Along with a story on aspirin and macular degeneration, my editors got a panicked note from me recently: “I struggled with the numbers on this one, and may not have gotten them right,”I wrote.
Odds are that I wasn’t the only reporter puzzling over the numbers in this study, which was published in JAMA Internal Medicine. That’s because some of them were made up, apparently by a journal editor.
The error added a scary statistic to research that was already generating some pretty alarming health headlines (i.e. “Aspirin Ups Risk of Age-Related Macular Degeneration,” courtesy of a website for doctors called Clinical Advisor.)
To make matters worse, the mistake got repeated in the press release. (Et tu, media relations team?)
Thankfully, it seems many health reporters noticed something was amiss and left the wrong numbers out of their stories.
Several news stories did repeat the bad stat, however. I won’t call them out, but the fact that it happened should remind us not to write on autopilot. It’s better to be confused and ask questions than to dismiss those feelings and simply parrot the press release to meet a deadline. It also underscores the importance of reviewing the findings carefully with the study researchers. They aren’t there only for quotes.
Here’s what happened: The study followed nearly 2,400 older adults in Australia. Study participants answered questions about aspirin use, and they got periodic eye exams to check for changes to their retinas, the light sensing part of the eye. The macula, incidentally, is a spot in the center of the retina.
Fifteen years later, 63 adults in the entire study, about 1 in 38 people (2.6 percent), had developed the aggressive “wet” form of the disease. Left untreated, “wet” macular degeneration can cause blindness.
But that’s not the statistic that made it into the text of the study. Both the abstract and results section say 63 of the 257 people in the study who were regular aspirin users (24.5 percent) developed “wet” macular degeneration. Think about that. If it were true, that would mean the study found that nearly 1 in 4 regular aspirin users developed the blinding eye condition.
Observational study or not (see correlation vs. causation), a statistic like that could lead plenty of people to rethink a daily aspirin regimen.
And it contradicts other statistics published in the article. According to the results, the 15-year cumulative incidence of “wet” macular degeneration among regular aspirin users was 9.3 percent.
How could 24.5 percent of regular aspirin users develop this form of macular degeneration over 15 years when just 9.3 percent of them do just a few paragraphs later?
Table 3 offers other clues that the 24.5 percent stat is off target. This table compared causes of macular degeneration in both nonregular and regular aspirin users who did and did not have cardiovascular disease. If you add across columns and rows, you see that only 15 regular aspirin users developed macular degeneration over the course of the study, the other 48 were nonregular aspirin users. That’s where the 63 comes from.
I was pretty sure table 3 explained things, but I still wasn’t sure. I emailed the study researchers for help. Because of the 16-hour time difference between us, I knew I wouldn’t hear back from them until after deadline (unless, fingers crossed, one of them had insomnia.)
I turned in a story that contradicted the press release and study text, but I explained the confusion to my editors and hoped they wouldn’t think I was crazy.
A few hours later, the study authors replied. They seemed just as confused as I was:
“The 24.5% figure is incorrect and is a post-acceptance editorial addition which we did not pick up in the proof. Thank you for pointing it out,” says lead study author Jie Jin Wang, Ph.D., a senior research fellow at the University of Sydney.
Wang and her co-authors have alerted the journal’s editor to the error, and they apologized for not catching the change during the proofing stage.
Rita Redberg, M.D., a cardiologist at the University of California San Francisco and editor of the journal, didn’t offer an explanation of what happened but said they would correct the article.
Still, with JAMA articles and correspondence about them are behind pay walls, reporters who wrote the original stories may not learn about the correction. It’s worth noting that AHCJ members do have access to the publications.
I contacted the media relations team to see if they might quickly alert reporters to the mistake so stories could be corrected in a timely fashion.
“Hard to reel the information back in after it’s out there, as you well know. And even if we put out a correction, very few media outlets would publish that information.”
Actually, it’s not that hard. Media outlets correct stories all the time. Perhaps the bigger worry here is that media relations would have to admit that the release wasn’t properly fact-checked in the first place. Now that is a blind spot.