On Friday October 6, shortly after our reviewers called its news release about a text messaging study “an exercise in spin,” George Washington University (GWU) withdrew that news release without explanation.
The following week, apparently after careful consideration, the news release was re-issued with a one-word change to the original: the addition of the word “some” to the headline.
Here’s the original headline: Text messaging program may help pregnant women kick the smoking habit
Here’s the revised headline: Text messaging program may help some pregnant women kick the smoking habit [emphasis added]
An asterisked footnote said, “The headline of this release was edited for clarity to match the content in the body of the release.”
In their evaluation of the original news release, our reviewers expressed major concerns about the overall framing of the study results and how this seemed to conceal the fact that the results for the primary outcome were negative. They wrote:
“The reduction in smoking rates for the study overall was not statistically significant — full stop. That should have been the headline. It’s not acceptable for the release to bury the lack of statistical significance in the body of the text while claiming benefits in the headline.”
While slightly recalibrated, the revised, corrected headline maintains the framing of the original. And the negative result for the primary outcome is once again buried in the fifth paragraph.
We reached out to public relations experts at several academic institutions to ask them whether they thought the new, corrected headline was an accurate reflection of the study results. They didn’t mince words.
“The paper is a negative study,” said Preeti Malani, MD, a professor of medicine at the University of Michigan and HealthNewsReview.org contributor. “And that’s not how this press release reads.”
Preeti Malani, MD
Malani is also an associate editor at JAMA and previously helped oversee the JAMA media relations department. She said the benefits touted in the headline were secondary outcomes that should be considered “exploratory and hypothesis-generating.” The fact that they are emphasized in the headline and lead sentence will create a misleading impression for readers, she said. “We would never want people to walk away from this thinking, ‘Wow, these text messages are great,’ because that’s not what the study found.”
She suggested creative ways to frame the negative findings that wouldn’t run the risk of misinforming readers. “The fact that there are so many pregnant women smoking, that’s a big problem,” Malani said. “And maybe that’s one way to present the findings while still keeping it interesting and true to the science. Maybe your headline could be something like, ‘No magic bullet to help women quit smoking during pregnancy.'”
Earle Holland, who was the senior science and medical communications officer at Ohio State University for 35 years before retiring, was even more pointed in his criticism.
“There’s a fundamental dishonesty here, first in the original release and now in their so-called remedy,” said Holland, who is also a HealthNewsReview.org contributor. “It suggests that GW media folks’ main interest is in painting a rosy picture rather than accurately explaining the outcomes of their research.”
Holland said that adding adding “some” to the headline makes it “more accurate, but not accurate enough,” adding that “the findings weren’t statistically significant, so the whole thrust of the release should have been reworked. If they were going to take it off EurekAlert! — which is basically a retraction — only to make a change, then they should have fixed everything.”
Michael L. Millenson, a health care consultant and researcher at Northwestern University who has studied the evidence base for health apps targeting consumers, said: “It’s distressing to see an academic press release with the kind of ‘spin’ on an app’s effectiveness the public might expect from a for-profit organization, not a university. While the details of the press release are careful not to make undue claims, the headline and first paragraph just as carefully imply more significant results than appear justified.”
Millenson added, “The FDA under the Trump administration has said it wants to pull back on premarket regulation of apps. The evidence base for these type of interventions is already modest, and the relaxing of regulation makes rigorous and reliable studies that much more vital.“
GWU’s Director of Media Relations, Kathy Fackelmann, did not respond specifically to our request to address this criticism. When she notified us that the corrected release had been posted, she that GWU was “in the process of reviewing our news release practices keeping in mind some of the points in your review. We will also reach out to the reporters who contacted us about the release and send them the revised version just so that they are aware of the updated release.”
Fackelmann’s comments raise the issue of how to deal with inaccurate news releases — a subject that also sparked strong opinions from experts. While some said it was appropriate to remove a flawed original and replace it with a corrected version — as GWU did in this case — others argued that the original should be maintained as part of the public record.
Joann Rodgers, a journalist and author who formerly led Johns Hopkins Medicine’s communications and public affairs division, said it was important to always consider the end user of the information in these types of situations.
“In this case, that would call for ‘repeal and replace,’ with a note from the university public information officer/editor that a release titled (original title) was in error, has been removed from the site and replaced with a corrected version. It’s also a best practice to spell out the nature of the original error, so long as that does not lead to confusion.”
Rodgers, a HealthNewsReview.org contributor, said the GWU correction was inadequate to address the magnitude of the initial error, and that the “headline/lede graphs should have emphasized the ongoing difficulties in getting even motivated groups to quit smoking.” A secondary message could be that “some interventions work for some people some of the time,” she said.
Matt Shipman, a public information officer at North Carolina State University and HealthNewsReview.org contributor, said that “minor errors should be corrected, and the relevant release updated, as quickly as possible. I also think it’s appropriate to include a note specifying that changes were made and what those changes were.” But “If the errors are so egregious that the release is effectively moot, I think the entire thing should be taken down,” Shipman added. “In either case, the institution should attempt to identify any news outlets that ran stories based on the release and notify them of the mistakes as quickly as possible.”
However, Earle Holland, formerly of Ohio State, said that it was bad practice to delete the original flawed version of a news release that has been retracted. He likened it to “trying to rewrite history, or at least the public record,” and said that the flawed version would continue to linger on the Internet despite being scrubbed from university websites and PR wires.
Moreover, he said that at public universities, news releases may be subject to the state’s public records laws, which in most cases require the university to maintain the integrity of its official documents. Altering a news release might represent a violation of such laws, Holland said.
He noted that GWU, as a private institution, would not fall under any legal requirement to safeguard public documents. But he believes it’s still in the best interests of the reader to maintain the original flawed version, so that it’s available for comparison against the corrected version.
“The best way to handle situations where a release is found to be wrong is to add an editor’s note to that release stating the error and, if available, a link to a revised/corrected version of the release. It’s key to make sure the original release and any ‘fixed’ version are inseparable.”
Holland said it was important to admit to errors when they occur, which helps reinforce to the public the institution’s commitment to honesty and transparency. This also will remind some readers that science is often self-correcting, he added, and that findings can change as new information is available.
“And that is a message that is well worth sending,” Holland said.