About 10% of adults who visited Massachusetts Eye and Ear because of reduced hearing learned that their hearing was clinically normal: of 106,787 patient records, 19,952 (19%) had normal audiograms and 45% of those patients came in complaining of decreased hearing (Parthasarathy et al., eLife 2020). Population studies point the same way: 12.0% of adults with normal thresholds in the Beaver Dam Offspring Study and 15% in a national NHANES sample report hearing difficulty. The leading biological suspect, cochlear synaptopathy, was first shown in mice in 2009, when a noise exposure that left thresholds fully recovered still destroyed more than half of the ribbon synapses in the hardest-hit region. This roundup sits next to our broader age-related hearing loss statistics and tinnitus statistics. We aggregated data from the Journal of Neuroscience, Hearing Research, eLife, Ear and Hearing, the NIH National Institute on Deafness and Other Communication Disorders (NIDCD), the World Health Organization, UK Biobank cohort studies and the other peer-reviewed sources listed in the methodology.
TL;DR
- 19% of 106,787 clinic records had normal audiograms; 45% of those patients complained of reduced hearing (eLife 2020).
- 12.0% of adults with normal thresholds report hearing difficulty (Beaver Dam Offspring Study, Ear and Hearing 2015).
- 15% of 2,015 US adults with normal thresholds self-reported hearing difficulty (NHANES 1999-2002, Hearing Research 2018).
- Noise causing only reversible threshold shifts destroyed more than 50% of cochlear nerve synapses in mice and guinea pigs (Hearing Research 2015).
- Mouse spiral ganglion cell counts near the damaged region fell by about 50% two years after one noise exposure (Journal of Neuroscience 2009).
- 7 of 11 human subjects over 60 had more than 60% loss of auditory nerve axons (Neuroscience 2019).
- Noise history added about 24% extra auditory nerve fiber loss in ears of people aged 50 to 74 (Journal of Neuroscience 2021).
- High-noise young Veterans had EFRs 2.7 to 3.4 dB smaller than non-Veteran controls (Hearing Research 2021).
- 10.6% of adults with normal thresholds have persistent tinnitus (NHANES 1999-2002, Hearing Research 2018).
- As many as 40 million US adults (24%) may have hearing test features suggesting noise-induced hearing loss (NIDCD 2025).
- Over 1 billion young adults are at risk of avoidable hearing loss from unsafe listening (WHO fact sheet 2026).
- Mild and severe speech-in-noise impairment carried dementia hazard ratios of 1.33 and 1.53 (UK Biobank, Alzheimer Disease and Associated Disorders 2026).
1. The Animal Evidence: Thresholds Recover, Synapses Do Not
The whole field rests on one uncomfortable finding: a normal audiogram after noise is not proof of an undamaged ear. In the 2009 mouse study that named the problem, a 2-hour octave-band noise at 100 dB SPL caused a 40 dB threshold elevation that returned to normal within 2 weeks, yet ribbon synapses in the 32 kHz region dropped from about 16 to fewer than 7 per inner hair cell (Kujawa and Liberman, Journal of Neuroscience 2009). The nerve cells themselves died slowly: ganglion cell numbers looked normal at 2 weeks, but near the 32 kHz region they had fallen by about 50% at 2 years.
The damage is selective for high-threshold nerve fibers, which do not matter for detecting quiet tones but are thought to matter for following speech in loud backgrounds. That is why the 2015 review argues primary neural degeneration “remained hidden” for decades (Kujawa and Liberman, Hearing Research 2015). Primate data narrow the species gap: rhesus monkeys needed higher noise levels than rodents to lose hair cells, but still lost synapses after exposures that produced only temporary shifts (Hearing Research 2017).
| Metric | Value | Source |
|---|---|---|
| Noise exposure in the founding mouse study | 8-16 kHz octave band, 100 dB SPL, 2 hours | Kujawa and Liberman, Journal of Neuroscience 2009 |
| Threshold shift at 24 hours, then recovery | 40 dB; back to baseline by 2 weeks, stable 8-16 weeks later | Kujawa and Liberman, Journal of Neuroscience 2009 |
| Ribbon synapses per inner hair cell, 32 kHz region | About 16 before, fewer than 7 at 24 hours | Kujawa and Liberman, Journal of Neuroscience 2009 |
| Ganglion cell loss near 32 kHz, 2 years post-exposure | About 50% | Kujawa and Liberman, Journal of Neuroscience 2009 |
| Synapse loss after reversible threshold shifts (mouse, guinea pig) | More than 50% | Kujawa and Liberman, Hearing Research 2015 |
| Rhesus monkey synapse loss, temporary threshold shift | Mean 12-27% in basal half of cochlea | Hearing Research 2017 (rhesus macaque study) |
| Rhesus monkey synapse loss, permanent threshold shift | Mean 50-75% in basal half of cochlea | Hearing Research 2017 (rhesus macaque study) |
| Mice used to test ABR and EFR as predictors of synapse counts | 57 | Journal of the Association for Research in Otolaryngology 2026 |
Context note: the 2026 mouse study found a rectangular amplitude-modulated EFR at 1000 Hz was the single best predictor of synapse numbers for broad synapse loss, while ABR alone best predicted focal loss (JARO 2026). That matters because these are the same noninvasive measures used in people.
2. Aging Without Noise: Synapses Go First
Noise is not required. In CBA/CaJ mice never exposed to loud sound, ABR wave I amplitude was about 65% lower at 80 weeks of age while the hair-cell summating potential was down less than 5% (Sergeyenko et al., Journal of Neuroscience 2013). Synapse loss started in youth, at 4 weeks, and ran to 144 weeks, appearing throughout the cochlea long before threshold or hair cell changes.
The gap between neural and sensory decline is the key interpretive point. A response that is two-thirds smaller with nearly intact hair cell output is exactly the profile an audiogram cannot see, because audiograms measure the quietest sound detected, not how many nerve channels carry a loud, complex signal. Thresholds in these mice stayed only mildly elevated out to 96 weeks before accelerating.
| Metric | Value | Source |
|---|---|---|
| Normal synapse density, mid-cochlea | 15-20 pairs per inner hair cell | Sergeyenko et al., Journal of Neuroscience 2013 |
| Window of progressive synapse loss | 4 weeks to 144 weeks of age | Sergeyenko et al., Journal of Neuroscience 2013 |
| ABR wave I decrement at 80 weeks | About 65% | Sergeyenko et al., Journal of Neuroscience 2013 |
| Summating potential decrement at 80 weeks | Less than 5% | Sergeyenko et al., Journal of Neuroscience 2013 |
| Spiral ganglion cell loss, oldest group (144 weeks) | About 40% | Sergeyenko et al., Journal of Neuroscience 2013 |
| Rodent meta-analysis on age-related synaptopathy | 3,008 records screened, 12 studies reviewed, 5 meta-analyzed | Neuroscience 2026 (systematic review) |
Context note: the 2026 rodent meta-analysis found a significant reduction in ABR wave I amplitude in aged rodents versus young controls for both click and tone-burst stimuli, and called wave I a sensitive marker of age-related synaptopathy in rodents (Neuroscience 2026).
3. Human Temporal Bones: The Strongest Human Evidence
Living humans cannot have their synapses counted, so post-mortem ears carry the most direct proof. In 20 normal-aging people aged 0 to 89 with no ear disease, 7 of 11 subjects over 60 had lost more than 60% of auditory nerve peripheral axons relative to the youngest subjects, while inner hair cell loss rarely exceeded 15% at any age (Liberman group, Neuroscience 2019). The age-related slope of axon loss outpaced inner hair cell loss by almost 3 to 1, meaning many surviving neurons are simply disconnected.
Noise history makes it worse. A comparison of 52 ears with documented noise exposure against 51 age-matched controls found noise caused an additional 24% loss of auditory nerve fibers in the 50-to-74 age group, and a model using histology alone explained about 36% of variance in word-discrimination scores (Liberman group, Journal of Neuroscience 2021). In other words, two people with the same audiogram can understand words very differently, and the count of surviving nerve channels is part of why. Many of these listeners also carry the threshold losses covered in our hearing loss and headphone statistics.
| Metric | Value | Source |
|---|---|---|
| First confocal synapse study of human ears | 5 normal ears aged 54-89, harvested within 9 hours post-mortem | Viana et al., Hearing Research 2015 |
| Outer hair cell loss, subjects over 60 | Mean 30-40% across 0.25-8 kHz | Liberman group, Neuroscience 2019 |
| Inner hair cell loss across audiometric frequencies | Rarely more than 15%, at any age | Liberman group, Neuroscience 2019 |
| Subjects over 60 with more than 60% axon loss | 7 of 11 | Liberman group, Neuroscience 2019 |
| Extra nerve fiber loss from noise history, ages 50-74 | About 24% | Liberman group, Journal of Neuroscience 2021 |
| Age-related loss per decade, nerve axons vs inner hair cells | 6.3% vs 3.4% | Liberman group, Journal of Neuroscience 2021 |
| Variance in word scores explained by histology (with audiogram) | About 36% (about 60%) | Liberman group, Journal of Neuroscience 2021 |
| Surviving ganglion cells lacking a peripheral axon, 62 aging ears | More than 50% apical, almost 66% basal | Journal of the Association for Research in Otolaryngology 2023 |
Context note: the 2015 human study concluded that synaptopathy and axon degeneration “despite a near-normal hair cell population” may be an important part of presbycusis (Viana et al., Hearing Research 2015). Temporal-bone samples are small and skew old, so they show that the pathology exists in humans, not how common it is in the general population.
4. Normal Audiogram, Real Complaint: How Many People
The epidemiology of hidden hearing loss is really the epidemiology of a mismatch: people who say they struggle and pass the standard test. 12.0% of adults with normal audiometric thresholds in the Beaver Dam Offspring Study (82 of 682) self-reported hearing difficulty, with noise from hobbies and firearms among the risk factors (Tremblay et al., Ear and Hearing 2015). A nationally representative US sample put the figure at 15% of 2,015 adults with normal four-frequency averages (Spankovich et al., Hearing Research 2018).
Clinic data show what this looks like on the ground. At Massachusetts Eye and Ear, about 10% of adults came in because of reduced hearing only to be told their hearing was normal, and the normal-audiogram group was young, predominantly 20 to 50 with a median age of 39 (eLife 2020). Standard categories also fit poorly: in 116,400 records over 24 years, conventional audiogram categories left 46% unclassified (Scientific Reports 2020).
| Metric | Value | Source |
|---|---|---|
| Normal-threshold adults reporting hearing difficulty, Beaver Dam | 12.0% (82 of 682) | Tremblay et al., Ear and Hearing 2015 |
| Normal-threshold adults reporting hearing difficulty, US national | 15% of 2,015 | Spankovich et al., Hearing Research 2018 (NHANES 1999-2002) |
| Clinic records with normal audiograms | 19,952 of 106,787 (19%) | Parthasarathy et al., eLife 2020 |
| Normal-audiogram patients whose main complaint was reduced hearing | 45% (about 10% of all adults seen) | Parthasarathy et al., eLife 2020 |
| Clinic records left unclassified by standard categories | 46% of 116,400 (24 years) | Parthasarathy et al., Scientific Reports 2020 |
| US adults reporting some trouble hearing | About 15% (37.5 million) | NIDCD Quick Statistics |
| UK adults aged 40-69 with significant speech-in-noise impairment | 10.7% of 164,770 | Dawes et al., Ear and Hearing 2014 (UK Biobank) |
Context note: self-reported difficulty with normal thresholds can have many causes, including depression symptoms, vision problems and neuropathic symptoms, all of which Beaver Dam also flagged as associated factors. Not every one of these listeners has synaptopathy. Most recent available data for NHANES normal-threshold complaints: Hearing Research 2018 using 1999-2002 survey years.
5. Noise Exposure in Young Adults and Veterans: Mixed Human Signals
If synaptopathy follows noise in people as it does in mice, young adults with heavy exposure and clean audiograms should show it. Some studies do. College students at high noise risk had a summating-to-action-potential ratio nearly twice that of low-risk peers (0.46 vs 0.26), consistent with selective neural loss, and scored worse on word recognition in noise and with time compression (Liberman et al., PLOS ONE 2016). Both groups scored above 96% in quiet, which is exactly the blind spot a standard word test in a silent booth would miss.
Other large studies do not. A Manchester team tested 126 young adults aged 18 to 36 and found no relation between lifetime noise exposure and ABR wave I amplitude, then found no significant behavioral deficits in 138 participants (Prendergast et al., Hearing Research 2017). The honest reading: synaptopathy may be measurable only at extreme exposures, or current tools are too blunt to see it. Military noise sits at the extreme end, and the population at risk is large, as our hearing protection statistics show.
| Metric | Value | Source |
|---|---|---|
| SP/AP ratio, high-risk vs low-risk students | 0.46 vs 0.26 (22 vs 12 participants) | Liberman et al., PLOS ONE 2016 |
| Young Veterans vs non-Veteran controls tested | 48 Veterans, 31 controls, aged 19-35 | Bramhall et al., Hearing Research 2021 |
| EFR deficit, high-noise Veterans vs controls | 2.7, 2.5 and 3.4 dB smaller (100%, 63%, 40% modulation) | Bramhall et al., Hearing Research 2021 |
| Veterans in high-hazard noise occupations | 23 of 48 | Bramhall et al., Hearing Research 2021 |
| Young adults with no noise-ABR wave I link | 126 tested, aged 18-36 | Prendergast et al., Hearing Research 2017 |
| US adults under 70 with test features suggesting noise damage | At least 10 million (6%), perhaps 40 million (24%) | NIDCD, Noise-Induced Hearing Loss (updated 2025) |
| Teens aged 12-19 with features suggesting noise damage | As many as 17% | NIDCD, Noise-Induced Hearing Loss (updated 2025) |
| Workers with 5+ years of loud workplace noise with bilateral speech-frequency loss | About 18% (vs 5.5% unexposed) | NIDCD Quick Statistics |
Context note: musicians are the other natural test group. A study of 76 early-career musicians and 47 non-musicians found greater ABR wave I/V ratios and more self-reported difficulties in musicians, but a two-year follow-up of 64 musicians and 30 non-musicians found few longitudinal changes (Hearing Research 2024). The 2019 consensus review led by Bramhall called for improved consistency across human studies (Hearing Research 2019).
6. Tinnitus, Hyperacusis and the Diagnostic Gap
Synaptopathy is a candidate explanation for more than speech-in-noise trouble. 10.6% of US adults with normal audiometric thresholds have persistent tinnitus (Spankovich et al., Hearing Research 2018, NHANES 1999-2002), and in young Veterans with tinnitus the EFR was 2.4 dB smaller than in controls, a gap the authors estimated as roughly equal to 21 years of aging (Bramhall et al., Journal of Speech, Language, and Hearing Research 2023). A meta-analysis of 11 studies found significantly reduced ABR wave I amplitude in tinnitus patients with normal thresholds (Frontiers in Neuroscience 2021).
The problem is diagnosis. There is no agreed clinical test, and the reviews keep finding the same short list of candidate measures. A 2025 systematic review of 21 studies named ABR as the most common method, with reduced wave I, reduced electrocochleography summating potential and abnormal middle ear muscle reflex as possible markers (Diagnostics 2025). Even the vocabulary is unsettled: of 49 articles in a 2023 scoping review, 61.2% used the term cochlear synaptopathy, 34.7% used both terms and 4.1% used only hidden hearing loss (CoDAS 2023).
| Metric | Value | Source |
|---|---|---|
| Normal-threshold adults with persistent tinnitus | 10.6% | Spankovich et al., Hearing Research 2018 (NHANES 1999-2002) |
| EFR deficit in young Veterans with tinnitus vs controls | 2.4 dB, about 21 years of aging (97 participants) | Bramhall et al., JSLHR 2023 |
| Tinnitus meta-analysis, normal hearing thresholds | 11 of 489 studies included; wave I significantly reduced | Frontiers in Neuroscience 2021 |
| US adults with tinnitus of 5+ minutes in past year | Roughly 10%, about 25 million | NIDCD Quick Statistics |
| Adults presenting with tinnitus as primary complaint | Approximately 20% | Otology and Neurotology 2020 (audiometric predictors of bothersome tinnitus) |
| Systematic review of hidden hearing loss diagnostics | 21 studies; ABR most common | Diagnostics 2025 |
| Terminology across 49 articles | 61.2% synaptopathy, 34.7% both, 4.1% hidden hearing loss | CoDAS 2023 scoping review |
| Earlier review of detection measures | 15 articles (7 human, 7 animal, 1 both) | Seminars in Hearing 2018 |
Context note: the evidence cuts both ways. A UK study of young adults with tinnitus and normal audiograms found greater lifetime noise exposure but no reduction in ABR wave I and no EFR effect, concluding tinnitus may relate to other effects of noise (Guest et al., Hearing Research 2017). The Liberman 2016 high-risk students also reported heightened reactions to sound consistent with hyperacusis.
7. Why Speech-in-Noise Hearing Matters: Brain, Health and Global Burden
Speech-in-noise performance is not a niche lab measure; it predicts outcomes far outside the ear. In 80,287 UK Biobank participants aged 40 to 69, mild and severe digit-triplet speech-in-noise impairment carried dementia hazard ratios of 1.33 (95% CI 1.13-1.56) and 1.53 (95% CI 1.15-2.02) over nearly 988,000 person-years (Peng et al., Alzheimer Disease and Associated Disorders 2026). A separate analysis of 159,395 participants found a 57% higher Parkinson’s risk for every 10 dB worse speech reception threshold (Readman et al., Parkinsonism and Related Disorders 2025).
Brain measures help explain the variability. In the Massachusetts Eye and Ear multi-talker study of normal-hearing adults, combining pupil-indexed listening effort, a neural fine-structure measure and behavioral FM thresholds explained 78% of the variability in speech intelligibility (eLife 2020). A 2026 study of 105 clinically normal-hearing adults aged 18 to 77 found pronounced age-related cochlear neural degeneration alongside cortical changes that were not significantly correlated with it, suggesting brain aging runs in parallel (Journal of Neuroscience 2026). For anyone who spends the day on calls in noisy rooms, the same physics drives the background-noise numbers in our noise pollution statistics.
| Metric | Value | Source |
|---|---|---|
| Dementia hazard ratio, mild / severe speech-in-noise impairment | 1.33 / 1.53 (80,287 participants, 1,086 cases) | Peng et al., Alzheimer Disease and Associated Disorders 2026 |
| Parkinson’s risk per 10 dB worse speech reception threshold | +57% (HR 1.57; 159,395 participants, 810 cases) | Readman et al., Parkinsonism and Related Disorders 2025 |
| Variance in multi-talker speech scores explained, normal hearers | 78% (pupil + neural + FM thresholds) | Parthasarathy et al., eLife 2020 |
| Normal-hearing adults showing age-related cochlear neural decline | 105 tested, aged 18-77 | Journal of Neuroscience 2026 |
| Hearing aid use, UK adults 40-69 | 2.0% | Dawes et al., Ear and Hearing 2014 (UK Biobank) |
| People projected to have some hearing loss by 2050 | Nearly 2.5 billion | WHO fact sheet, March 2026 |
| Young adults at risk from unsafe listening | Over 1 billion | WHO fact sheet, March 2026 |
| Annual global cost of unaddressed hearing loss | Almost 1 trillion US dollars | WHO fact sheet, March 2026 |
Context note: the UK Biobank studies measure speech-in-noise hearing, not synaptopathy, and they are observational, so they show association, not cause. WHO reports that over 5% of the world’s population, about 430 million people, need rehabilitation for disabling hearing loss (WHO, Deafness and hearing loss fact sheet).
Summary: Hidden Hearing Loss and Speech-in-Noise by the Numbers
| Metric | Value | Source |
|---|---|---|
| Clinic adults told hearing was normal despite complaint | About 10% | Parthasarathy et al., eLife 2020 |
| Clinic records with normal audiograms | 19% of 106,787 | Parthasarathy et al., eLife 2020 |
| Normal-threshold adults with hearing difficulty, Beaver Dam | 12.0% | Tremblay et al., Ear and Hearing 2015 |
| Normal-threshold adults with hearing difficulty, NHANES | 15% | Spankovich et al., Hearing Research 2018 |
| Normal-threshold adults with persistent tinnitus | 10.6% | Spankovich et al., Hearing Research 2018 |
| Threshold shift that fully recovered in founding mouse study | 40 dB | Kujawa and Liberman, Journal of Neuroscience 2009 |
| Synapse loss after reversible shifts, rodents | More than 50% | Kujawa and Liberman, Hearing Research 2015 |
| Rhesus synapse loss after temporary shift | 12-27% basal cochlea | Hearing Research 2017 |
| Mouse ABR wave I drop at 80 weeks, no noise | About 65% | Sergeyenko et al., Journal of Neuroscience 2013 |
| Humans over 60 with more than 60% axon loss | 7 of 11 | Liberman group, Neuroscience 2019 |
| Extra nerve fiber loss from noise, ages 50-74 | About 24% | Liberman group, Journal of Neuroscience 2021 |
| Nerve axon vs inner hair cell loss per decade | 6.3% vs 3.4% | Liberman group, Journal of Neuroscience 2021 |
| SP/AP ratio, high- vs low-noise-risk students | 0.46 vs 0.26 | Liberman et al., PLOS ONE 2016 |
| EFR deficit, high-noise young Veterans | 2.5-3.4 dB | Bramhall et al., Hearing Research 2021 |
| EFR deficit in Veterans with tinnitus | About 21 years of aging | Bramhall et al., JSLHR 2023 |
| US adults with features of noise-induced loss | 6% to 24% | NIDCD 2025 |
| UK adults 40-69 with significant speech-in-noise impairment | 10.7% | Dawes et al., Ear and Hearing 2014 |
| Dementia hazard ratio, severe speech-in-noise impairment | 1.53 | Peng et al., 2026 |
| Parkinson’s risk per 10 dB worse SRT | +57% | Readman et al., 2025 |
| Young adults at risk from unsafe listening | Over 1 billion | WHO 2026 |
Methodology and Sources
Every figure above was read during research for this article in a peer-reviewed paper (full text on PubMed Central or the abstract indexed by Europe PMC), an NIH NIDCD page or a WHO fact sheet. Statistics that appeared only in secondary blogs, or that a paper cited from another study we could not read, were excluded. This is a factual summary of published research, not medical advice.
- Kujawa and Liberman: Adding insult to injury, Journal of Neuroscience 2009; Synaptopathy in the noise-exposed and aging cochlea, Hearing Research 2015
- Sergeyenko et al.: Age-related cochlear synaptopathy, Journal of Neuroscience 2013
- Noise-induced cochlear synaptopathy in rhesus monkeys, Hearing Research 2017; Predicting cochlear synaptopathy in mice, JARO 2026; Rodent ABR systematic review and meta-analysis, Neuroscience 2026
- Human temporal bones: Viana et al., Hearing Research 2015; Neuroscience 2019; Journal of Neuroscience 2021; JARO 2023
- Normal-audiogram complaints: Tremblay et al., Ear and Hearing 2015; Spankovich et al., Hearing Research 2018; Parthasarathy et al., eLife 2020; Parthasarathy et al., Scientific Reports 2020; Scientific Reports 2022, word recognition and cochlear nerve degeneration
- Noise-exposed young adults: Liberman et al., PLOS ONE 2016; Bramhall et al., Hearing Research 2021; Prendergast et al., Hearing Research 2017 (I) and (II); Musicians longitudinal study, Hearing Research 2024; Bramhall et al. consensus review, Hearing Research 2019
- Tinnitus and diagnostics: Bramhall et al., JSLHR 2023; Guest et al., Hearing Research 2017; Frontiers in Neuroscience 2021 meta-analysis; Otology and Neurotology 2020; Diagnostics 2025; CoDAS 2023; Seminars in Hearing 2018
- Health outcomes and burden: Peng et al., Alzheimer Disease and Associated Disorders 2026; Readman et al., Parkinsonism and Related Disorders 2025; Dawes et al., Ear and Hearing 2014; Journal of Neuroscience 2026
- NIH NIDCD: Quick Statistics About Hearing; Noise-Induced Hearing Loss (updated April 16, 2025)
- World Health Organization: Deafness and hearing loss fact sheet (March 2026)
- Data watch: there is no population prevalence figure for cochlear synaptopathy itself, because it cannot be measured directly in living people; the 10% to 15% figures describe hearing complaints with normal audiograms, which have several possible causes. Human evidence conflicts: Liberman 2016 and Bramhall 2021 report group differences consistent with synaptopathy, while Prendergast 2017 (126 and 138 young adults) and Guest 2017 found none; the 2019 Bramhall-led consensus review attributes much of this to inconsistent methods. Several key figures are older than three years and remain the most recent available data: NHANES 1999-2002 (published 2018), Beaver Dam (2015), UK Biobank baseline speech-in-noise prevalence (Dawes 2014) and the animal studies (2009-2017). Temporal-bone samples are small (5 to 103 ears). The Neuroscience 2026 rodent meta-analysis and the 2026 UK Biobank dementia paper were read as indexed abstracts only. The 2024 Lancet Commission on dementia could not be read (403 on the publisher site), so its hearing-loss figures are not used here. Sample-level details such as the 21-year aging equivalent are authors’ model estimates, not direct measurements.
Last updated: October 3, 2026. We update this roundup quarterly, and the next refresh is expected when new NHANES audiometry cycles and further UK Biobank speech-in-noise follow-up analyses are published.