PA Media corrects error in reporting on research for prostate cancer test

A number of news articles misreported new research into a test for prostate cancer. The Telegraph, Independent and ITV News all incorrectly stated that all the men involved in the research had prostate cancer. 

In fact, the study shows, albeit unclearly, that the majority of men–97 of the 147–enrolled in the research did not have prostate cancer.

This is important because the study did not just look at how successful the test was at detecting prostate cancer in men who have it, but also at how accurate it was at identifying men who did not have prostate cancer.

Full Fact identified other errors in the reporting which stem from the original research, some of which stemmed from the reporting and some from the paper itself.  

As a result of Full Fact’s enquiries, PA Media updated its copy and the Daily Mail has also amended its article. The study authors have advised us they will add a comment to the webpage of their study to note their error.

What the study was about

The study was conducted by a team at Oxford BioDynamics, a company that produces medical tests, in collaboration with Imperial College and the University of East Anglia (UEA).

The researchers looked at tests for prostate cancer. The study claimed that using the company’s new “PSE” test alongside the currently used blood test (Prostate Specific Antigen or PSA) improved the accuracy of both positive and negative results. 

PSA, the test used currently, is imperfect as most patients who test positive will end up not having cancer, but it is reasonably good at not giving “false negatives”, meaning it rarely tells men who genuinely do have prostate cancer, that they don’t. 

The study authors wrote: “The widely used prostate-specific antigen (PSA) blood test does not have sufficient accuracy, resulting in numerous unnecessary prostate biopsies in men with benign disease and false reassurance in some men with cancer.”

The results of the study found that 93% of positive results with the combined PSA/PSE tests were true positives (in which the patients actually had cancer) and 95% of negative results were true negatives (in which the patients definitely didn’t have cancer). The average of this, 94%, was used by most news outlets to describe the “accuracy” of the tests.

What the outlets got wrong

As mentioned, many media outlets incorrectly reported that all the men in the study had cancer. 

Full Fact found that the articles which made this error appeared to use the PA wire copy in their reporting. We contacted PA Media, and the wire service subsequently told Full Fact it had updated its copy.

But this wasn’t the only error. The Daily Mail, in its original write up, wrote that, with the combined tests, 92% of positive results were true positives, and 94% of negatives were true negatives (not 93% and 95% respectively as reported by others). 

The Mail had correctly copied these numbers from the study’s abstract (or summary). However, the numbers in the abstract did not match the numbers reported further down in the study. 

Full Fact asked the lead researcher, Professor Dmitry Pshezhetskiy from the UEA, about this discrepancy. He told us the abstract was not going to be corrected but that a comment would be added to the webpage for the article. At the time of writing, this comment was not visible. 

The Mail also stated that “most of the 147 men in the study had a positive result for PSA” but this was incorrect as only 57 of them did. It has corrected the article on both of these points.

It should also be noted that the study uses a small sample size of 147, and the numbers analysed using the statistical test appear to be even smaller with 101 participants included in the results table for the headline results. The authors told Full Fact that the sample size was big enough to reach statistical significance but that further studies were definitely required, and with a lower proportion of cancer patients. 

Dr Alexandre Akoulitchev, Chief Scientific Officer at Oxford BioDynamics told Full Fact “It is everyone’s intention to expand the validation cohort and accumulate real life clinical readouts with early adopters in clinical practice.”

We have previously written about other studies reported in the media, commonly including errors such as mistaking correlation for causality, and failing to pick up on flaws in the research itself.

Image courtesy of Testalize.me