Skip to content

An Estimate for Hidden Hearing Loss

In This Article

  • In a recent study, researchers at Mass Eye and Ear examined data from nearly 96,000 ears to develop a word-score model that can help estimate the amount of hidden hearing loss in humans
  • Led by Stéphane F. Maison, PhD, CCC-A, researchers collected thousands of recorded word recognition scores among patients who underwent hearing evaluations at Mass Eye and Ear from 1993 to 2017
  • The scores were then compared to what they should be based on patients' hearing thresholds
  • Findings from the study confirmed an association between poorer speech scores and larger amounts of cochlear nerve damage
  • The study was one of the largest retrospective studies of its kind, and its findings could one day help clinicians determine who is a candidate for regenerative therapies, as well as traditional and newer sound amplification products

This article was written by Mike Kotsopoulos and republished from the Fall 2022 Harvard Otolaryngology Magazine

More than 1.5 billion people worldwide live with some degree of hearing loss, according to the World Health Organization (WHO). The severity of these cases varies from mild to profound, and the vast majority are attributed to sensorineural hearing loss, which affects the ability of the inner ear to transmit sound signals to the brain.

If left untreated, hearing loss can affect the social-emotional wellbeing of patients; the constant struggle to communicate with peers can lead to overwhelming feelings of frustration and isolation. While traditional hearing devices can help treat sensorineural hearing loss by amplifying sounds entering the ear, these devices are not perfect. Many patients hear the amplified sound but report great difficulties understanding what the sound means. Noisy environments only exacerbate the issue; the cluster of sounds heard in a restaurant, for example, might blend into an indiscernible din, making it impossible for someone to follow a conversation.

The discovery of a new type of inner-ear damage at Mass Eye and Ear in 2009 has helped make sense of this phenomenon. Referred to as cochlear synaptopathy, or hidden hearing loss, the nerve damage affects how clearly sounds are heard instead of the inability to hear sounds at all.

"This might explain why so many patients with normal hearing exams report difficulties understanding a conversation in noisy environments," said Stéphane F. Maison, PhD, CCC-A, associate professor of Otolaryngology–Head and Neck Surgery at Harvard Medical School and a principal investigator of the Eaton-Peabody Laboratories at Mass Eye and Ear. "Likewise, it explains why most hearing aid users still struggle with intelligibility of speech."

Identifying how much hidden hearing loss exists in a living human could one day help clinicians determine who is a candidate for regenerative therapies, as well as traditional and newer sound amplification products. However, the only method to date for quantifying the amount of nerve damage is through harvesting the temporal bones of cadavers.

Inside the Eaton-Peabody Laboratories, a team of researchers led by Dr. Maison made an important first step toward clearing this hurdle by developing the first-ever word-score model capable of predicting the extent of hidden hearing loss in living human ears. According to a study in Scientific Reports, the research team built its model using the records of nearly 96,000 ears examined at Mass Eye and Ear, making it one of the largest retrospective studies of its kind.

"Prior to this study, we could either estimate neural loss in a living patient using a lengthy test battery or measure cochlear nerve damage by removing their temporal bones when they've died," said Dr. Maison. "Using ordinary speech scores from hearing tests—the same ones collected in clinics all over the world—we can now estimate the number of neural fibers that are missing in a person's ear."

Figure 1

Stéphane F. Maison, PhD, CCC-A, evaluating hearing scores inside his laboratory. Image courtesy of Dr. Maison.

Uncovering Hidden Hearing Loss

Two main factors determine how well a person can hear: audibility and intelligibility. Hair cells, the sensory cells inside the inner ear, contribute to the audibility of sounds—or how loud a sound must be to be detected. Upon receiving a sound, hair cells pass electrical signals to the cochlear nerve, which then passes those signals to the brain. How well the cochlear nerve relays these signals contributes to the intelligibility, or clarity, of sound processed within the central nervous system.

For years, scientists and clinicians focused on hair cell deterioration as the primary cause of hearing loss and believed cochlear nerve damage was widespread only after hair cells were destroyed. In 2009, Sharon Kujawa, PhD, and M. Charles Liberman, PhD, principal investigators in the Eaton-Peabody Laboratories, upended this notion when they discovered permanent cochlear nerve damage existing in the ears of mice that had recovered hair cells and hearing thresholds after noise exposure.

Their finding called into question whether the audiogram—long considered the gold standard of hearing exams—had actually measured the full extent of damage to the ear. Audiograms, they found, had only provided information about the health of hair cells and not information about the cochlear nerve. Because scientists believed nerve loss was secondary to hair cell loss or dysfunction, patients with a normal audiogram were typically given a clean bill of health despite reporting difficulties hearing in noisy environments.

"We used to think that if thresholds recovered to normal after noise, or had not yet elevated with aging, then there was no lasting ear damage, which isn't necessarily the case," said Dr. Kujawa. "Rather, the neural loss is 'hidden' in the normal audiogram."

In 2017, the National Institutes of Health (NIH) awarded Drs. Kujawa, Liberman, Maison, and Daniel Polley, PhD, of the Eaton-Peabody Laboratories, a five-year $12.5 million P50 grant to investigate the new type of hearing damage from multiple angles. Through four separate projects, including one studying donated human temporal bones, one focusing on animal models of human hearing loss, and two working with human subjects, the researchers have since found widespread evidence of nerve damage hidden behind normal audiogram measurements.

Building a Predictive Model

Mass Eye and Ear possesses one of the largest collections of audiometric data in the world. In the early 1990s, Aaron Thornton, PhD, CCC-SLP/A, former director of Audiology at Mass Eye and Ear, developed the Audiology Operating System (AOS), a computer-controlled audiometer interface that incorporated a patient-specific model of word understanding. By 1993, the AOS evolved into Mass Eye and Ear's Audiology Database, which has collected the speech recognition scores and audiograms of patients for nearly two decades.

"Aaron was a visionary who knew how to think outside the box," Dr. Maison said. "Thanks to the database he established, we were able to retrieve a tremendous number of hearing evaluations that few other research hospitals could ever obtain."

Dr. Maison's team began building its predictive model for hidden hearing loss by taking a deep dive into the database. The researchers collected thousands of recorded word recognition scores among patients who underwent hearing evaluations at Mass Eye and Ear from 1993 to 2017. They compared them to what they should be based on patients' hearing thresholds. Since word recognition scores are obtained at a presentation level well above a patient's hearing threshold—where audibility is not an issue—any difference between the predicted and the measured score should have reflected deficits in intelligibility, Dr. Maison explained.

After considering a number of factors, including the cognitive deficits that may accompany aging, Dr. Maison argued that the size of these discrepancies reflected the amount of hidden hearing loss a person had. He then collaborated with Dr. Liberman who, through his NIH-funded hidden hearing loss research project, had obtained measurements of neural loss from the existing pathological data of human temporal bones. By applying the observed word-score discrepancies to the neural loss measurements provided by Dr. Liberman, Dr. Maison and his team created a predictive model based on a standard hearing exam.

Findings from the study published in Scientific Reports confirmed an association between poorer speech scores and larger amounts of cochlear nerve damage. Among the word scores of older patients, the researchers observed major deficits in intelligibility. These deficits were even worse in patients with Ménière's disease and other sensorineural etiologies consistent with temporal bone studies showing a dramatic loss of cochlear nerve fibers.

Assessing the Road Ahead

By 2030, the WHO predicts a staggering increase in hearing loss among the global population. The organization predicts the number of people affected by the condition will jump to nearly 2.5 billion, 700 million of whom will require hearing rehabilitation.

To ensure clinicians can better estimate the effectiveness of hearing loss interventions in future patients, Dr. Maison would like to further refine his team's model. He will begin by modifying current hearing evaluation protocols and implementing tests of word recognition performance in noise. A refined model, he believes, could one day help clinicians pinpoint optimal treatments in the most timely manner. In theory, it could serve as a tool to determine who may benefit from regenerative therapies and perhaps predict the efficacy of hearing aids versus cochlear implants.

In the meantime, separate research on hidden hearing loss will continue to flourish at Mass Eye and Ear. In August 2022, the NIH renewed the initial five-year $12.5 million P50 project awarded to Drs. Kujawa, Liberman, Maison and Polley. Its renewal will allow the investigators to extend their projects on hidden hearing loss to human subjects with different forms of clinical hearing loss, as well as to subjects who report tinnitus, a persistent "ringing" in the ears, and hyperacusis, a hypersensitivity to sound. Both conditions can be debilitating for patients, and neither has a known cause or cure.

According to Dr. Kujawa, the evaluation of human and animal subjects with overt hearing loss under the new P50 grant will bring investigators a step closer to devising sensitive diagnostic tools, thereby making it possible to apply future therapeutics to prevent, limit or reverse cochlear neural degeneration.

"Right now, we have therapeutics on the horizon that have shown the potential to repair or reconnect these nerve fibers in the inner ear," said Dr. Kujawa, who will oversee the administrative portion of the grant. "If we can improve our tests so they show us who has neural degeneration and who doesn't, then we can test the efficacy of these therapeutics, which would bring us closer to a long-sought treatment for hearing loss."

Learn more about the Department of Otolaryngology–Head and Neck Surgery at Mass Eye and Ear

Refer a patient to Mass Eye and Ear/Mass General Brigham

Related topics


Matthew G. Crowson, MD, MPA, MASc, Michael S. Cohen, MD, Christopher J. Hartnick, MD, and colleagues trained a neural network to diagnose pediatric middle ear effusion using a novel approach: the training set of images were tympanic membrane photos taken during myringotomy. When applied to a test set, the algorithm's diagnostic accuracy was 84%.


Steven D. Rauch, MD, and colleagues at Mass Eye and Ear found word recognition scores improve after cochlear implantation for deafness caused by Ménière's disease regardless of age at implantation and whether patients received previous medical or surgical Ménière's treatment.