Media release

From: Bond University

A large global study published today in the leading US medical journal Jama Internal Medicine finds the media is often failing to cover the potential harms of new medical tests which are designed to detect the early signs of disease in healthy people.

The study also found media stories often fail to report on conflicts of interest of researchers promoting the new tests.

Researchers from the University of Sydney and Bond University analysed over 1100 recent news stories about five new early detection tests including:

* Liquid blood biopsies to detect cancers

* Apple Watch to detect Atrial Fibrillation

* Blood tests for dementia

* Artificial Intelligence tests for dementia

* 3D mammography for breast cancer

Failure to cover important potential harms

While almost all the 1173 analysed stories reported on the potential benefits of these tests, over 60% of stories failed to make mention of any potential harms.

One story claimed: “A simple blood test can now detect dementia decades before any symptoms appear...” and advocated using the test during routine health checks.

Another story provided unbalanced coverage of a new test for multiple cancers: “Dubbed the ‘holy grail’ for cancer care, liquid biopsy could significantly shorten diagnosis time (and) find very early-stage cancer in those with no symptoms.”

“For all five of the tests covered in our JAMA Internal Medicine study, there is concern that some of the healthy people tested will be wrongly classified as sick,” says Dr Ray Moynihan, Assistant Professor at Bond University and senior author on the study. “Yet this potential harm of overdiagnosis was only mentioned in 5% of stories.”

Lead author, Dr Mary O’Keeffe from the University of Sydney, said: “While these tests sound appealing, and can bring benefits, whenever we test healthy people there’s a danger that some people will be wrongly classified as sick. This can cause overdiagnosis and overtreatment.”

Overdiagnosis occurs when people receive a diagnosis of a disease or condition that will never develop to cause any symptoms or early death. It is an unnecessary diagnosis that does more harm than good. Evidence published last year in the Medical Journal of Australia shows that for common cancers, around 1 in 5 cancers may be overdiagnosed.

Failure to cover important conflicts of interest

Another key finding from the study is the widespread failure of media stories to cover important conflicts of interest, such as commentators receiving payments from the companies marketing the new tests. For example, 19 of 22 authors of a key trial examining the ability of the Apple Watch to detect atrial fibrillation disclosed taking grants or personal fees from Apple, yet this information was rarely reported in news stories.

While over half the 1173 stories in the study included the views of commentators with important financial conflicts of interest, these conflicts were only covered in 12% of stories.

Strategies to improve media coverage needed

The paper concludes: “Strategies to improve media reporting so that professionals, patients and the public receive more balanced information about early detection tests are urgently needed.”

“Higher quality medical reporting is more complete medical reporting – covering benefits, harms, and conflicts of interest,” says Dr Moynihan, who is working with Dr O’Keeffe and colleagues on a pilot study of new training interventions for journalists.

Their work is supported by the Wiser Healthcare Research Collaboration investigating overdiagnosis and overtreatment in Australia, funded by the National Health and Medical Research Council.

Key information and findings about the study published today in JAMA Internal Medicine

* Researchers analysed 1173 English-language media stories from around the world about 5 new tests, between 2016 and 2019

* 97% of stories reported on benefits

* 37% of stories reported any harms

* Harms were mentioned, but de-emphasized in 27% of stories

* Overdiagnosis was mentioned in 5% of stories

* 55% of stories included the views of commentators with conflicts of interest, but conflicts were only disclosed in 12%