Early stages of hearing loss — when hearing is still considered normal — were tied to cognitive decline, an analysis of two cross-sectional studies showed.
Surprisingly, associations between hearing and cognition appeared stronger or equivalent in people with normal hearing than in people with hearing loss, reported Justin Golub, MD, MS, of Columbia University Vagelos College of Physicians and Surgeons in New York City, and co-authors, in JAMA Otolaryngology-Head and Neck Surgery.
Links between hearing loss and cognition have been reported many times in observational studies, but the question here was whether the relationship between hearing and cognition begins before people reach “that kind of arbitrary 25 dB threshold,” Golub noted in a JAMA audio interview. “When you’re studying a biological phenomenon, there’s really no need to label an artificial threshold,” he said. “We wanted to look at the entire [spectrum] of hearing, from perfect to severe loss, without defining any strict cut points.”
To investigate this, the team analyzed data from two U.S. epidemiologic studies: the Hispanic Community Health Study (HCHS) from 2008-2011 and the National Health and Nutrition Examination Study (NHANES) 1999-2000, 2001-2002, and 2011-2012 cycles. The sample included 6,451 people age 50 or older, including 5,190 people from HCHS and 1,261 from NHANES. The average age was about 59, and 59.5% were women.
All participants had hearing assessments with pure-tone audiometry and cognitive testing. Hearing loss was defined as a pure-tone average (PTA) exceeding 25 dB; subclinical hearing loss was a PTA of 1 to 25 dB. Neurocognitive performance was measured in both groups by the Digit Symbol Substitution Test (DSST); score range 0-113, a measure of psychomotor speed and attention. The HCHS cohort also had four other cognitive tests: the Word Frequency Test, Spanish-English Verbal Learning Test (SEVLT) 3 trials, SEVLT recall, and Six-Item Screener. In all cognitive tests, higher scores indicated better cognitive performance.
The researchers used three strategies to analyze data: linear regression; generalized additive model (GAM) regression, a technique to look at nonlinear associations; and separate linear regressions for people with and without hearing loss.
“First we confirmed what people have found before: that as hearing gets worse, cognition gets worse,” Golub said. But using GAM regression, the researchers found that “the relationship actually existed in people who had normal hearing,” Golub said. “Going from zero dB — perfect hearing — to 24 dB — borderline hearing loss — there was a drop in cognition.”
Linear regression showed that, for every 10-dB decrease in hearing, DSST scores decreased by 3.12 (95% CI 2.72-3.52) in the combined cohort. This association held for each cohort individually and across different cognitive tests in the HCHS.
As hearing decreased, DSST scores declined in GAM analysis. Visual patterns were qualitatively similar in HCHS and NHANES in both univariable analysis and multivariable analysis that adjusted for age, sex, education, and cardiovascular disease. The drop in cognitive performance was steeper in people with normal hearing: in the second linear regression model, DSST scores in the combined cohort dropped by 2.28 points (95% CI 1.56-3.00) per 10-dB decrease in people with normal hearing, compared with a 0.97-point (95% CI 0.20-1.75) drop in people with hearing loss.
When GAM regression analysis was repeated in the HCHS group for the four other cognitive tests, associations between hearing and cognition also appeared stronger in people with normal hearing.
Most studies of hearing and cognition have found a dose-dependent response between hearing loss and decreased cognitive function, and these new findings run counter to that, noted Danielle Powell, AuD, of the Cochlear Center for Hearing and Public Health at Johns Hopkins University in Baltimore, and co-authors, in an accompanying editorial.
“We caution the reader about the assertion by Golub and colleagues that an association between hearing loss and cognition begins in those with a pure-tone average greater than 25-dB hearing loss,” the editorialists wrote. “To our knowledge, no previous research has attempted to define the threshold at which hearing may influence cognition: this threshold is unknown. Likewise, the cut point for hearing loss has not always been clearly defined at 25-dB hearing loss and has a complicated history.”
“Before any formalized conclusions regarding this association can be drawn, more work is needed on how hearing that is classically defined as normal is associated with cognitive performance,” Powell and co-authors added. “Because of the surprising nature of these results, speculation for external factors that could lead to the present study’s findings is warranted.”
The analysis has several limitations, Golub and colleagues noted. It is based on cross-sectional studies and does not show causation. Both early declines in hearing and cognitive performance may be tied to common aging processes, and unknown confounders may have influenced results.
Still, the findings raise the question of whether the 25-dB threshold to define adult hearing loss is too high, Golub suggested. “We need to really think about hearing loss as a bigger problem with adults, and we need to treat earlier,” he said.
The research was supported by grants from the National Institutes of Health.
Researchers reported relationships with Cochlear, Advanced Bionics, Auditory Insight, Optinose, Decibel Therapeutics, Abbott, Stryker, Acclarent, 3NT, and vTv Therapeutics.
One of the editorial authors reported receiving support as an employee of the Cochlear Center for Hearing and Public Health from Cochlear Ltd.