Lancet journal retracts article on hearing aids and dementia after prodding
retractionwatch.com
retractionwatch.com
I think one issue we have is that authors of a paper shouldn’t be forced to provide “tech support” for life under the sword of retraction. I try to reimplement analysis and algorithms in the literature all the time, and it’s hard to get it right. I don’t want to write to the authors for help without trying my best first. And I don’t get to write to the editors complaining about no response unless I’m thoroughly convinced of an error.
That’s why I think the comment article triggering a real investigation is the right process. It showed that Mur put enough effort into his replication effort by thoroughly describing the process.
The final exchange with the authors was confusing to me, I don’t get what happened. But I do think Mur was right to be frustrated at that moment.
To add some context: a postdoc position typically lasts 1 or 2 years. The investigation plus discussion with the journal easily took 4 months - at least a sixth of the good dr.'s time. And he's got nothing to show for it - he isn't even named in the retraction notice. In this case, it would have been trivial for the journal to do better... but they opted to do nothing, even after being notified of problems, until they received a thorough description. Which they then rejected.
TLDR: the journal's editors, likely all tenured academics, refused to take action when notified and refused to give credit once a postdoc spent months of their time.
Yikes. And the Lancet is generally regarded as a very reputable source. After reading the linked post I think I understand their initial defense, a little anyway.
But Jennifer Ann McCarthy and Oprah Winfrey were a lot more pivotal in starting this problem with real and avoidable children deaths more so than a fraudulent RCT from a now unlicensed charlatan.
I believe it is important to remember people's contribution to problems and where they sat on which side of history.
Cantuaria ML, Pedersen ER, Waldorff FB, et al. Hearing Loss, Hearing Aid Use, and Risk of Dementia in Older Adults. JAMA Otolaryngol Head Neck Surg. Published online January 04, 2024. doi:10.1001/jamaoto.2023.3509
Conclusions and Relevance: The results of this cohort study suggest that hearing loss was associated with increased dementia risk, especially among people not using hearing aids, suggesting that hearing aids might prevent or delay the onset and progression of dementia. The risk estimates were lower than in previous studies, highlighting the need for more high-quality longitudinal studies.
For life changing decitions people should wait until the study has been replicated a few times, and preferabely destiled to a general advice by the FDA or whoever is in charge.
Isolated papers are better than blogpost in allcaps, but they are not "setled" science.
The problem is that the press release of the universities and the report by science "journalism" present these single paper result as a definitive result.
“In people with hearing loss,” the authors wrote, “hearing aid use is associated with a risk of dementia of a similar level to that of people without hearing loss.” They proposed “up to 8% of dementia cases could be prevented with proper hearing loss management.
From the article, describing the replication work which led to its retraction:
Mur and his colleagues intended to build on the article with a related analysis on the same UK Biobank data. But when he couldn’t replicate the main findings, Mur scrutinized the paper more deeply. [...] Most notably, he found that hearing aid use did not correspond to a lower rate of dementia for people with hearing loss, as the authors reported. He found the opposite: among people with hearing loss, the dementia rate was higher for those using hearing aids.
I don't see this one retraction proves/disproves causality one way or the other; without looking at all other studies on the same thing.
It does however prove that The Lancet isn't very scrupulous about forcing authors to retract even when a paper is discredited, or properly crediting third-parties with discovering serious errors (rather than fogging things to suggest the authors voluntarily initiated the retraction, which they didn't).
Second: in my complete and utter layman's view, it is far more likely that the causation is the other way around. That is: dementia could come across as hearing loss for the affected individual (or their nearest). As such, people with dementia would be more likely to get hearing aids - and therefore people with hearing aids would be more likely to suffer from dementia than people without.
This is complete and utter speculation, but this line of reasoning makes more sense to me than the reverse (using hearing aids would aggravate dementia).
Nevertheless, can't tell without data.
Ehh, that's debatable.
There's been a decades-long campaign to malign science and science-related institutions by big tobacco, big oil, various religious sects, and political groups because science regularly gets in the way of those groups' profits/beliefs/legal maneuvers.
While scientists do regularly pick up a footgun, it's important to acknowledge the various powerful entities that want them distrusted and pay big money to further that goal.
Or did you mean the resistance to retraction was what is killing science?
It's true there will probably be an increase in this kind of bad article retraction before we can see a decrease, and that wouldn't necessarily indicate an increase in bad science -- similar to certain crime statistics when reporting starts to be taken seriously. But I still think the article raises quite a few ongoing concerns with the current academic system, which does have capacity pressure/incentive problems to a much greater degree than previous generations.
I do agree retractions in general should be more normalized, as there are also plenty of cases where honest non-reckless mistakes were made, and sometimes even quality work that turns out to have issues in retrospect e.g. new information invalidates a previously common assumption. But I'm not sure any of that applies to this particular paper, and regardless the original authors apparently avoided addressing any of it.
If "most" means "most published papers" and "scrutiny" means "can replicate" then no, it doesn't withstand scrutiny: most published research is wrong. https://youtu.be/42QuXLucH3Q (2016)
(the exact percentages of non-replicatable articles may be different in different fields)
EDITED to add:
What I mean by progress, is measurable changes in society e.g. the improvement of medical science leading to measurably better survival, longevity etc outcomes, or the improvement in chemistry, materials and solid-state science effecting significant change in electronics, communications, batteries etc.