When do you really need to use a data safety monitoring board?

Experts discuss staying on top of adverse events

More clinical trials are using data safety monitoring boards (DSMBs) in recent years as investigators, sponsors, and clinical trials managers increasingly see the value of added checks and balances to subject protection, experts say.

Strictly speaking, few clinical trials are required by federal regulations to have a DSMB or data monitoring committee, but FDA guidance suggests a number of other types of trials should consider forming such a board. And this guidance is where clinical trials managers should begin looking when they want answers to the big question of whether to form a DSMB.

From a regulatory perspective, the only type of trial that’s required by the FDA to have a DSMB is a trial in which there is a waiver of informed consent for emergency research, says Susan Ellenberg, PhD, director of biostatistics and epidemiology at the Center for Biologics Evaluation and Research of the FDA in Rockville, MD.

The NIH and the VA have their own requirements regarding DSMBs, but the FDA’s requirement for an independent DSMB pertains to emergency research in which extra patient protections are necessary, she notes.

"The FDA put out a guidance document a few years ago, and we are in the process of revising it based on comments we received that suggest that in other areas, a DSMB could be an addition to the trial to ensure protection," Ellenberg says. "We recommended these kinds of data monitoring committees for trials that address mortality endpoints."

For instance, a data monitoring committee could assess the trial’s information for signs that one treatment is saving lives while another is not, and so that trial may need to be stopped early, Ellenberg says.

FDA guidance also suggests that a DSMB might be considered in clinical trials in which an innovative treatment creates concerns that there might be serious adverse events, she adds.

It might be a mistake to routinely create a DSMB for every clinical trial because not every study needs one, says Patricia Hibberd, MD, PhD, professor of medicine at the Tufts University School of Medicine, division of clinical research resource, and Tufts New England Medical Center in Boston.

"If every study needed a DSMB, we wouldn’t have enough people to sit on these DSMBs," she says. "So I’m not advocating that every one have one."

While every trial needs oversight and monitoring, most don’t need a DSMB, Ellenberg points out. "The vast majority of clinical trials do not need this extra layer of oversight," she says. "This is one more committee to have in a clinical trial, and with the complexity of doing a trial, you should only have them when they’re really needed."

Likewise, it’s a mistake to think that clinical trial managers and investigators can stop worrying about subjects protection because they have a DSMB, Hibberd notes.

"You still have to do your homework to make sure safety of subjects is maintained throughout the study," she says.

Very few clinical trials used DSMBs until the 1990s when the National Cancer Institute started to incorporate the boards into their trials, notes Ellenberg.

"So there’s been a gradual increase [in use of DSMBs] as more people became aware that these committees could provide extra information," she explains. "Also there are more industry trials that have serious end points than there used to be."

Ellenberg and Hibberd provide these thoughts to consider when deciding whether or how to most effectively use a DSMB:

• A DSMB makes recommendations to sponsors. Data monitoring committees receive raw data regarding adverse events and a trial’s progress, analyze the data, and make recommendations in a report to the sponsor, Ellenberg explains.

The information also could be made available to the IRB, she adds. "Because the sponsor is blinded to the results, you have to have somebody looking at the interim results, and it cannot be people who could change the trial," Ellenberg says. "So when the report from the DSMB goes back to the sponsor it doesn’t reveal what interim data look like; it makes recommendations."

In trials where subjects might die before the trial ends, it’s important to have a third party looking at the interim results to make certain the trial hasn’t produced dramatic results that should result in a change, she says.

"The DSMB might make a recommendation to the sponsor to make a change in the trial to reduce adverse events, such as reducing dosage or adding another drug to mitigate or — in the worse case — to stop the trial," Ellenberg explains.

"You could have a trial where you see unusual adverse events only in women," Ellenberg says. "That’s why you have a DSMB to have an independent judgment that is not colored by how much money it costs to get the trial going and how much money the company would spend to make it successful."

• The DSMB’s responsibilities need to be spelled out in advance. "If you write a research protocol and say this needs to be monitored by an independent DSMB then it’s your job to put together a DSMB and include experts of relevance for the particular study," Hibberd says.

For instance, here are some questions to consider:

— What will the DSMB do?
— Will the DSMB review adverse events?
— Does the DSMB need to look at summaries of adverse events?
— How will the DSMB be successful in having the appropriate oversight?
— Should the DSMB have regular meetings with the investigator?
— How will the meetings be conducted?
— How will the DSMB’s recommendations be handled?
— How does information get from the DSMB to the IRB?

"All of those issues need to be sorted out, and that’s the investigators’ responsibility," Hibberd says. "If the pharmaceutical company is setting up the DSMB, their responsibility is the same."

With regard to the IRB, it’s important to know who will be sending the DSMB’s information to the IRB, or whether that will be done, she adds.

Extra steps 

• Best strategy is to set up DSMB early in process. "I think honestly the best way to set up a DSMB is — before patient accrual occurs — for there to be a meeting between the principal investigator and the DSMB to review the processes and to agree upon the processes," Hibberd says.

"There may be certain trials where the DSMB wants to hear about every single serious adverse event as they occur, and if that’s what the DSMB would like to do then we’ve got to make sure that happens," she adds. "In other circumstances, they may say we don’t need to hear anything unless a certain threshold is crossed, and if you don’t set up the process of when and how before the study starts then you may have a problem."

• Sponsors may circulate DSMB report to investigators. "There could be an agreement that the sponsor would circulate the DSMB report to investigators, and if there are any recommendations that impact how the trial is conducted, then that would have to be circulated," Ellenberg says.

If the DSMB says that things look good, but the data aren’t up to date, then the clinical trials staff need to know that there is a problem in the process that needs to be corrected, she says.

"The DSMB might recommend a dose be lowered because of toxicity, and if the sponsor agrees, that information goes out to all cities because the protocol has been changed," Ellenberg says. "If they say everything is going well, there’s no reason why the sponsor couldn’t circulate that as well."

• DSMBs pay attention to time limit issue. "From the perspective of someone who has had experience in managing data monitoring committees and serving on them, one thing a DSMB will pay attention to is the time limit issue," Ellenberg says. "If a DSMB sees a high rate of dropouts, they’ll be concerned about that because those kinds of problems harm the eligibility of the trial, and the DSMB will make a recommendation to the sponsor to tighten up the trial."

Also, DSMBs often consider aspects of trial quality in addition to data on safety because if a clinical trial reaches its end and doesn’t produce useful information because the trial wasn’t conducted rigorously enough then this becomes an ethical issue, she notes.

"Is it ethical to ask people to participate in a trial for which there is no useful result?" Ellenberg says. "So frequently DSMBs will make recommendations regarding quality."

• Trials using DSMBs still need to include extra steps to ensure safety. If the trial’s risk will be high enough to require a DSMB, then investigators need to include more details to explain how they are going to be sure a study is going to be safely conducted, Hibberd notes.

"It’s not enough to say, We’re going to report adverse events,’" she says. "How do you know an adverse event has occurred? Are you going to ask patients, and how do you ask patients?"

For instance, should clinical trials staff ask subjects whether they’ve had any adverse events, or should they ask, "How have you been since the last time we saw you?" Hibberd says.

"What I see in a lot of protocols are descriptions of what an adverse event is and various definitions depending on how the protocol is regulated," she notes. "I couldn’t care less what the term is, but I know the spirit is the same in all circumstances, so move on to the next level: How are you going to explain to the IRB that you’re going to be on top of them instead of giving a long definition of some regulatory definition of adverse events pasted in off a web page?’"

Also, clinical trials staff need to provide context to DSMBs regarding adverse events, Hibberd says.

For example, if there are three patients reporting headaches, it’s not as useful to provide that information without the denominator of the number of people enrolled in the study, she explains.

"If there are three patients and every one has a headache, that’s likely a major problem," says Hibberd. "If there are 3,000 patients and only three have a headache, then that’s a very different matter."