Case report forms can more accurately determine ADEs
Better method for finding causal connections
Adverse drug event (ADE) reporting often is inaccurate, has omissions, and sends unnecessary information to IRBs, an expert says.
The problem is that clinical trial researchers and staff often take histories in an unstructured way that leads to mistakes, says Steven Belknap, MD, an assistant professor at Northwestern University in Chicago, IL.
"My career has always involved research and clinical medicine," Belknap says.
"One thing I noticed in taking histories from patients is if I ask a patient 'Are you having problems with your medications, they say 'No,'" he says. "Then if you ask them, 'Since you started taking Medicine X are you having Y?' and they say, 'Yes.'"
Often patients have experienced a change but are not aware it's associated with a medication, he adds.
Clinicians and researchers need to ask about adverse drug events in a structured way, asking specific questions about symptoms. This generates different information than when they ask a global or general question, Belknap says.
"There's still a widespread use of global introspection for ADEs in clinical trials," he says.
Investigators evaluate some information from subjects and make a determination as to whether there's a causal connection between some exposure to drugs and toxicity.
"Take the same experts and have them use a structure for these questions, and they'll get a more reproducible assessment than they will with an unstructured approach," Belknap says.
For example, Belknap has seen cases where a patient has a severe rash such as toxic epidermal necrolysis, which can be fatal. The patient's rash began on May 1st. Then on May 15th, the patient receives the investigational drug.
"Somehow, when data are analyzed, it's the drug that is given credit for the rash," he says. "Maybe that drug was administered in an earlier course of treatment, and there are a lot of possible explanations, but if the description of data is correct, the attribution is incorrect."
Investigators who ask ADE questions in a structured way reduce the likelihood of making that mistake, he adds.
Belknap and co-investigators found in a study of 49 National Cancer Institute (NCI) designated centers that many use global introspection, despite its proven inaccuracy, to report ADE causality to their IRBs.1
Investigators reviewed the tools and methods used at all of the NCI centers. They found that none of the forms used a validated method for assessing causality, and nearly 80% of them prompting for global introspection.1
The tools used by the cancer centers did not prompt for the information needed to give useful reports to the IRB. From the IRB's perspective, it would be helpful to know whether the consent form needs to be changed or if there is a need for revising the protocol or excluding additional subjects from a study, Belknap says.
"In some cases a protocol might need to be terminated, and to make those sorts of decisions and work with investigators, we need accurate information," he explains.
The tools many clinical trial sites and the 49 NCI centers use to capture ADE information are woefully inadequate and give poor information to IRBs, he adds.
"Getting good data requires meticulous attention to how you gather, describe, and check data," Belknap says. "One reason why it's not done correctly is because it's left to statisticians: 'We'll just gather these data and leave the information in the database and statisticians will figure them out.'"
While this can work sometimes, too often it creates so much noise that the true ADEs are difficult to find, he adds.
For instance, some adverse events are common in a general population. One example is deep vein thrombosis. If a study population has some incidents of deep vein thrombosis (DVT), then this becomes noise to statisticians. They would have a difficult time separating the DVTs that occurred because of the investigational drug and those that appeared irrespective of the drug, Belknap explains.
More precise and accurate ADE reporting will give IRBs more accurate information and give data safety monitoring boards (DSMBs) better quality data on which to base their decisions.
"So the approach we're taking is to try to use methodologies and tools that have been validated and shown to work," he says. "It's a simple idea, and it's surprising that it has not been used more widely."
Some examples of tools that produce better results include the Naranjo Score, which has been validated, and also the Common Terminology Criteria for Adverse Events (CTCAE), Belknap suggests.
"In theory, the NCI centers were using Naranjo and the CTCAE, but when you looked at the forms and tools they used, you could see they weren't capturing the information in a structured way," Belknap says. "Some of the investigators were reporting things perfectly, but it increases the likelihood of omission and error if you don't have a structured way of capturing data."
IRBs can assist investigators and clinical trial sites with improving their ADE data collection through three steps, he says:
Reduce administrative burdens: IRBs need to understand the enormous administrative burden placed on research staff, Belknap says.
"This is a major problem," he explains. "The first thing that needs to be done at most medical centers is to carefully review all things required of investigators and eliminate those that are not providing any value."
Capture data automatically: IRBs and research institutions need to automate data capture as well as possible.
Research sites and IRBs often have cumbersome electronic system interfaces. Sometimes a research coordinator will need to type data from one computer screen into another, a process that places administrative burden on research staff and can introduce errors, Belknap says.
"One thing a center can do is if they have an electronic medical record, then instead of having research staff do double entry of everything, let's repopulate the form and send the information to many places, including the IRB, sponsor, coordinating data center," he adds. "Things need to be done in a thoughtful way that respects the study staff's time and resources."
Identify and capture key information: The third step is to identify information the IRB needs and add those elements in the data capture. They can use evidence-based tools that are validated and reproducible.
"They could use our forms if they want," Belknap says. "We went to a lot of effort to put this together and make it evidence-based, that it might be a good resource for people to see what we did and base their process on what we did."
For example, the model adverse drug event reporting form to which Belknap refers has research staff describe an adverse event according to CTCAE version 3.0 choices and list whether the event is expected.
The model form also includes this partial list of items:
- Describe the event's severity according to the CTCAE's grades 1 through 5;
- Describe the event's seriousness from the choices of death, life-threatening experience, inpatient hospitalization, prolonged hospitalization, significant disability, congenital anomaly, not serious;
- Describe patient outcome from the choices of resolved, resolved with sequelae, ongoing, automatic display of death if chosen in "seriousness of event" query;
- Suspected agent;
- Strength of association:
- Date drug was last administered;
- Medication(s) prescribed to improve symptoms of event;
- Dosage of treatment medication(s);
- Date treatment medication last administered;
- Duration of treatment medication administration;
- Date toxic effect improved, resolved, worsened;
- According to answers given above, did the toxic effect improve with drug discontinuation or treatment or was the association tested with a placebo or did the toxic effect re-appear when a placebo was administered?
- Was the drug administered at a previous time?
- Did the event occur with previous administration of the drug?
- Concurrent medications with risks similar to event?
- Are the patient's underlying condition or other co-morbidities possible causes?
- Anything in the patient's history that indicates a possible cause?
- May any element of the research process be a possible cause?
- According to the answers given above, are there alternate causes that could have caused the event?
If research sites follow these steps, beginning with reducing administrative burdens, then they'll see that the quality of their ADE information will increase at the same time the total time burden on staff decreases, Belknap says.
Reference:
- Belknap SM, Georgopoulos CH, West DP, et al. Quality of methods for assessing and reporting serious adverse events in clinical trials of cancer drugs. Clin Pharmacol & Therap 2010; 88(2):231-236.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content