A team of Boston- and Stanford-based researchers has devised a less costly, clinic-based approach to randomized clinical trials (RCT) that shifts patients to a more successful therapy as the trial progresses. The method, based on the Veterans Affairs (VA) electronic medical records system (EMR), is currently being tested in a comparative effectiveness trial of two insulin regimens in the Boston VA system.
The new approach, called a point-of-care clinical trial, aims to apply the statistical strength of an RCT to the real-world settings of observational studies.
“We’re trying to build in [to medicine] a process of learning, and we’re trying to do that in a way that’s both valid and not terribly expensive or slow,” said Philip Lavori, professor of biostatistics at Stanford.
Lavori designed the new system in collaboration with the Boston team. The fruits of their research efforts were published in Clinical Trials this month.
In clinical medical research, randomized clinical trials are recognized as the gold standard, because they can determine without bias whether drugs or medical technologies work and which among them are most effective. Since these trials are often prohibitively long and expensive, investigators instead use observational data from everyday care with wide generalizability. Without randomization, however, observational studies cannot eliminate inherent biases that arise from knowing which patients use which therapy.
“The problem we have right now is that the two methods we have for learning are either just to look at data that occur in the ordinary course of treating patients…or we can do these big, expensive, but very valid studies and get very good answers,” Lavori said.
“We’re in the middle,” he said of the new method.
According to researchers, the new method will be used for pragmatic clinical trials, which examine the comparative effectiveness of approved drugs in everyday patients.
“We’re not looking to introduce new treatments — we’re interested in looking at which treatments are more effective than others,” said the study’s lead author Louis Fiore, who is the director of the Department of Veterans Affairs Cooperative Studies Program Coordinating Center in Boston.
The point-of-care approach embeds randomization into everyday clinical care. When a physician at the VA prescribes an insulin regimen to a diabetic using the EMR, a dialog box usually gives two options: one for the control treatment and one for the experimental treatment. The box now includes a “no preference” option, which, if selected, explains the trial to the patient. Furthermore, if consent is given, the EMR-based system randomizes the patient to one of the two therapies.
The data inputted to the EMR during physician visits is automatically gathered for the trial and used to track the overall progress of patients in the two insulin therapy groups. If one type of therapy consistently outperforms the other, the randomization tool begins to put more patients into the successful therapy pool. This process continues until the probability of one therapy being better than the other exceeds 99 percent.
All of the randomization and data collection is done within the EMR, making it easier for investigators to carry out comparative effectiveness studies.
In light of the growing focus on health care costs, comparative effectiveness research can help determine if more expensive therapies actually outperform their cheaper or generic counterparts. The government has therefore increased funding for the research, creating a huge new market for the VA team’s design.
“Comparative effectiveness research really has taken on new life in the last couple of years,” said study co-author Ryan Ferguson, who is a member of the Boston team.
The team eventually wants to expand this system to different arenas of medical research. Fiore and Ferguson said they were investigating the possibility of conducting a larger study on PTSD, the prevention of acute kidney injury, pain management or infectious diseases. They also aim to expand the VA system beyond Boston and the Northeast region.
“In five or 10 years, I think the only surviving health care systems will be the ones that could do something like this,” Lavori said. “And then we hope it will be seen as a comparative advantage.”
He noted that the long-term goal is to have “automated and rapid improvements in care and to build that into the system so that the care just keeps getting better.”