Cancer Screening Trials Rarely Tally Harms, Says Study

Nick Mulcahy

September 19, 2013

Investigators of cancer-screening interventions such as mammography and colonoscopy "seldom quantify the harms" of their efforts, according to the authors of a quantitative analysis of randomized trials.

The "most important harms" — overdiagnosis and false-positive findings — were quantified in only 7% and 4%, respectively, of the 57 cancer-screening trials reviewed, they note.

"We found that the harms were poorly reported," write the authors, led by Bruno Heleno, PhD, a fellow in the Department of Public Health at the University of Copenhagen in Denmark.

The results were published online September 16 in BMJ.

However, an expert in mammography screening disagrees with this conclusion — strongly.

"In my view, [cancer-screening] trialists have, in the main, done their best to describe both positive and negative effects of screening," said Stephen Duffy, MD, professor of cancer screening at the Wolfson Institute of Preventive Medicine in London, United Kingdom. He was not involved in the study.

Furthermore, "this paper gives a false impression about the reporting of negative effects," Dr. Duffy said in an email to Medscape Medical News.

He criticized the study on a number of points, including its faulty accounting.

"Without even looking at the literature, I can think of 4 mammography trials that estimate overdiagnosis, whereas the table in the paper reports only 2," he said.

Even if the study criteria for excluding certain trials were less stringent, it would "not change the conclusion" about substantial underreporting, Dr. Heleno told Medscape Medical News.

The benefits of cancer screening were quantified much more often than the harm-related outcomes in trials, the authors report. For instance, cancer-specific mortality and cancer-specific incidence were each tallied in more than 80% of the trials.

The pluses and minuses of screening should be "equally well quantified," they argue, because screening typically occurs in healthy people who deserve balanced information.

The authors also employed a "crude" quantitative method to assess harms reporting, according to Dr. Duffy. They measured, with a ruler, the amount of physical space that is devoted to discussing harms in the published papers.

The authors found that the median percentage of space in the results section devoted to harms was only 12%.

Seven Harms Assessed

The authors searched a variety of databases to find trials that evaluated cancer screening. They looked, for example, for screening trials for breast cancer with mammography, self exam, or clinical exam; colon cancer with endoscopy or fecal occult blood testing or virtual colonoscopy; and lung cancer with chest x-ray or low-dose spiral computed tomography.

Individual articles were eligible if they provided data for both the screening and control groups.

They found 57 trials involving 10 different screening interventions and more than 3 million study participants. The trials were then assessed for the inclusion of 7 different types of harm.

The harms were assessed in a low percentage of the trials, with a couple of exceptions: overdiagnosis (7%), false-positive findings (4%), somatic complications (19%), negative psychosocial consequences (9%), invasive procedures (47%), all-cause mortality (60%), and withdrawal because of adverse events (2%).

However, to be counted as having fully assessed a specific harm, a trial had have quantified that specific harm for both the screened and control groups.

Dr. Duffy criticized the authors' claim that only 4% of screening trials reported false-positive results.

He used his experience in mammography trials to roughly gauge the accuracy of the finding in the larger set of 57 trials. "All the mammography trials have reported quantitative false-positive results. Therefore, again, the paper is misleading."

The authors point out that only 4% of trials reported false-positive results for both the screened and control groups, although they acknowledge in their study that 32% of the 57 trials analyzed provided false-positive information for the screened group alone.

They chose to emphasize the 4% finding because they believe that randomized cancer-screening trials should be held to the highest standards.

"Reporting harm outcomes for the intervention and control groups is a central recommendation in the guidelines for reporting randomized trials," they write, referring to trials for drugs and other medical interventions.

The authors have disclosed no relevant financial relationships.

BMJ. 2013;347:f5334. Abstract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.