Journal Logo

Speech Perception Gap

Speech Perception Gap Part Four: Closing the Gap

McCraney, Anna L. AuD

Author Information
doi: 10.1097/01.HJ.0000719792.45586.79
  • Free

Editor's note: This is the fourth installment of a four-part series. Read part one in our April 2020 issue (, part two in our June 2020 issue (, and part three in our August 2020 issue (

A speech perception (SP) gap is the discrepancy between a patient's cochlear potential for word recognition (WR), namely the PB max, and the actual aided WR (WRaided).1 A significant SP gap is present when a patient's WRaided at 60 dBA is ≥18 points below the PB max (based on the Thornton-Raffin criterion2), indicating a failure to achieve full cochlear potential as defined by the PB max score. It's not surprising that patients with undiagnosed SP gaps are often the same patients who have a history of being difficult, if not impossible, to satisfy. Many of them have tried multiple sets of hearing aids (HAs) over the years only to relegate each pair to the dresser drawer. It is thought that SP gaps occur as the result of a simple anatomical inefficiency, not a deficiency in the HA or its programming, which explains why these patients haven't been satisfied with any hearing aids they've tried. SP gaps affect about a quarter of HA patients with sensorineural hearing loss (SNHL), with both the largest and most prevalent SP gaps affecting those with moderate to severe high-frequency (HF) SNHL.3 Two simple tests are needed to calculate SP gaps: PB max (measured at UCL-5 if testing at a single level) and WRaided at 60 dBA. It can be expressed as SP Gap = PB max WRaided. In the preceding articles in this series, we explored SP gaps, who they affect, why they occur, and how to identify them. In this article, we'll discuss ways clinicians can manage patients for whom HAs have proven ineffective.

Figure 1
Figure 1:
The significant and very strong correlation between PB max and MAXUM WRS (r=0.86, p=0.001) enables the use of PB max to reasonably predict word recognition outcomes with MAXUM (adapted from Laryngoscope. 2018 Feb;128[2]:456-460. doi: 10.1002/lary.26605. Epub 2017 Jun 5). Audiology, hearing loss, hearing aids.
Figure 2
Figure 2:
Comparison of an electromagnetic middle ear implant and hearing aid word recognition performance to word recognition performance obtained with earphones (adapted from Otol Neurotol. 2017 Oct;38(9):1308-1314. doi: 10.1097/MAO.0000000000001554). Audiology, hearing loss, hearing aids.


Historically, hearing loss treatment has been a binary proposition: HAs or cochlear implants (CIs). When a patient is not doing well with HAs, a CI is recommended. While CIs have changed the lives of many patients, the surgery is both invasive and irreversible, demanding thoughtful consideration and significant commitment. To mitigate the associated loss of residual hearing following cochlear implantation, hybrid cochlear implants (HCIs) using combined electro-acoustic stimulation (EAS) were developed to provide adequate stimulation for severely damaged high-frequency cochlear regions while sparing the largely intact low-frequency regions. However, according to the original U.S. FDA clinical trial data, 44 percent of the HCI recipients lost most, if not all, residual hearing during the first year post-implantation.4 In a 2019 study, it was reported that “half of the implanted ears had minimal hearing preservation or total hearing loss (HL) at 5.5 years and the cumulative risk of total HL was 50 percent at seven years.”5 So while EAS may offer improved speech understanding over HAs for many patients, the possibility of losing significant residual hearing remains a legitimate consideration.

A middle ear implant (MEI) is another implantable alternative to CIs and HAs. Research into MEIs began about 30 years ago. Two FDA-approved MEIs are marketed in the United States: Esteem and MAXUM. Esteem is totally implantable while MAXUM is partially implantable, requiring the use of a custom canal-worn processor that is approximately the size of a completely-in-the-canal (CIC) or invisible-in-the-canal (IIC) HA. While Esteem and MAXUM are uniquely different from each other, they both replace the function of the tympanic membrane and the middle ear by directly driving the cochlea through electromechanical energy, not acoustic energy. In other words, they don't make sound. By providing amplification through direct mechanical stimulation of the cochlea rather than acoustic stimulation, MEIs offer significantly better HF gain6 (up to 60 dB)7 and fidelity than HAs with little to no feedback while also posing no risk to residual hearing since their implantation does not disturb the cochlea.


While the unobtrusive hardware, powerful gain, high fidelity, and preserved residual hearing make MEIs an intriguing treatment option, they are particularly compelling for cases in which an SP gap has been identified. Considering that SP gaps occur in ears with an inefficient middle ear transfer function that prevents adequate acoustic energy from reaching the cochlea and that MEIs don't transmit acoustic energy, we can see how MEIs’ direct mechanical stimulation of the cochlea may be particularly advantageous for patients with SP gaps. To test this theory, Dyer, et al., conducted a retrospective analysis of clinical trial data to see if SP gaps existed among patients who received an MEI, specifically the MAXUM.1 As we can see in Figure 1, not only were the patients’ MAXUM word recognition scores (WRS) consistent with their PB max whether they were tested with an analog or digital processor, the scores were also very strongly correlated, which means that we can use one score to predict the other. In other words, we can use a patient's pre-operative PB max to predict the postoperative MAXUM WRS.

While it was an important discovery to learn that PB max scores can predict MAXUM WRS in clinical trial patients, everyday patients are not so neatly homogenous as those patients who participate in a clinical trial. So the question remains: Do pre-operative PB max scores predict postoperative MAXUM WRS in everyday, highly variable patients who represent a typical caseload? To answer that question, we turn to Chang, et al.,6 who reported on the outcomes of the first 12 MAXUM recipients implanted in a clinical practice. The patients’ audiograms were variable, including both flat and steeply sloping losses and PB max scores ranged from 30 percent to 90 percent. WRaided measured with their real ear measurement (REM) verified HAs also varied from zero to 84 percent. One patient was lost to follow-up and dropped from the study. Figure 2 shows that the remaining patients except for “10R” had a significant SP gap. Not only were these patients’ MAXUM WRS 42 percent higher than their WRaided on average, but their MAXUM WRS were also largely in good agreement with their preoperative PB max scores, supporting the finding of Dyer, et al., that the strong correlation between PB max and MAXUM WRS allows clinicians to predict MAXUM outcomes when they know the PB max score.

This is a significant finding. Recall from the study by McRackan and Dubno (see part one of this series: that aided outcomes among patients with the same PB max can vary by over 50 points even when they are fitted with REM-verified HAs.8 Reported CI outcomes are similarly variable and therefore similarly unpredictable.9-11 We've not been able to predict treatment outcomes with any other device beyond vaguely assuring patients that we expect that they'll “do better” as a result of the proposed treatment. While published studies on SP gap have largely been of MAXUM recipients, a recently published study12 shows that the Esteem MEI also closes SP gaps, which is no surprise since, like MAXUM, Esteem directly stimulates the cochlea electromechanically rather than acoustically.

In recent years, audiology has enjoyed the release of treatment modalities that were previously unavailable, making treatment decisions no longer binary. With the introduction of each new device, there is greater overlap in the devices’ indications. As such, we have both the luxury and burden of choice.

Having more choices means clinicians are better able to fit technology to a patient rather than having to make a patient fit a technology. However, it also means we must carefully weigh the potential benefits and risks of each option to ensure that we're recommending the most appropriate treatment for each patient. As discussed in the previous articles in this series, research has shown that roughly 75 percent of patients with up to severe SNHL can reach their full cochlear potential through HA use. For the remaining 25 percent, however, an inefficiency in their ear's middle ear transfer function precludes effective HA use no matter how sophisticated or well-fit the HAs are. These patients have an SP gap. Historically, we've assumed that patients with good word understanding and appropriately fit HAs have good aided benefit. Consequently, we rarely test WRaided unless or until a patient is being evaluated for a CI. If we have not conducted a WRaided test, we're unaware that these patients’ aided benefit is not in line with what their PB max has proved their cochlea is capable of understanding. As a result, these patients are considered the “walking wounded” in audiology. While they might have good PB max scores and the best HAs money can buy, functionally they're no better off than traditional CI candidates. By simply testing the WRaided and comparing the results with the patient's cochlear potential for WR (PB max), we can identify patients who have an SP gap and offer them the choice of an MEI to close the gap while also sparing their residual hearing.


    Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.