Electronic medical databases could replace or complement random-controlled clinical research trials, Agency for Healthcare Research and Quality Director John Eisenberg, MD, recommended Nov. 20.

With the onset of technological advances, industry should consider "instances in which large databases can replace or at least live side-by-side with RCTs," he said at NIH/HHS' "The Science of the Placebo" workshop, held in Bethesda, Md., Nov. 19-21.

"I don't think that we've adequately investigated this alternative to RCTs," Eisenberg said. Today, large databases "ask questions that only an RCT could have answered a half century ago, to deal with the challenge of controls and to simulate, in some ways, a placebo," he said.

There may be opportunities where observational and RCT trials can "peacefully coexist at some times," the AHRQ director suggested. Observational studies can, for example, "help to generate hypotheses for more focused RCTs" or "extrapolate from RCTs to more heterogeneous populations."

Databases are wrought with inherent problems, FDA Associate Director for Medical Policy Robert Temple, MD, countered. While he expressed hope for database use, Temple questioned whether such analyses are useful to measure small differences between two active drugs or to "tease out effects among subsets."

"What we need to do," Eisenberg responded, "now that we have access to large data sets, is to do a better job in collecting the information....What I'd like to see us do is to collect real-time information that's linked" with the care provided to the patients.

Edward Davis, PhD, University of North Carolina, said the "main problem" with databases is the "potential bias" in how patients are selected to receive therapy. He pointed to estrogen observational studies, noting that women who received estrogen were upper-middle class and much more likely to be health-conscious, which may have skewed the results.

The differences between patients offered entry in placebo-controlled trials compared to those invited into active-controlled trials is one area for further research, Temple maintained during a breakout session.

Shepard Siegel, PhD, McMaster University, concurred, adding, "you might either over- or under-estimate the amount of the placebo response because you're dealing with a very needy population."

Since physicians are less likely to enroll patients who are expected to respond in an active-control trial, physicians may be selecting treatment-resistant subjects for placebo-controlled trials, Charles Weijer, MD/PhD, Dalhousie University, said. "Similarly, patients who have had bad drug experiences may be more likely to enroll in a placebo-control trial."

Physicians may be "singling out" subjects "who are refractory to the standard treatment" for placebo-controlled trials, Robert Levine, MD, Yale University School of Medicine, agreed. These patients, "particularly in the mental health field, might be biologically very different from those who respond to treatment."

Eisenberg's recommendations were made as a proposed option for the ongoing debate on the ethics of placebo-controlled clinical trials. The deliberations have been heightened by the October amendments to the World Medical Association's Declaration of Helsinki ethical principles on clinical trials.

The amendments recommend the effectiveness of a new drug therapy be tested against the "best current prophylactic, diagnostic and therapeutic methods."

Temple reiterated FDA's concern with the Helsinki principles, saying the document undermines the assumption that individuals can decide on their own whether to join a placebo-controlled study. Active-control studies also make it difficult to measure efficacy and require large cohort groups, he said.

Temple acknowledged problems with placebo-controlled trials, however, and suggested a number of improvements to such studies. Offering an "early escape" by defining patients who fail on therapy and allowing them to quickly leave can reduce the time spent without treatment, he suggested.

Other options include measuring time-to-first-occurrence, in which subjects are taken off the therapy and after reoccurrence is measured, are removed from the trial. Non-responders could also be enrolled, although since such patients may not respond to any therapy, that could be "tricky," Temple said.

Electronic Databases Could Replace Random-Controlled Trials, AHRQ Says

More from Archive

More from Pink Sheet