Randomized controlled trials are considered to be the gold standard for evaluating new therapies as well as proving the effectiveness of screening exams. In this type of study, participants are randomly assigned to an experimental group or a control group. The evaluation includes all participants.
However, when the purpose is to discover the value of screening exams such as lower GI endoscopy, a problem arises: participants in an experimental group frequently do not undergo the procedure. On the other hand, participants from the control group may do so outside the study protocol – if for example, they have a health concern.
“This means that the difference between experimental and control groups is diluted, making the protective effect of endoscopic gastrointestinal screening appear smaller than it actually is," says Professor Hermann Brenner from the German Cancer Research Center. Brenner’s team at DKFZ has now carried out calculations to quantify the amount by which this effect is likely underestimated. The team based their calculations on data from four completed randomized trials on the effectiveness of an endoscopic examination called flexible sigmoidoscopy.
The actual participation rates in from members of the experimental groups were generally 70 percent or even lower. On the other hand, during the relevant period, up to 50 percent of participants from the control groups underwent lower gastrointestinal endoscopy outside the screening programs.
Under these conditions, if lower gastrointestinal endoscopy were 70 percent effective in coping with the incidence and mortality of colorectal cancer, the usual study evaluation method would reduce that figure to approximately 38 percent. If the procedure prevented the incidence of cancer cases by 50 percent, the result in the standard evaluation would be a much less convincing 23 percent.
“Randomized controlled trials do provide information about the effect that screening programs have on the colorectal cancer incidence in the whole population that is offered the examination," Hermann Brenner says. “However, when considering the protective effect of examinations that have actually been conducted, the rate is much higher and should be correctly communicated in colorectal cancer screening consultations."
Brenner expects even greater discrepancies to emerge during ongoing studies with a focus on full colonoscopies, because participation rates in the experimental groups are yet lower and, at the same time, the proportion of people in the relevant age group who receive lower gastrointestinal endoscopy outside screening programs continues to rise.
Hermann Brenner, Christian Stock, Michael Hoffmeister: In the era of widespread endoscopy use, randomized trials may strongly underestimate the effects of colorectal cancer screening. Journal of Clinical Epidemiology 2013, DOI: 10.1016/j.jclinepi.2013.05.008