Study questions true extent of trial effect’
No evidence to support long-held belief
A new study from researchers at the Dana-Farber Cancer Institute in Boston indicates that a long-held belief among oncologists — that patients who participate in clinical trials have better outcomes overall than those who do not — may not be supported by empirical evidence.
In a report published in the Jan. 24 issue of the journal Lancet, investigators reviewed 26 studies that compared outcomes among trial and non-trial participants. Fourteen studies showed some evidence that trial participants had better outcomes, but just nine of the trials were designed to compare the outcomes of the participants with those nonparticipants who also would have been eligible for the trials. Of these, only three studies suggested better outcomes among trial participants than among nonparticipants. No studies showed that participants had worse outcomes than nonparticipants.
The findings seem to indicate there is little strong evidence of a "trial effect" in cancer research, the researchers say.
"Clinical trials are critical to the advancement of cancer care, but it is important that people who enroll in a study understand that their participation is intended primarily to benefit future patients," says lead author Jeffrey M. Peppercorn, MD, MPH, a clinical research fellow at Dana-Farber.
Many clinicians believe that patients who participate in clinical trials frequently have better outcomes than nonparticipants regardless of the efficacy of the experimental treatment involved. Known as the trial effect, it’s believed that the increase in clinical monitoring and supportive care can help improve outcomes among research participants.
Although investigators found a few instances in which trial participants may have had better outcomes than nonparticipants, the limitations of the data did not allow for a definitive link between participation and improved outcomes, adds study co-author Steven Joffe, MD, MPH, instructor in pediatrics at Dana-Farber and Children’s Hospital in Boston.
Although about half of the studies found some evidence of a trial effect, methodological difficulties with the studies’ design makes the resulting interpretations questionable, the authors state.
They contend that there are four possible reasons that trial participants could experience an improved outcome vs. nonparticipants:
- Experimental treatment effect — the experimental treatment offered is an improvement over standard therapy.
- Participation effect — aspects of trial participation other than the investigational treatment improve outcomes (changes in doctor-patient relationship, improved monitoring, etc.).
- Confounding case characteristics — differences in baseline characteristics (age, gender, socioeconomic status) that might influence both outcome and trial participation.
- Bias in collection of data — follow-up might be more complete in trial participants that nonparticipants, or publication bias might result in dissemination of studies that find an effect in certain populations whereas researchers have little incentive to publish investigations that show no effect.
Collection bias and reporting bias can work both ways, the researchers note, concealing some trial effects while indicating effects where none truly exist.
To accurately evaluate a trial effect, it is particularly important to identify an appropriate comparison group — that participation be the only difference between participants in a trial and nonparticipants.
This might best be accomplished by comparing outcomes among participants in a trial and those who were offered study participation, but declined. This strategy is not foolproof, but offers several advantages over the other methods used, they said.
"More research is still needed to determine whether trial participation can confer a predictable benefit," Joffe reports.
However, oncologists should be sure to emphasize the true goals of research trials when recruiting patients and not any perceived direct benefit of participation, say Joffe and Peppercorn.
Public statements that indicate trial participation is a key in obtaining better outcomes for patients can be misleading for both patients and their physicians, they add.
However, the belief that clinical trials offer the best treatment for patients with cancer is wide-spread in the oncology community, the authors note.
For example, the American Federation of Clinical Oncologic Societies states, "Treatment in a clinical trial is often a cancer patient’s best option."
Such claims suggest that trials are viewed not only as a way to improve future treatment, but also as the best treatment for current patients. The view that trials lead to better outcomes, if correct, has important implications. First, that more than 95% of adults and perhaps 40% of children with cancer do not enroll in trials would constitute evidence of substandard care, the authors add.
- Jeffrey M. Peppercorn, MD, MPH, Clinical Research Fellow, Dana-Farber Cancer Institute, 44 Binney St., Boston, MA 02115.