federal research guidance, it is the responsibility of the sponsor to send site monitors to evaluate the conduct of the study at each site, and the responsibility of principal investigators to monitor the conduct of the subordinate investigators and research coordinators to ensure that subjects receive appropriate informed consent, are treated according to the study protocol, and that any adverse events are appropriately reported.

It is unrealistic, Goebel contends, to place the lion’s share of the burden on an IRB already in charge of monitoring possibly hundreds of other ongoing studies at the same time.

"The people closest to the study are the clinical investigator and the study coordinator. They are the ones who are involved with the study every day," he notes. "The study monitors are charged with repeated oversight of the study. They visit the study site several times a year. The IRB is a distant third. A routine study is reviewed once a year, excluding changes in the protocol and adverse effect reports. So if we want IRBs to now assume the responsibility for close monitoring of studies, it will be expensive."

Improvement in oversight cannot be the sole responsibility of individual IRBs, but they can be empowered through changes in regulation or guidance to have more of a say in how drug studies are conducted, Goodyear notes.

"Yes the issues raised are outside the IRB scope. Yet the IRB is a component of a social structure set up to protect the public, and is entitled to a voice at the table on drug policy," he explains. "But ideally, IRBs should have a strong voice through a central umbrella organization. In terms of what we now know, IRBs can simply refuse to approve studies that do not meet minimal standards, such as an independent DSMB. The can also lobby to have these made mandatory, and for other national standards. However, to do this they will need the ear and support of the regulator. Usually what happens is sponsors play IRBs off against each other."


1. Associated Press. "NIH Dismissed Concerns About Drug Treatment." Dec. 14, 2004. Accessed on-line at: