Hidden hearing loss (HHL) is a popular topic referring to complaints of hearing difficulty or perceived hearing loss despite having “normal” audiometric thresholds. Within the scientific literature, this term has most recently been used to refer specifically to the reduced amplitude of sound-evoked neural responses that occurs with loss of synapses that connect the inner hair cells (IHCs) to the auditory nerve. In other words, the patient’s audiometric difficulties are hidden behind a normal audiogram. However, perceived hearing loss despite normal audiometric thresholds may be a complaint related to numerous factors. The phenomenon of HHL including its epidemiology, pathophysiology, and clinical implications are under intense study and debate. In this review, we will address these subjects with primary emphasis in adult populations. Recent reviews of the work in animals and implications for patients are available, e.g., Liberman et al (2017). In addition, the review of this topic by Pienkowski (2017) is clinically oriented, and is highly recommended to the interested reader.
Epidemiology of HHL
Epidemiology refers to the study of patterns and causes of health and disease. To establish common epidemiological outcomes such as prevalence and determinants (factors that predict or are associated with outcome of interest), we must operationally define HHL. Here, we will define HHL as perceived hearing loss despite normal audiometric thresholds, i.e., normal hearing. Which raises the question, what is “normal hearing?” This sounds simple enough, but can be quite complex. The majority of epidemiological studies in the literature utilize the four-frequency pure tone average (PTA5124) of 500, 1000, 2000, and 4000 Hz and cutoff for normal PTA at greater than or equal to 20 or 25 dBHL.
Currently, no large population-based study has examined extended high frequencies, i.e., greater than 8000 Hz, although several large data sets have become available. The reason that it is important to consider EHF frequencies is that depending on how you define “normal hearing,” you may over or under estimate the prevalence of HHL. Furthermore, even if an individual has “normal” hearing, that does not mean they have not acquired some threshold shift, as 10 years ago their hearing may have been as much as 15–20 dB better (lower thresholds) even if current thresholds are still within many common definitions of “normal” hearing.
What do we know about HHL epidemiology? Anecdotally, patients presenting with this phenotype to otolaryngology and audiology practices (normal audiometric thresholds and reported hearing difficulty) are not uncommon. Gates et al (1990), analyzed data from the Framingham Heart Study reported 20.2 percent of those who self-reported hearing loss had a pure-tone average less than 26 dB HL (PTA=0.5, 1.0, and 2.0). Hind et al (2011) suggested one to five percent of children and adults under the age of 60 have difficulties understanding speech, particularly, in noisy challenging environments despite normal thresholds (defined as < 20dB HL at 0.5, 1.0, 2.0, 3.0, 4.0, 6.0, and 8.0 kHz). More recently, Tremblay et al (2015) analyzed data from the Beaver Dam Offspring Study (BOSS) and found that 12 percent of participants had normal audiometric thresholds but reported hearing difficulty (defined as < 20dB HL at 0.5, 1.0, 2.0, 3.0, 4.0, 6.0, and 8.0 kHz). Interestingly, the prevalence estimates are consistent with suggested prevalence accounts of auditory neuropathy spectrum disorder (ANSD) in pediatric populations (Berlin et al, 2010).
Hearing difficulty despite normal audiometric thresholds is not a new concept. Numerous names have been suggested (e.g., King-Kopetzky syndrome, obscure auditory dysfunction, etc.). So what causes HHL? Again, this is as complicated as explaining what causes age-related hearing loss (for which age itself may only be a small factor). We have many potential factors that may contribute to HHL, both auditory and non-auditory.
Auditory Peripheral Deficits
Changes in cochlear mechanics (amplification and non-linearity) can be observed prior to changes in audiometric thresholds (for review see Dhar et al, 2012). Animal studies have indicated thresholds may be insensitive to loss of the outer hair cells when up to 20–30 percent of outer hair cells are damaged (Bohne et al, 1992; Davis et al, 2005). In addition, changes in neural integrity can also be observed prior to changes in audiometric thresholds. There can be significant loss of afferent neural function without obvious changes to audiometric threshold (Kujawa et al, 2009; Lobarinas et al, 2013; Schuknecht et al, 1953). In addition, auditory neuropathy spectrum disorder (ANSD) represents a spectrum of pathologies with sites of lesions ranging from inner hair cells to the auditory nerve and displaying compromised neural function (neuropathy to dys-synchrony) despite (at some point) normal cochlear function as measured by otoacoustic emissions and cochlear microphonic (Starr et al, 1996).
Auditory Central Deficits
Central auditory issues can include temporal processing deficits, tinnitus, hyperacusis, etc. Central auditory processing disorder (CAPD) is broadly defined as a deficit in the processing of information that is specific to the auditory modality (Jerger et al, 2000). Central auditory deficits should be considered in patients presenting with HHL complaints. It also makes sense that persons with tinnitus would report hearing difficulty even in the absence of other hearing issues (e.g., speech understanding in noise). The very presence of tinnitus may create an impression that something is wrong with a person’s hearing.
Cognitive function, neuropathy/health issues, head/brain injury, stroke, attention deficit and hyperactivity disorders (e.g., attention deficit/hyperactivity disorder), and medications are all factors that may contribute to HHL complaints.
Noise-Induced Synaptopathy and HHL
Permanent peripheral damage observed after noise exposure previously thought to be benign is the underlying driver of recent discussions about HHL. Data from animals and human temporal bones show that neural degeneration can be observed with minimal loss of hair cells (in addition to the classic observations of neural degeneration accompanied by and thought to result from loss of the hair cell targets). Animals raised in quiet lose a subset of the afferent neural innervation with age, with the primary compromised population of cells being the higher-threshold low-spontaneous-rate neural fibers (Schmiedt et al, 1996). In addition, animals that are exposed to noise that elicits a robust temporary threshold shift (TTS), from which there is a complete recovery of neural thresholds, have compromised suprathreshold neural response (e.g., reduced ABR wave-I amplitude) secondary to the immediate loss of synapses and delayed neuronal loss that occurs later with aging of the damaged cochlea (Kujawa et al, 2006; Kujawa and Liberman, 2009).
However, this synaptopathy is dependent on both the noise dose and the corresponding severity of the TTS (Fernandez et al, 2015) and the noise required to elicit synaptopathy appears exceedingly greater with higher animal species (e.g., mouse 100 dB octave band of noise for two hours and non-human primate 108 dB 50 Hz band of noise for four hours) (Valero et al, 2017). Data from Hickox et al (2017), Fernandez et al (2015), Jensen et al (2015) and Lobarinas et al (2017) are consistent in that TTS of less than 30 dB at 24 hours post noise exposure did not result in synaptopathy, whereas exposures resulting in greater than or equal to 40 dB TTS at 24 hours post noise exposure did result in synaptopathy and wave-I amplitude changes. It has also been shown that synaptopathy has the potential to influence signal-in-noise performance in different listening conditions (Lobarinas et al, 2017). The human evidence is more complicated. Human temporal bone studies do support the existence of synaptic and neural loss with minimal evidence of hair cell loss (Makary et al, 2011; Viana et al, 2015). However, these studies not only lack measures of audiometric threshold and function, but noise history information is also mostly unavailable. Even if noise history were known, a major challenge for studies that rely on histology in the absence of functional data is that intact cells do not always indicate functioning cells.
In other words, an ear that looks normal, as OHCs are present, may not be normal if there is damage to stereocilia that is not revealed in whole-mount processing. In the small number of studies specifically including temporal bones of individuals with known noise exposure history, the analyses have shown either limited loss of neural populations and primary loss of hair cells or damage to both hair cells and neural populations, but not damage to neural populations alone. Though these earlier studies did not examine synaptic elements (Igarashi et al, 1964; McGill et al, 1976). The challenge of mixed pathology in human cochlear tissues has recently been discussed in detail by Hickox et al (2017).
Beyond temporal bone studies, there were several studies in humans examining physiological markers of HHL; these studies included participants with normal thresholds (thresholds of 25 dB HL or better) between 250–8000 Hz. Stamper et al (2015) demonstrated a relationship between reported noise history and ABR wave-I amplitude. However, the relationship was not conserved with variation in recording site (mastoid vs. canal) and was significantly influenced by sex [males tend to have smaller ABR amplitude independent of noise exposure history, but males also tend to report higher noise exposure creating risk for confound] (Stamper and Johnson, 2015). Since then, Prendergast et al (2016), Spankovich et al (2017), Fullbright et al (2017), and Grinn et al (2017) have all failed to replicate the originally reported relationship between noise and ABR wave-I amplitude in other young adult populations exposed to similar patterns of recreational noise.
Liberman et al (2016) examined individuals with higher and lower noise exposure, assessing conventional pure-tone thresholds, but also EHF thresholds, DPOAE amplitude, AP and SP amplitude, and performance on a hearing-in-noise test. They found persons with higher noise exposure had an altered SP/AP ratio, poorer extended high frequency thresholds, and poorer hearing-in-noise. Interestingly, the AP was not significantly different; rather the SP was larger in the higher noise group. Bramhall et al (2017) found evidence of lower wave-I amplitude in veterans and non-veterans with higher noise exposure. Bramhall et al (2015) had considered the possibility that wave-I amplitude may be related to performance on the QuickSIN test, however, relationships were statistically significant only when pure-tone average was in the model.
In summary, data supports the existence of synaptopathy with age and noise exposure in animals and humans. What remain unclear are the risk criteria and timeline for humans and determinant factors that may influence susceptibility.
Clinical Evaluation and Management
How do we evaluate and manage these patients? First, we must determine if the patient truly has normal hearing (how you define normal hearing is important). For the purpose of this writing, if the patient has thresholds equal or below 25 dBHL at 250–8000 Hz without notching (e.g., thresholds at 2000 and 8000 Hz 10 dB better than at frequencies of 3000, 4000 or 6000 Hz), we will call it normal. To get to this point you needed to perform an audiogram, but let’s step back to the case history. What are the complaints? Do they have tinnitus? Do they have sound sensitivity or other abnormal auditory perceptions? A tinnitus patient may report hearing difficulties, even with normal audiometric thresholds, simply due to the fact that they have tinnitus. Are the reported difficulties primarily in noise? Is the patient on medication, or have other co-morbidities that may influence attention, awareness, cognitive function, etc., all of which are expected to contribute to difficulties in challenging listening conditions?
Self-assessment measures of hearing are positively correlated with psychophysical measures of auditory function. Numerous options exist. Self-assessment measures can determine the perceived impact on quality of life and help differentiate complaints. The authors most commonly use the Hearing Handicap Inventory for the Elderly and Adults (HHIE and HHIA) short versions (Newman et al, 1991; Newman et al, 1988) and the Tinnitus and Hearing Survey (Henry et al, 2015).
Let’s move on to diagnostic evaluation. It is plausible that a patient with HHL may have reduced cochlear function, neural function (including pre-synaptic, synaptic, and neural), central auditory processing, and/or non-auditory deficits. A good place to start is with the patient’s complaints.
For example, a common complaint in this patient population is difficulty understanding speech in noisy environments. Speech-in-noise testing (e.g., QuickSIN, word-in-noise test) may be good places to start. This testing may confirm the patient’s perception or performance may be within the normative range. Speech-in-noise test selection should be based on patient ability. Although some of the easier tests may have a ceiling effect, the word-in-noise and QuickSIN tests are considered more challenging tests with the potential to reveal evidence of more subtle issues (Wilson et al, 2007). The WIN may further minimize confounding factors such as memory and attention as it involves repeating single words rather than sentences.
Site of lesion testing may provide some further insight. For differential diagnosis of HHL, you should include middle-ear testing, extended high-frequency threshold tests, otoacoustic emissions (both distortion product and transient and possibly multiple levels), OAE suppression, and auditory-evoked potentials (e.g., neural auditory brainstem response, electrocochleography, complex-ABR, middle and late auditory-evoked potentials).
Possible etiologies to consider or rule out include middle-ear dysfunction or history of chronic middle-ear dysfunction (potential amblyopia see Whitton et al, 2011) subclinical cochlear dysfunction, auditory neuropathy spectrum disorder, central auditory deficits, and tinnitus/sound sensitivity. In addition, medical pathologies such as space-occupying lesions should be excluded.
Test battery considerations in addition to comprehensive audiometric evaluation and speech-in-noise testing with more commonly available equipment and material include the following:
Elevated extended high-frequency thresholds (i.e., > 8.0 kHz): These have been associated with reduced speech understanding in noise (Badri et al, 2011; Liberman et al, 2016).
Otoacoustic emissions (OAEs): These include distortion product and transient-evoked otoacoustic emissions. Numerous studies suggest that OAEs are sensitive to small amounts of OHC loss despite limited changes to thresholds. In particular TEOAEs may be sensitive to subtle changes in cochlear amplification.
- Consider also examining DPOAEs at multiple levels L1/L2=65/55 dBSPL and 55/40 dBSPL. Compare findings to normative data (suggest collection only on subjects with normal thresholds and limited noise history).
Tympanometry and middle-ear muscle reflexes (MEMR): These may have normal admittance with absent or elevated MEMR may be suggestive of neural dysfunction/pathology such as ANSD.
Auditory-evoked potentials: These have been a primary measure in differential diagnosis of sensory versus neural pathology. Absence of the auditory brainstem response (ABR) despite presence of cochlear responses (OAEs and CM) is suggestive of ANSD.
- Comparison of response with condensation and rarefaction stimuli can help differentiate cochlear and neural responses as CM reverses direction with polarity changes, while neural responses do not (Berlin et al, 1998). More subtle effects have been suggested for the compound action potential (CAP)/wave-I amplitude (reduced amplitude) and summating potential/action potential ratio (SP/AP) (larger ratio). However, wave-I amplitude in the literature supporting evidence of compromised neural response are mostly within 1 to 1.5 standard deviations of the normative data (Spankovich et al, 2017) and data examining SP/AP ratio have mainly been driven by changes in the SP (Liberman et al, 2016 and Grinn et al, 2017).
- Complex ABR (brainstem response called the frequency-following response, which is driven by the ability to follow a longer-duration stimuli) or middle and late AEP (e.g., MMN, P300) measures may also provide some insight into dysfunction and are an underutilized tool. There is a wealth of literature and on-line courses on applications of auditory-evoked potentials in diagnosis of central auditory deficits (see Atcherson et al, 2015) for recent review).
Central auditory processing disorder (CAPD): There are variable approaches to the diagnostic CAPD battery, but in general five broad types of measures are included (Weihing et al, 2015).
- Dichotic processing
- Temporal processing
- Monaural low-redundancy
- Binaural interaction
- Spatial processing
Checkout the Handbook of Central Auditory Processing Disorder Volume 1 for more information on CAPD diagnostics (Musiek et al, 2013).
Other cognitive screening: An additional measure to consider is a cognitive screen. Numerous options exists, see Beck et al (2016) for a recent review.
The etiology underlying HHL will likely influence management recommendations. For example, evidence supporting a central auditory processing deficit may prompt auditory training exercises and environmental modifications. In cases where significant noise history is reported, hearing conservation approaches should be discussed in an effort to prevent additional damage. Nonetheless, even if we identify evidence of subtle cochlear dysfunction, synaptic or neural loss, at this time there is no way to regenerate these elements. So, what can we recommend?
- Protect Your Ears: Though the jury is out on risk for noise-induced synaptopathy in humans, noise as a risk factor is a preventable determinant of hearing loss in general. Valero et al (2017) found evidence of synaptopathy in non-human primates (monkeys), but the level of exposure was 108 dB SPL for four hours, not taking into account the narrow-band nature of the noise used in the study, which may theoretically increase risk for damage, this would be a noise dose of over 600 percent according to OSHA regulations and over 10,000 percent with NIOSH recommended exposure guidelines. We may expect humans to be even less susceptible to this damage. No matter, use of hearing protection around loud sounds is always “sound” advice (Dobie et al, 2016).
- Recommend Auditory Training: Depending on the outcomes of the test battery proposed earlier, specific auditory training recommendations may be made. Numerous computer-based options exist and choice of training is dependent on the target population and specific deficits being targeted. Depending on the tool, there is mixed data on the generalization of auditory training, but those incorporating combined auditory-cognitive training (Ferguson et al, 2015) and using stimuli with frequent communication patterns (Tye-Murray et al, 2016) have been suggested to increase real-world translation. See Wiehing et al (2015), Olson (2015), and Musiek and Chermak (2013) for further review.
- Pick Up an Instrument: Musical experience may have a profound influence on auditory skills and speech-in-noise ability. Older normal-hearing musicians have faster brainstem timing and greater representation of speech syllable harmonics compared to age-matched peers. However, passive music listening does not get the job done, rather active engagement is necessary. Check out the work of Nina Kraus and colleagues for more information (great review article Anderson et al, 2013).
- Practice Basic Communication Strategies: We are all familiar with basic communication strategies, i.e., environmental strategies, repair strategies, advocacy, clear speech, etc. We are all familiar because they can help.
- Implement Healthy Living: Eating healthy, exercise, and healthy living (e.g., not smoking) are not likely to resolve current HHL complaints. However, healthy living can potentially reduce the risk of developing hearing loss and tinnitus (Spankovich et al, 2013, 2017), as well as perhaps preventing cognitive decline (Phillips 2017).
- Use Mild Gain Amplification/Remote Microphone: Would a hearing aid help someone with “normal” hearing and speech-in-noise difficulty? The literature on the effectiveness of hearing aids in real-world noisy environments suggests features such as noise reduction and directional microphones may buy a few additional dB of signal to noise (see review by Beck et al, 2016). In addition, there is the potential benefit of reduced listening effort (Wendt et al, 2017) and self-perceived benefit. Nonetheless, the most effective means to improve SNR in a noisy background is use of a remote microphone coupled to a closed ear-level device.
A comprehensive examination is critically important for patients reporting difficulty hearing in noise, or other deficits that might be identified as a “hidden hearing loss.” After a comprehensive evaluation, HHL may not be so hidden. Though site of lesion testing is near and dear to most audiologist, once significant pathology is ruled out, the key concern is how we manage these patients to improve their communication concerns.
Though there is great interest in the specific pathophysiology contributing to HHL complaints, in humans there will likely be numerous factors at play. A large acute loss of synaptic elements after TTS in humans has not been demonstrated to date. Based on the human electrophysiological and psychophysical data to date, correlates of synaptopathy (because we cannot directly measure synaptopathy in vivo) are not highly evident in young adults, suggesting that loss of synaptic and neural elements may take high levels of TTS (as in rodent models), or potentially, repeated exposures over an extended time. If humans require higher levels of TTS or noise dose or repeated exposures over time to develop synapse loss and corresponding functional deficits, there will be a corresponding increase in the risk for concomitant cochlear damage (i.e., hair cell loss).
If pathology is both sensory (affecting hair cells) and neural (affecting synapses or neurons), we have a term for this type of hearing loss (sensorineural), and the question is then rehabilitating both the sensory and neural loss components using hearing aids, noise-reduction processing, and auditory training. Specific site of lesion diagnostics will become even more important as sensory and neural regenerative treatments become clinically feasible. However, until that time we still have tools at our disposal to help these patients.
Christopher Spankovich, AuD, PhD, is an associate professor at the University of Mississippi Medical Center in Jackson, Mississippi and an associate editor at Audiology Today at www.audiology.org.
Colleen Le Prell, PhD, is a professor and head of the audiology program at the University of Texas at Dallas in Dallas, Texas.