Hearing Aids are great, even if they don't save lives
Not even a week passed in 2024, yet we already have a contender for the most problematic medical paper of the year.
The Study of the Week is back after a nice holiday break. Peace to all. And Happy New Year.
The journal LANCET Healthy Longevity now has the early lead for publishing the year’s most dubious study.
The top-line result of this observational study was that use of hearing aids in adults with hearing loss was associated with a 24% lower rate of death compared with nonusers of hearing aids. This was statistically significant.
Grin. My friends, statins don’t reduce death when used for primary prevention. The anointed new heart drugs, sacubitril/valsartan and SGLT2i don’t reduce overall death in some forms of heart failure.
But hearing aids? The devices slash death by 24%!
And this was the lead from nearly 150 news outlets—including… gulp…national public radio.
Here is what the authors did: They used data from a longitudinal study called NHANES. NHANES is one of the many registries that measure health parameters in adults and then follow them for years. Such registries create huge databases.
In this study, among 10,000 adults, nearly 2000 had completed hearing tests and had hearing loss. Over the many-year study, hearing loss occurred in slightly less than one in five of these middle-aged adults. Roughly 13% of these adults regularly used hearing aids. Hearing loss was an independent risk factor for higher mortality.
None of this anodyne information would make news.
Here comes the newsmaker:
The authors created three groups of adults based on hearing aid use—never, regular, and non-regular users. They gained this information from asking patients. Never-users of hearing aids is like the control arm.
Boom. Those adults who regularly used hearing aids had a 24% lower rate of death. (Hazard ratio 0·76 [0·60–0·95])
Here were the authors’ conclusions. I highlighted the causal language.
Regular hearing aid use was associated with lower risks of mortality than in never users in US adults with hearing loss when accounting for age, hearing loss, and other potential confounders. Future research is needed to investigate the potential protective role of hearing aid use against mortality for adults with hearing loss.
Here was the NPR story. It made the All Things Considered show.
Note the causal language, boost longevity. The NPR story contained no comment on limitations. But added this sentence for emphasis.
The effect held up even when the researchers accounted for differences such as age, ethnicity, education, and medical history.
I singled out NPR because I would have thought that they would have been above this level.
This is a broken study. One that should never have been done never-mind published or covered in news. The problems are many.
Implausibility: It is utterly implausible that a hearing aid can lower mortality by 24%. JAMA-Internal Medicine just published a large meta-analysis showing that nearly all cancer-screening tests, including colonoscopy, mammography and PSA screening, have zero effect on lifespan gains. Hearing aids are great for people because it’s nice to hear. But everyone knows that a hearing aid is not going to reduce death.
Imbalanced Groups: Table 1 of patient characteristics reveals that there are nearly 1500 adults in the never-user group but only 237 adults in the regular-user group. Regular users have much more favorable characteristics: higher education, less poverty, rarely uninsured, lower rates of smoking and they are more likely to be of normal body weight.
The authors say they adjust for these differences in their model. But regular readers of Sensible Medicine know that researchers can only adjust for characteristics on a datasheet. If there are that many differences to begin with, you know that there are many unknown factors that cannot be controlled for.
Unmeasured differences: This is the fatal flaw of this and many similar studies: It is not randomized. Without randomization, you don’t balance the compared groups. So, in this case, adults who use hearing aids have many other reasons for their lower rates of death. It is surely NOT hearing aid use.
We call this residual confounding. It is often the main problem with all look-back (retrospective) non-random observational studies. The authors list this as their main limitation.
Our results accounted for all available potential confounders including demographics and medical comorbidities. However, potential confounders and mediators that have implications in understanding the mechanism behind the associations between hearing loss or hearing aid use and mortality, such as depression, social engagement, and cognition were not available in this study.
There are many other limitations, but it’s hardly worth discussing these because this is the fatal flaw.
Another way to write this limitation would go something like this: potential unmeasured confounders preclude any reliable causal inference from this type of study. And acting on such observations has a chance of leading to wasteful research.
I highlight this paper because it exposes three big problems with science and science communication.
A) That this study was done. Surely the researchers know that the methods were too limited to make any conclusions. But they did it anyway. That speaks to the misguided incentives in science. Using the number of publications is a bad measure.
B) That a journal published it. I don’t know about this journal, but I have to believe that the editors, too, know that the methods were broken. Again. Bad incentives. To publish.
C) That this paper was covered uncritically in the healthcare news media. In this case, I don’t know if the journalists actually knew how limited a study this was. I find it hard to believe that an NPR-level journalist would not know about the problems with causal inference from observational studies. Again, bad incentives. Getting clicks.
In sum…Hearing aids are great tools for adults. We don’t need flawed research to promote hearing aids.
The continued publication of flawed studies and news coverage is shredding trust in science and science news coverage.
Sensible Medicine is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber
Our goal at Sensible Medicine is to help educate the public regarding critical and rational thinking. We mean no malice towards the authors, the editors or the journalists. Our goal in criticizing such studies is to foster science education. Thanks for your support. We remain independent and free of advertisements thanks to this support.