Audiologists have Been Testing Hearing the Wrong Way!

We've Been Testing Hearing the Wrong Way! - YouTube

If, like me, you've reached seniority, and found the need to seek help to hear, you may have been dissappointed at the performance of prescribed hearing aids given you by very expensive audiologists in stressful situations such as cocktail parties and restaurant dinners. This video, and the article below may explain why. As is said in the military, train like you're going to fight.

Changing the Way We Test Hearing Forever 

For decades, audiologists have been testing hearing the wrong way, according to a new study published in the Journal of Ear and Hearing. The traditional hearing tests, such as pure tone testing and word recognition in quiet, may not accurately reflect an individual's ability to understand speech in noisy environments. In fact, research shows that even individuals with good word recognition scores in quiet may still struggle in background noise. The QuickSIN test, which simulates real-life scenarios, provides a more accurate assessment of an individual's ability to understand speech in noise. This shift in approach can lead to better diagnosis and treatment of hearing loss, and audiologists can benefit from incorporating this new approach into their practice.

QuickSIN v. WRQ Testing Article Summary

This study aimed to evaluate the feasibility of replacing word-recognition testing in quiet (WRQ) with speech-in-noise (SIN) testing using the QuickSIN in routine audiologic practice. The authors analyzed data from 5,808 patients who underwent audiometry, WRQ testing, and QuickSIN testing at the Stanford Ear Institute.

Key findings:

1. QuickSIN SNR loss increased and WRQ scores decreased with greater degrees of hearing loss, but QuickSIN was abnormal at milder hearing losses compared to WRQ.

2. Many patients with normal or near-normal hearing had abnormal QuickSIN performance despite having excellent WRQ scores.

3. Using logistic regression, the authors could predict which patients were likely to have good/excellent WRQ scores (≥88% or ≥76%) based on their high-frequency pure-tone average (HFPTA) and QuickSIN SNR loss with high accuracy.

4. However, WRQ scores and HFPTA were poor predictors of abnormal QuickSIN performance.

The authors conclude that these findings support replacing WRQ with QuickSIN testing in most patients during routine audiologic evaluations. They provide potential guidelines, such as only conducting WRQ testing if HFPTA is ≥40 dB HL and QuickSIN SNR loss is ≥7-8 dB. This approach would make audiologic testing more aligned with patient complaints of difficulty hearing in noise while saving time by avoiding WRQ testing in most patients predicted to have good/excellent scores. The findings have important implications for updating routine audiologic practice.

ncbi.nlm.nih.gov

Preliminary Guidelines for Replacing Word-Recognition in Quiet With Speech in Noise Assessment in the Routine Audiologic Test Battery

78–99 minutes

---

• Journal List
• Lippincott Open Access
• PMC10583951

2023 Nov-Dec; 44(6): 1548–1561.

Abstract

Objectives:

For decades, monosyllabic word-recognition in quiet (WRQ) has been the default test of speech recognition in routine audiologic assessment. The continued use of WRQ scores is noteworthy in part because difficulties understanding speech in noise (SIN) is perhaps the most common complaint of individuals with hearing loss. The easiest way to integrate SIN measures into routine clinical practice would be for SIN to replace WRQ assessment as the primary test of speech perception. To facilitate this goal, we predicted classifications of WRQ scores from the QuickSIN signal to noise ratio (SNR) loss and hearing thresholds.

Design:

We examined data from 5808 patients who underwent audiometric assessment at the Stanford Ear Institute. All individuals completed pure-tone audiometry, and speech assessment consisting of monaural WRQ, and monaural QuickSIN. We then performed multiple-logistic regression to determine whether classification of WRQ scores could be predicted from pure-tone thresholds and QuickSIN SNR losses.

Results:

Many patients displayed significant challenges on the QuickSIN despite having excellent WRQ scores. Performance on both measures decreased with hearing loss. However, decrements in performance were observed with less hearing loss for the QuickSIN than for WRQ. Most important, we demonstrate that classification of good or excellent word-recognition scores in quiet can be predicted with high accuracy by the high-frequency pure-tone average and the QuickSIN SNR loss.

Conclusions:

Taken together, these data suggest that SIN measures provide more information than WRQ. More important, the predictive power of our model suggests that SIN can replace WRQ in most instances, by providing guidelines as to when performance in quiet is likely to be excellent and does not need to be measured. Making this subtle, but profound shift to clinical practice would enable routine audiometric testing to be more sensitive to patient concerns, and may benefit both clinicians and researchers.

Keywords: Audiology, Speech in noise, Word recognition, Hearing loss

INTRODUCTION

Approximately 14.3% of Americans 12 years of age or older have some degree of bilateral hearing loss, defined as elevated audiometric thresholds, with 22.7% of Americans having hearing loss in at least one ear (Goman & Lin 2016). The widespread nature of such deficits is of crucial importance, because hearing loss can directly impede the ability of an individual to communicate with others. Hearing loss is linked to increasing rates of isolation and depression in adults (Arlinger 2003; Dawes et al. 2015 ), and is associated with higher risk of cognitive decline (Lin 2011; Lin et al. 2011). Hearing loss is routinely diagnosed through audiometric testing by audiologists who specialize in the diagnosis and management of hearing disorders. Despite the numerous advances in the management and treatment of hearing loss in recent decades, much of the routine audiologic test battery in use today was developed over 50 years ago (Carhart & Jerger 1959; Peterson & Lehiste 1962; Tillman & Carhart 1966). Standard audiometric assessment consists primarily of pure-tone audiometry, measurement of tympanometry and acoustic reflexes, and some measure of speech understanding ability. Since the inception of audiology over 60 years ago, the default measure of speech understanding has been word-recognition in quiet (WRQ) (Egan 1948; Hirsh et al. 1952; Peterson & Lehiste 1962; Tillman & Carhart 1966).

WRQ has a long and storied history in audiology. Generally it involves the repetition of phonetically or phonemically balanced monosyllabic words (Egan 1948; Hirsh et al. 1952; Peterson & Lehiste 1962; Tillman & Carhart 1966). The goal of such measures has been to find the best possible performance for a given individual (PBmax). Excellent WRQ scores are generally considered to mark someone as a good candidate for amplification in presence of elevated pure-tone thresholds. The assumption underlying this recommendation is that hearing aids are designed to make speech sounds audible; if a patient can understand speech when it is maximally audible, then they should benefit from hearing aid use. To that effect, PBmax has been characterized by some scientists as the “information carrying capacity” of the auditory system (Halpin & Rauch 2009). WRQ scores have also been used to help identify the presence of retrocochlear disorders such as vestibular schwannomas, as poorer than expected WRQ scores for a given degree of hearing loss may be a sign of retrocochlear pathology. Such disorders are more likely when “rollover” is observed, meaning that WRQ scores decrease with increasing intensity of the speech signal (Jerger & Jerger 1971; Dirks et al. 1977).

Although pure-tone audiometry and acoustic immittance measurements remain useful diagnostic procedures, there is increasing awareness that WRQ scores may contribute little to the diagnosis and rehabilitation of patients with hearing loss. For example, in modern clinical practice, WRQ scores play a small role in the detection of retrocochlear pathology, as magnetic resonance imaging (MRI) has become the gold standard for detection of vestibular schwannomas and another retrocochlear disorders.

With regard to rehabilitative audiology, WRQ scores often provide very little information which can be used by the audiologist beyond a general deficit in understanding speech. For example, speech recognition in quiet has no relationship with hearing aid satisfaction or continued device usage (Humes 2003; Killion & Gudmundsen 2005). In contrast, speech in noise (SIN) abilities appear to better correlate with hearing aid satisfaction (Walden & Walden 2004; Saunders & Forsline 2006; Davidson et al. 2021). Further underscoring the importance of speech understanding in noise versus quiet, it has been known for decades that speech understanding in quiet has little bearing on real-world communication abilities (Davis 1948; High et al. 1964; Giolas et al. 1979). Rather, the primary concern of most patients is an inability to understand speech in the presence of background noise (Le Prell & Clavier 2017). Consistent with this view, when questionnaires are used to assess the perceived communication difficulties of patients, those questions focus predominately on understanding SIN, rather than in quiet (Newman et al. 1990, 1991, 1993; Cox & Alexander 1995, 2002; Gatehouse & Noble 2004; Noble et al. 2013).

Given the recurrent complaints of individuals with hearing loss, it is somewhat remarkable that SIN testing in audiology has not entered routine clinical use. The concept of doing so; however, has been discussed for decades with only minimal changes to clinical practice. For example, in 1966 the Northwestern University Test No. 6 (NU-6; Tillman and Carhart). Word lists were published (Tillman & Carhart 1966), and these lists continue to be widely used in clinical practice today. In 1970, these same authors recommended SIN testing also be conducted as part of standard clinical practice (Carhart & Tillman 1970), yet that recommendation has never been consistently adopted outside of preoperative and postoperative assessment of performance by recipients of cochlear implants. The continued lack of SIN testing in clinical audiology has persisted despite numerous researchers having developed several tools for clinical use such as the hearing in noise test (HINT, Nilsson et al. 1994), the words in noise (WIN) test (Wilson 2003), and the QuickSIN (Killion et al. 2004). The WIN or the QuickSIN were shown to be more sensitive to the effects of hearing loss than the HINT or the Bamford-Kowal-Bench (BKB)-SIN (Wilson et al. 2007), in large part because of the differences between low-context (QuickSIN) versus high-context (BKB-SIN or HINT) sentence materials, or the lack of context in the WIN. For these reasons, Wilson et al. (2007) recommended either the WIN or the QuickSIN for clinical use. Further supporting the idea that the WIN was appropriate for clinical practice, the WIN was measured in over 3400 veterans (Wilson 2011). In this study, Wilson noted several key findings. Most important among them are that both WRQ and WIN scores decreased with increasing age and degree of hearing loss, but performance on WRQ could not predict the WIN score. The inability to predict SIN abilities from speech recognition in quiet indicates that SIN abilities in patients need to be measured directly, rather than inferred from traditional audiometric measures (Wilson 2011; Vermiglio et al. 2018).

Taken together, there is a substantial and decades-long disconnect between the primary concerns of patients (e.g., challenges understanding SIN), and the routine measurement of WRQ in standard audiologic practice. This raises the question of why WRQ continues to be ubiquitous in clinical use. Wilson et al. (2007) identified several key factors that have hindered the widespread adoption of SIN measures into routine audiologic assessment. One is the longstanding tradition of monosyllabic word-recognition in clinical practice. Audiologists have performed such measures for decades, and both audiologists and physicians routinely expect to see a word-recognition score as part of the standard audiometric evaluation.

A second key factor that may hinder widespread adoption of SIN measures is that most tests of SIN abilities provide output in terms of signal to noise ratio (SNR) required to obtain 50% correct recognition, rather than a measure of percent correct. SNR-based measures have the potential to more accurately characterize patient performance because they avoid floor or ceiling effects which can occur with fixed SNR levels that are too easy or challenging for a given patient. Such flexibility is likely to be important for routine clinical practice in which patients will present with a wide array of hearing thresholds and auditory pathologies. Unfortunately, audiologists, physicians and other healthcare professionals lack guidance as to how to evaluate performance in terms of the SNR.

A third and vital factor is that it is unclear how the information from SIN measures can be utilized in clinical practice. For example, clinicians need to know how SIN is influenced by auditory pathology, whether retrocochlear in nature (Qian et al. 2023) or resulting from outer or middle ear disorders (Smith et al. in revision). Clinicians also need to know how SIN abilities relate to perceived auditory disability (Fitzgerald et al. in revision), and the age of the patient (Fitzgerald and Ward in revision). Clinicians would also benefit from a better understanding of how SIN performance influences candidacy for hearing aids, effectiveness of varying signal processing, and subsequent satisfaction with amplification. Finally, a better understanding of SIN performance is required for pediatric patients or other vulnerable populations. Taken together, these issues speak for the need for additional research on SIN performance and how it can be used to guide clinical practice for audiologists and physicians.

A final, and crucial factor limiting the implementation of SIN measures into routine audiologic practice is that most audiologists view performing SIN measures as an addition to the routine audiologic test battery. By this logic, measuring both SIN and WRQ scores is often perceived to be too time-consuming for traditional audiologic practice, particularly in healthcare systems that prioritize volume of patients seen in their metrics of clinical efficiency.

One approach to alleviate many of these concerns, particularly regarding the duration of testing, would be to make SIN measures the default test of speech perception in routine audiologic testing. In this scenario, SIN measures would replace WRQ in most instances, rather than performing both measures simultaneously. Making this subtle, but profound transition to clinical practice would solve the pragmatic concerns of clinicians having insufficient time to perform both measures. Perhaps more important, it would make routine audiologic assessment more sensitive to the concerns elicited by patients as to how hearing loss affects their everyday life. To make this transition; however, requires additional information about the relationship between hearing loss, WRQ, and speech understanding in noise. As noted previously, SIN abilities cannot be reliably predicted from WRQ scores or degree of hearing loss (Wilson 2011; Vermiglio et al. 2018). However, if WRQ performance could be predicted by SIN abilities or degree of hearing loss, then it provides a path by which SIN measures can replace WRQ in the routine audiologic test battery. Our assumption underlying this approach was that audiologists and physicians need to know when WRQ scores are likely to be excellent or poor, as those classifications are likely to drive decisions regarding patient care (e.g., hearing aid candidacy, attempt to preserve hearing during surgery, and so on). In contrast, small differences in WRQ scores obtained with 25-word lists (e.g., 4%) are generally not relevant for patient care. If SIN abilities and/or hearing thresholds can predict whether a patient is likely to have excellent WRQ scores, then WRQ would not need to be measured in these patients, as it would be highly unlikely to add anything to the diagnostic or rehabilitative management of that patient. In contrast, an algorithm that can predict suboptimal WRQ scores could then provide guidelines as to when WRQ may provide relevant information for patient management, and therefore should be measured. For example, patients with poor word-recognition scores may be candidates for cochlear implantation (Zwolan et al. 2020) and thus could be assessed when appropriate.

Here, we address these issues by reporting data from 5808 patients who underwent monaural QuickSIN testing in addition to traditional measures of WRQ and pure-tone audiometry. Our first goal of this investigation was to characterize QuickSIN performance in routine clinical practice across a wide range of hearing abilities and auditory pathology. Such information is necessary for clinicians seeking to integrate SIN measurement in routine clinical practice. Our second goal was to use the QuickSIN SNR loss and high-frequency pure-tone average (HFPTA) to predict whether a given patient is likely to have good to excellent WRQ scores; we used HFPTA because that has been reported to align better with speech understanding in noise than the traditional PTA (Wilson 2011). Using the nomenclature from Bossuyt et al. (2003), our reference test was WRQ. The index test was the QuickSIN, audiometric threshold measurement, or both. Finally, the target condition was the WRQ score. Our final goal was to use these predictions to generate clinical guidelines for making SIN measures the default test of speech perception in routine audiometric testing, with guidelines for when WRQ scores should be measured.

MATERIALS AND METHODS

Participants

All participants were patients undergoing audiometric evaluations at the Stanford Ear Institute. Clinical data from 5808 individuals were included in this study. All participants were either native speakers of English, or highly fluent. Their ages ranged from 18 to 101 years of age; children under the age of 18 were excluded from this study to avoid developmental effects observed with performance on some tests of SIN (Holder et al. 2016). Many of these patients were seen in conjunction with a team of otologists and neurotologists. Therefore, these data include patients with a wide array of auditory pathologies including presbycusis, various middle ear pathologies, and retrocochlear pathology such as vestibular schwannomas. Women comprised 49.99% of the participants. According to U.S. Census Bureau (2010), the ethnic demographic breakdown in the San Francisco Bay area is 23.3% Asian, 23.5% Hispanic or Latino, 6.7% African American, 0.7% Native American, 0.6% Native Hawaiian or Pacific Islander, and 52.5% Caucasian. Taken together, these data suggest that this is a diverse patient base with regard to auditory pathology, sex, and ethnicity.

Procedures

All data were obtained as part of routine clinical audiologic evaluations at the Stanford Ear Institute. These evaluations consisted of the traditional audiologic test battery (otoscopy, tympanometry and acoustic reflex measurements, air-conduction and bone-conduction thresholds, speech-reception threshold, and WRQ). Air-conduction and bone-conduction thresholds were obtained using the modified Hughson-Westlake method (Carhart & Jerger 1959). Interoctave thresholds at 3000 and 6000 Hz were regularly obtained; other interoctave thresholds were measured when the thresholds differed by ≥20 dB HL between neighboring octaves. Some individuals also underwent additional testing, such as otoacoustic emissions, vestibular assessment, and auditory brainstem responses. Those data are not considered here. All audiologic testing was completed in double-walled sound booths using GSI-61 (Grayson-Stadler, Eden Prairie, MN, USA) audiometers and fed through either ER-3A insert earphones or circumaural headphones (Sennheiser HD 200, Sennheiser USA, Old Lyme CT, USA).

We obtained WRQ scores using NU-6 lists (Tillman & Carhart 1966), and SIN abilities were obtained from the QuickSIN (Killion et al. 2004). WRQ scores were obtained unilaterally in each ear. In most cases, 25-words were used to obtained WRQ (Auditec NU-6 Form A). In some instances, the difficulty-weighted words were used (version II, Hurley & Sells 2003), and if a patient scored either 90% or 100% across the first 10 words, then word-recognition was discontinued, and the percent correct value was reported. Regardless of the number of presentations, we computed the percentage of words correctly repeated by the patient and reported that as their WRQ score. Following measurement of WRQ, the QuickSIN was obtained unilaterally in each ear. The QuickSIN is a test which measures the SNR at which 50% of key words in low-context sentences can be repeated in the presence of multitalker babble. This test consists of six low-context sentences, with each sentence containing five key words. Each sentence is presented at a different SNR, beginning at +25 dB, and decreasing in 5 dB steps to 0 dB SNR (Killion et al. 2004). Two lists were presented in each condition. In each condition, the QuickSIN SNR loss was the average SNR loss of those two lists. Note that unlike the WRQ scores, the QuickSIN SNR losses reflect the difference in SNR relative to norms to repeat 50% of key words in a sentence (Killion et al. 2004)

To minimize effects related to presentation level, we used the same level for both WRQ and speech in noise measurement in each patient. The default presentation level was 70 dB HL unless that level would have resulted in some part of the signal being inaudible. In that case, the audiologist increased the signal presentation level to maximize audibility while not exceeding the uncomfortable loudness level of the patient. In this way, we attempted to observe the best possible performance for a given individual (PBmax for WRQ measures). In all patients we used recorded stimuli for the speech material.

Data Analysis

Our approach for replacing WRQ with SIN in routine clinical environments is to use audiometric data to predict when WRQ scores are likely to be optimal (in which case it need not be conducted), or suboptimal (in which case it may have diagnostic and rehabilitative significance). To complete this goal, our first step toward was to perform a paired-sample t test on the QuickSIN data from the right and left ears to determine whether there was any “right-ear advantage” for SIN abilities; this was previously observed in a large-scale clinical study with the WIN (Wilson 2011). Here, we did not observe a significant difference in the QuickSIN SNR loss between right and left ears t(_11,614) = 1.408, p = 0.16. Thus, further statistical analyses were reported on data from the right ear only.

We then attempted to predict two classifications of WRQ scores as measured by the NU-6. The first classification was WRQ scores ≥88% correct, defined here as “excellent.” We selected this value because, for a 25-word list, scores between 88% and 100% do not differ statistically (Carney & Schlauch 2007). The second classification to be predicted was WRQ scores ≥76%. We defined this classification as “good.” We chose this value of 76% because it is often used anecdotally by audiologists to separate “good” from “fair” word-recognition abilities (Lawson and Peterson 2011). For both the 88% and 76% cutoffs, we built a statistical model using logistic regression to determine the sensitivity and specificity of five different criteria. These criteria were prespecified according to values readily interpretable by audiologists, and are listed as follows:

1. HFPTA ≥ 40 dB HL

2. QuickSIN SNR loss ≥ 7 dB

3. QuickSIN SNR loss ≥ 8 dB

4. QuickSIN SNR loss ≥ 7 dB and HFPTA ≥ 40 dB HL

5. QuickSIN SNR loss ≥ 8 dB and HFPTA ≥ 40 dB HL

The HFPTA was calculated by averaging audiometric thresholds at 1 kHz, 2 kHz, and 4 kHz. A HFPTA of 40 was chosen as one criterion because it reflects the transition between mild and moderate hearing loss. A similar rationale was chosen for the QuickSIN SNR losses. According to the QuickSIN manual (Etymotic Research 2006), an SNR loss of 7 dB reflects a “moderate SNR deficit.” However, the data used to generate these criteria were presented binaurally in many instances (Killion et al. 2004). Within our own data we observed an average improvement of 1 dB between binaural and monaural presentation in a smaller subset of patients who completed both unilateral and bilateral assessments. Thus, one could argue that a 7 dB SNR loss obtained binaurally may be considered to be equivalent in performance to an 8 dB SNR loss obtained monaurally. To account for either interpretation, we utilized both criteria when building our statistical model. In this way, we determined the sensitivity and specificity of the index test (QuickSIN, HFPTA, or both) using five different cutoff points. Logistic regressions were also computed to predict patients with QuickSIN SNR loss ≥7 dB using HFPTA and WRQ as predictors. Finally, in addition to our logistic regression analyses, we also computed linear regressions and second-order polynomial functions to describe the relationship between two variables (i.e., HFPTA versus WRQ or HFPTA versus SNR loss).

All data, statistical analyses, and code for plotting of figures for this project are publicly available through the Stanford data repository.

RESULTS

Effects of Hearing Loss on Speech Understanding in Quiet and Noise

On average, hearing loss is associated with decreases in WRQ scores, but this decrease in part reflects increasing variability in WRQ scores with decreases in hearing acuity. Table ​1 displays mean, median, and SD values, as well as the range of WRQ scores, for a given degree of hearing loss for the right ear. These data are shown in Figure ​1 with HFPTA as a continuous variable, and in Figure ​2 with different degrees of hearing loss widely used in audiologic practice as determined by the HFPTA (1 kHz, 2 kHz, and 4 kHz). Results of a one-way analysis of variance reveal a significant decrease in WRQ scores as a function of degree of hearing loss (F_{5, 5803} = 1294, p < 0.001). Post hoc Holm-Sidek tests reveal that WRQ scores did not differ for HFPTA <15 and 16 to 25 dB HL (p = 0.251); all other WRQ scores differed significantly from one another (p < 0.001). The results of a linear regression on WRQ scores as a function of HFPTA was also significant (F_{1, 5807} = 3750, p < 0.001; R² = 0.39). However, these data were better fit by a second-order polynomial (R² = 0.65).

TABLE 1.

Word-recognition scores as a function of the degree of hearing loss

Right Ear HFPTA	Mean WRQ % Correct	SD	Median WRQ % Correct	WRQ Range
Normal (<15 dB HL)	98.96	2.90	100	84–100
Normal (16–25 dB HL)	98.58	3.30	100	80–100
Mild (26–40 dB HL)	96.34	6.45	100	0–100
Moderate (41–55 dB HL)	90.59	10.55	92	24–100
Moderately severe (56–70 dB HL)	73.51	24.11	80	0–100
Severe to profound (≥71 dB HL)	45.22	31.52	40	0–100

Word-recognition scores (y axis) as a function of HFPTA values (x axis) for 5808 patients. Here, different classification of HFPTA hearing loss are represented by different colors. Each circle reflects an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. HFPTA, high-frequency pure-tone average.

Word-recognition scores (y axis) as a function of HFPTA values (x axis) for 5808 patients. Here, different classification of HFPTA hearing loss are represented in different columns. Each circle represents an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. For a given degree of hearing loss, the mean word-recognition score is depicted by the solid black line. HFPTA, high-frequency pure-tone average.

Although there is a significant decrement in WRQ scores as a function of hearing loss, it should be noted that WRQ scores remain almost universally excellent for individuals with normal hearing through a moderate loss, as the mean WRQ score remained over 90%. In contrast, WRQ scores deteriorate rapidly for moderately severe losses or greater (Table ​1 and Fig. ​2). and the range of observed WRQ scores increased with the degree of hearing loss. For example, with moderately severe or severe hearing losses, some individuals continued to display excellent WRQ scores, whereas others had extremely poor degrees of performance. In other words, the range of WRQ scores are clustered near excellent when hearing is largely normal, and progressively increase in variance with increasing hearing loss.

As with WRQ, increasing amounts of hearing loss also results in decrements in QuickSIN SNR losses, but QuickSIN SNR losses have a greater likelihood of being abnormal, even in the presence of normal hearing. Table 2 shows the mean, median, SD and range of QuickSIN SNR losses for a given degree of hearing loss. Figure 3 shows the QuickSIN SNR losses as a function of HFPTA for right ears, whereas Figure ​4 shows QuickSIN SNR losses categorized according to differing degrees of hearing loss. There are several points to note in these data.

QuickSIN SNR loss (y axis) as a function of HFPTA values (x axis) for 5808 patients. Here, different classification of HFPTA hearing loss are represented in different columns. Each circle represents an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. For a given degree of hearing loss, the mean word-recognition score is depicted by the solid black line. HFPTA, high-frequency pure-tone average; QuickSIN SNR, QuickSIN signal to noise ratio; SNR, signal to noise ratio.

First, QuickSIN SNR losses decrease significantly as a function of hearing loss (F_{5, 5803} = 1192, p < 0.001), with significant decrements observed for each successive change in HFPTA (p < 0.001 in each case). Results of a linear regression on WRQ scores as a function of HFPTA were also significant (F_{1, 5807} = 5847, p < 0.001; R² = 0.50). Second, many individuals (23.8%) with HFPTA <15 dB HL had QuickSIN SNR losses >3 dB suggesting abnormal speech in noise abilities despite having normal hearing. If we assume that all QuickSIN criteria for different degrees of SNR loss should be shifted by ~1 dB to account for monaural versus binaural differences in performance, 13.2% of individuals with HFPTA <15 dB HL showed QuickSIN SNR losses that may be interpreted as abnormal. The number of individuals with elevated QuickSIN SNR losses were even higher for individuals with HPFTA values ranging from 16 to 25 dB HL, as 44% had QuickSIN SNR losses >3 dB SNR, and 30% had QuickSIN SNR losses >4 dB SNR.

A second key point from these data are that, for a given degree of hearing loss, both the average performance was worse, and the variance of that performance was greater, for the QuickSIN SNR losses than the traditional WRQ scores. Here, the QuickSIN SNR losses outlined in Table ​2 and shown in Figures ​3 and ​4 demonstrate that, on average, performance deteriorated with increasing degrees of hearing loss. Mean QuickSIN SNR losses were abnormal with mild HFPTA values and continued to deteriorate with additional hearing loss. Moreover, there was considerable variance in QuickSIN SNR loss across all degrees of hearing acuity, even in patients with normal hearing. In contrast, WRQ scores were on average excellent (>88%) through moderate HFPTA values, and only became abnormal on average for moderately severe losses. Consistent with the difference in the rate of decline on these two tests, the slope of the regression lines for QuickSIN SNR losses and WRQ scores significantly differed (F_{1, 11,604} = 6735, p < 0.001).

TABLE 2.

QuickSIN SNR loss as a function of the degree of hearing loss

Right Ear HFPTA	Mean QuickSIN SNR Loss	SD	Median QuickSIN SNR Loss	QuickSIN SNR Loss Range
Normal (≤15 dB HL)	2.16	2.06	2.06	−3.5 to 13.5
Normal (16–25 dB HL)	3.14	2.46	2.46	−3.5 to 16.5
Mild (26–40 dB HL)	5.09	3.45	3.45	−3 to 25.5
Moderate (41–55 dB HL)	8.21	4.61	4.61	−3.5 to 23.5
Moderately severe (56–70 dB HL)	12.80	6.40	6.40	−2.5 to 25.5
Severe to profound (≥71 dB HL)	17.05	7.17	7.17	−0.5 to 25.5

QuickSIN SNR loss (y axis) as a function of HFPTA values (x axis) for 5808 patients. Here, different classification of HFPTA hearing loss are represented by different colors. Each circle reflects an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. HFPTA, high-frequency pure-tone average; QuickSIN SNR loss, QuickSIN signal to noise ratio loss; SNR, signal to noise ratio.

Relationship Between Performance in Quiet and Noise

Taken together, these data provide evidence that clinical measurement of monosyllabic word-recognition ability in quiet has at best a moderate relationship with a clinical measurement of the ability to understand low-context sentences in noise. Figure ​5 depicts right-ear WRQ scores and QuickSIN SNR losses for 5808 patients who completed both measures. This figure illustrates that WRQ scores are good or excellent in most patients. Notably, 90.6% of patients had WRQ scores from their right ear that could be interpreted as excellent (≥88%), whereas 94.6% could be interpreted as good or better (≥76%). The high percentage of patients with excellent or good WRQ scores is notable given that these patients were seen at a tertiary medical center because of suspected or confirmed audiovestibular pathology, and suggests that performance on this test is at ceiling in most patients.

QuickSIN SNR loss (y axis) as a function of word-recognition score (x axis) for 5808 patients. Each circle represents an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. QuickSIN SNR, QuickSIN signal to noise ratio; SNR, signal to noise ratio.

Despite most patients having excellent or good WRQ scores, a significant percentage demonstrated difficulties understanding SIN (see Fig. ​1). Here, 45.8% of patients with WRQ scores ≥88% had SNR losses >3 dB in their right ear, suggesting a mild or greater deficit in performance according to the QuickSIN manual (Etymotic Research 2006). Of these patients 66.8% of individuals had QuickSIN SNR losses ranging from 3 to 6.5 dB, while 31.6% had SNR losses ranging from 7 to 14.5 dB. Finally, 1.6% of patients with excellent WRQ scores had right ear QuickSIN SNR losses ≥15 dB. Thus, in hundreds of cases, individuals had SNR losses suggesting a moderate or severe difficulty understanding SIN (Etymotic Research 2006), despite possessing good to excellent WRQ scores. Finally, a significant linear regression of QuickSIN SNR loss on WRQ scores was observed (p < 0.001; R² = 0.53), consistent, although these data were better fit by a second-degree polynomial function (R² = 0.66) These regression analyses suggest that individuals with poor WRQ scores almost universally have poor QuickSIN SNR losses, whereas individuals with excellent WRQ have a wide range of QuickSIN SNR losses.

Predicting Classifications of WRQ Scores

Although our data show considerable variance in QuickSIN SNR losses, HFPTA values, and WRQ scores, we can predict with high accuracy whether an individual is likely to have a WRQ score that is excellent or good. Figure ​6 shows a scatterplot of QuickSIN and HFPTA values for patients with WRQ scores either below (left panel) or above (right panel) 88% correct. Figure ​7 shows similar scatterplots for WRQ scores ranging from 76% to 100%, or below 75% correct. In both figures, individuals with WRQ scores below those cutoff points largely had HFPTA values ≥40 dB HL, and QuickSIN SNR losses ≥7 dB SNR. As described in section Data Analysis, logistic regression analyses were performed on different audiometric criteria to determine whether they could predict whether an individual had a WRQ score below the specified values of 88% and 76%. Table ​3 shows the results of this multiple-logistic regression analyses for the 88% criteria; Table ​4 shows the results for the 76% criteria. Both tables show the odds ratio,¹ the 95% confidence interval, and the p value for each set of QuickSIN and HFPTA values. From these statistical analyses, we then generated the area under the Receiver Operating Characteristic curve (AUC), the sensitivity and the specificity. These data are shown in Tables ​5 and ​6 for the 88% and 76% criteria, respectively. Because results were virtually identical for the right and left ear, only the right ear analyses are shown here.

TABLE 3.

Odds ratio, 95% confidence interval, and p value for various audiometric criteria at predicting classification of WRQ scores with 88% as a cutoff value

Audiometric Criteria	Odds Ratio	95% Confidence Interval	p
HFPTA ≥ 40 and SNR ≥ 7	52.03	41.44–65.32	<0.001
HFPTA ≥ 40 and SNR ≥ 8	61.54	48.05–78.81	<0.001
HFPTA ≥ 40	57.16	42.99–76.00	<0.001
SNR ≥ 7	30.71	21.78–43.29	<0.001
SNR ≥ 8	19.65	14.98–25.78	<0.001

TABLE 4.

Odds ratio, 95% confidence interval, and p value for various audiometric criteria at predicting classification of WRQ scores with 76% as a cutoff value

Audiometric Criteria	Odds Ratio	95% Confidence Interval	p
HFPTA ≥ 40 and SNR ≥ 7	105.13	69.21–159.70	<0.001
HFPTA ≥ 40 and SNR ≥ 8	110.21	70.69–171.84	<0.001
HFPTA ≥ 40	103.85	61.81–174.49	<0.001
SNR ≥ 7	84.98	40.19–179.67	<0.001
SNR ≥ 8	66.47	35.43–124.73	<0.001

TABLE 5.

Sensitivity, specificity, and area under the curve for various audiometric criteria at predicting classifications of WRQ scores with 88% as a cutoff value

Audiometric Criteria	AUC	Sensitivity (%)	Specificity (%)
HFPTA ≥ 40 and SNR ≥ 7	0.88	89.35	86.12
HFPTA ≥ 40 and SNR ≥ 8	0.89	88.43	88.95
HFPTA ≥ 40	0.87	82.56	92.35
SNR ≥ 7	0.78	61.56	95.04
SNR ≥ 8	0.78	64.19	91.64

TABLE 6.

Sensitivity, specificity, and area under the curve for various audiometric criteria at predicting classifications of WRQ scores with 76% as a cutoff value

Audiometric Criteria	AUC	Sensitivity (%)	Specificity (%)
HFPTA ≥ 40 and SNR ≥ 7	0.90	86.58	94.22
HFPTA ≥ 40 and SNR ≥ 8	0.90	85.45	94.94
HFPTA ≥ 40	0.88	79.57	96.39
SNR ≥ 7	0.79	59.32	98.31
SNR ≥ 8	0.80	62.14	97.59

Both the left and right panels depict QuickSIN SNR losses (y axis) as a function of HFPTA (x axis). The left panel displays individuals with WRQ scores <88% correct; the right panel shows data for individuals with WRQ scores ranging from 88% to 100%. Reference lines with darker shades represent the boundaries between mild and moderate performance on the QuickSIN, and mild vs. moderate HFPTA hearing losses. Each circle represents an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. HFPTA, high-frequency pure-tone average; QuickSIN SNR, QuickSIN signal to noise ratio; SNR, signal to noise ratio; WRQ, word-recognition in quiet.

Both the left and right panels depict QuickSIN SNR losses (y axis) as a function of HFPTA (x axis). The left panel displays individuals with WRQ scores <76% correct, whereas the right panel shows individuals with WRQ scores ranging between 76% and 100%. HFPTA, high-frequency pure-tone average; QuickSIN SNR, QuickSIN signal to noise ratio; SNR, signal to noise ratio; WRQ, word-recognition in quiet.

The first key finding from these analyses is that the logistic-regression model for QuickSIN and HFPTA values tested here were all significant (p < 0.001 in each case). Perhaps more important, the sensitivity, specificity, and AUC values were all quite high for this analysis. This suggests that audiometric values, whether from HFPTA or QuickSIN, can be used to identify patients with good to excellent WRQ scores with a high degree of accuracy.

It is worth noting that the relative effectiveness of each model varied when attempting to predict WRQ scores that were not excellent (e.g., <88%). Here, the confidence intervals for models built only on QuickSIN SNR losses ≥7 or ≥8 dB fell outside of those resulting from either a combination of SNR and HFPTA, or HFPTA alone. Consistent with these analyses, the AUC was lower when basing the statistical model on the QuickSIN losses alone (~0.75) relative to models which either combine QuickSIN SNR and HFPTA, or examine HFPTA alone (AUC ranged from 0.88 to 0.89). This decrease in effectiveness is likely because of the poor specificity observed when fitting a model to the QuickSIN SNR losses alone (~54% to 57%) as opposed to parameter values that include the HFPTA (all specificity values ranged between 80% and 87%). In contrast, the sensitivity for identifying WRQ scores <88% was largely similar across all values tested (ranging from 90% to 96%). Similar results were observed when attempting to predict WRQ scores <76%. Once again, the AUC was lower (0.76 to 0.77 for QuickSIN SNR alone versus 0.87 to 0.9 when including HFPTA in the statistical model). Sensitivity values were similar across all models (ranging from 95% to 99%), whereas specificity values were lower when examining the QuickSIN alone (53% to 55%) as opposed to models including the HFPTA in some capacity (ranging from 77% to 84%).

Another key finding is that WRQ scores are almost universally excellent in patients with lower HFPTA values, and with smaller QuickSIN SNR losses. For example, 98.9% patients with HFPTA values <40 dB HL, had WRQ scores ≥88%, and 99.7% had WRQ scores ≥76%. Similarly, 99.2% of patients with QuickSIN losses <7 dB SNR had WRQ scores ≥88%, whereas 99.9% had WRQ scores ≥76%. Finally, 99.8% of patients with both HFPTA <40 and QuickSIN <7 had WRQ scores >88%. Taken together, these results suggest a high likelihood of excellent WRQ scores in listeners with hearing thresholds ranging from normal to a mild loss.

Predicting Classifications of QuickSIN SNR Loss

Finally, we attempted to predict which patients had a moderate QuickSIN SNR loss from the HFPTA and WRQ scores. In the previous section, we demonstrated that HFPTA and QuickSIN SNR loss can predict patients with good or excellent WRQ scores. If WRQ scores cannot predict which classifications of QuickSIN SNR loss, then then this provides further evidence that measures such as the QuickSIN provide information beyond what WRQ can provide in clinical practice, and provides additional support for the idea that SIN could replace WRQ in routine audiologic assessment.

Figure ​8 shows scatterplots from the right ear with QuickSIN SNR losses either above or below 7 dB. Results of the logistic regression analyses are shown in Tables ​7 and ​8. The logistic-regression model for the ability of WRQ scores and HFPTA values to predict QuickSIN SNR losses >7 were all significant (p < 0.001 in each case). However, the predictive power of these models were poor, and were considerably worse than the ability of QuickSIN SNR loss and HFPTA to predict classifications of WRQ scores. Table ​8 shows the AUC values, sensitivity, and specificity for these audiometric variables for predicting QuickSIN SNR losses >7 (e.g., significant SIN deficit). Note that all of these variables have very high sensitivity (90% to 99%), but extremely poor specificity (13% to 44%). The AUC value was 0.67 with HFPTA >40 as a cutoff value, whereas AUC values of AUC values of 0.6 and 0.57 were observed with WRQ cutoffs of 88% and 76% correct, respectively. AUC values of 0.5 are at chance level. AUC values between 0.7 and 0.8 are considered acceptable. AUC values of 0.8 to 0.9 are excellent, whereas values >0.9 can be viewed as outstanding (Hosmer & Lemeshow 2000). Thus, the predictive power of WRQ to identify a significant QuickSIN SNR loss is very poor despite the statistical significance of the model. In contrast, QuickSIN SNR losses, particularly when used in conjunction with HFPTA, provides excellent or outstanding predictive power for identifying good versus suboptimal word-recognition abilities in quiet.

TABLE 7.

Odds ratio, 95% confidence interval, and p value for various audiometric criteria at predicting categories of QuickSIN SNR losses with 7 dB as a cutoff value

Audiometric Criteria	Odds Ratio	95% Confidence Interval	p
HFPTA ≥ 40	6.93	6.12–7.85	<0.001
WRQ ≥ 88	30.71	21.78–43.29	<0.001
WRQ ≥ 76	84.98	40.19–179.67	<0.001

TABLE 8.

Sensitivity, specificity, and area under the curve (AUC) for various audiometric criteria at predicting categories of QuickSIN SNR losses with 7 dB as a cutoff value

Audiometric Criteria	AUC	Sensitivity (%)	Specificity (%)
HFPTA ≥ 40	0.67	89.95	43.64
WRQ ≥ 88	0.60	99.12	21.33
WRQ ≥ 76	0.56	99.82	12.97

Here, both the left and right panels depict word-recognition scores (y axis) as a function of HFPTA (x axis). The left panel displays individuals with QuickSIN SNR losses ≥7 dB. The right panel shows data for individuals with QuickSIN SNR losses <7. The black reference lines depict the boundary between mild and moderate hearing loss (x axis) and between excellent (top y axis line) and good (bottom y axis line) word-recognition scores. Each circle represents an individual data point. To minimize the effects of overplotting, darker shading reflects values with more data points, whereas lighter shading reflects values with fewer data. HFPTA, high-frequency pure-tone average; QuickSIN SNR, QuickSIN signal to noise ratio; SNR, signal to noise ratio.

DISCUSSION

For over 70 years, the default test of speech perception in routine audiologic testing has been the repetition of monosyllabic words in quiet (Egan 1948; Hirsh et al. 1952; Peterson & Lehiste 1962; Tillman & Carhart 1966). This has occurred in spite of widespread acknowledgment that difficulties understanding SIN are the primary complaint of individuals with hearing loss. Here, we provide QuickSIN data from over 5800 individuals, and describe how they relate to the degree of hearing loss, and traditional WRQ scores. Consistent with previous reports, we observed only a modest relationship between a clinical measure of the ability to repeat monosyllabic words in quiet, and a clinical measure of the ability to understand speech in the presence of background noise. Performance on the QuickSIN decreased with hearing loss, albeit with considerable variability with all degrees of hearing loss. Most important, we provide strong evidence for the idea patients with excellent or good WRQ scores can be accurately identified by the HFPTA thresholds and QuickSIN SNR losses. In contrast, WRQ and HFPTA are poor predictors of a significant SNR loss on the QuickSIN. To our knowledge, these data reflect the first large-scale study of QuickSIN SNR losses obtained in a clinical population. Our ability to therefore make these predictions in this diverse patient base allows for new clinical guidelines for speech recognition testing which enable SIN to be the default test of speech perception in routine audiometric testing. Taken together, these data suggest that integrating SIN measures into routine clinical practice and replacing WRQ measures in most patients is feasible, and would provide additional information relative to traditional WRQ scores, because WRQ scores are a poor predictor of SIN abilities.

QuickSIN and WRQ Performance With Varying Degrees of Hearing Loss

One noteworthy aspect of these data is that QuickSIN SNR losses tended to deteriorate far more rapidly than WRQ scores with virtually all degrees of hearing loss. For example, although WRQ scores deteriorated significantly with increasing hearing loss, WRQ scores remained excellent on average for individuals with normal hearing through moderate degrees of hearing loss according to the HFPTA. These data suggest that most patients have excellent WRQ scores when the signal is audible until significant hearing loss is observed. In contrast, QuickSIN SNR losses deteriorated far more rapidly, and with greater variability in performance for virtually all degrees of hearing ability. Numerous individuals displayed abnormal QuickSIN SNR losses even with HFPTA values <15 dB HL, and the median QuickSIN SNR loss was indicative of a mild SNR deficit with mild HFPTA values (see Table ​2). Performance continued to deteriorate on the QuickSIN with increasing hearing loss, albeit with considerable variance in performance. Taken together, these data are consistent with prior work showing that deficits in SIN abilities can be observed in individuals with little to no difficulty understanding speech in quiet (Saunders & Haggard 1989; Middelweerd et al. 1990; Vermiglio et al. 2018).

Our findings in this clinical population are largely consistent with previous research examining the influence of hearing acuity on QuickSIN SNR losses. For example, in individuals with normal hearing, our mean QuickSIN loss of ~2.0 dB SNR (see Tables ​1 and ​2) was similar to that of several previous investigations using smaller sample sizes (Killion et al. 2004; Duncan & Aarts 2006; Phatak et al. 2018). In contrast, a separate investigation revealed mean QuickSIN SNR losses of ~4 dB SNR for individuals with normal hearing (Wilson et al. 2007). Less data are available for comparison in individuals with varying degrees of hearing loss, and there appears to be greater variability in reported outcomes relative to the data in individuals with normal hearing. For example, one recent report indicated values of approximately 5 dB SNR for individuals with a moderate high-frequency loss (Phatak et al. 2018). In contrast, other investigations report QuickSIN SNR losses of approximately 12 dB with moderate high-frequency losses (McArdle et al. 2005; Wilson et al. 2007). Here, we observed mean QuickSIN SNR losses of 5.09 dB SNR for individuals with a mild HFPTA, 8.21 dB SNR loss for moderate HFPTA, 12.80 dB SNR loss for moderately severe HFPTA, and 17.05 dB SNR loss for individuals with severe HFPTA values (see Tables ​3 and ​4).

We speculate that the variance in SNR losses between our investigation and others reflect two key factors. First, there are small methodological differences across several of these studies which may influence the QuickSIN SNR losses. For example, the presentation level varied somewhat across different investigations, and some investigations used only a prespecified subset of QuickSIN lists to better control differences across participant groups (McArdle et al. 2005; Wilson et al. 2007). Another key methodological difference is how different hearing losses were defined. Here, we utilized HFPTA as a metric for classifying degree of hearing loss, which differed from other investigations. This could have slightly influenced the results across these investigations, presumably due to effects of audibility, or possibly due to the signal being presented at a level which would reach different points on the psychometric function for sentence-recognition in noise for different individuals (Sobon et al. 2019). The other key difference between the present data and those obtained in previous investigations is the participant population. Here, the data in the present study were obtained from individuals referred for audiometric testing in a tertiary medical center because of suspected auditory and/or vestibular pathology. Many of these patients were also being followed by neurotologists for medical treatment. For example, some patients had retrocochlear pathology such as a vestibular schwannoma, whereas others had conductive components of varying etiology. Thus, it is possible that these other pathologies influenced the QuickSIN SNR losses in some participants beyond the audibility of the signal. In contrast, other investigations using the QuickSIN were largely prospective, laboratory-based studies, or were drawn from a population of veterans seeking care through the VA system. These subtle, but key differences in our participant populations may have contributed to these differences in QuickSIN SNR losses across different investigations. Nonetheless, we contend that the present data set are likely to reflect what would be observed in routine clinical practice, with the possibility of slightly overestimating mean QuickSIN performance for moderate or worse hearing losses. Although we attempted to obtain QuickSIN SNR losses on as many patients as possible, the time constraints of clinical practice precluded this for all individuals. Thus, audiologists may have been more likely to complete the QuickSIN on patients they were certain could perform the test in a reliable and timely manner. Doing so may have slightly reduced the QuickSIN SNR losses for some populations, particularly those with moderate or greater degrees of hearing loss by omitting patients for whom the test was likely to be particularly challenging.

Perhaps more important than the mean values are the wide range in WRQ and QuickSIN SNR losses observed across all degrees of hearing loss. Here, we observed considerable variability in QuickSIN SNR losses across all degrees of hearing loss (see Figs ​4 and ​5, and Tables ​3 and ​4). Previous investigations using the QuickSIN have not consistently reported individual data or the variability of performance on this measure with varying degrees of hearing loss. Wilson et al (2007) measured QuickSIN performance for lists 1 and 8 in individuals with normal hearing and with individuals who had a mean moderate HFPTA loss (~47 dB from visual inspection of their data). Visual inspection of their data suggests a range of ~2 to 7 dB SNR for individuals with normal hearing, and a range of ~6 to 22 dB SNR for the individuals with moderate HFPTA. Our data were somewhat more variable than those reported by Wilson et al (2007). Here, we observed ranges of −3.5 to 17.5 for individuals with HFPTA <15 dB HL, and from −1.5 to 23.5 dB SNR for the individuals with moderate HFPTA. As before, we anticipate this increase in variability largely reflects the patient population from which these data were obtained, and small methodological differences in measurement of the QuickSIN. Nonetheless, the variability observed in this clinical population suggest that viewing the effects of hearing loss in terms of average decrements in performance may be misleading in some instances. Rather, the effects of hearing loss may be best characterized by a greater likelihood of poor performance, albeit with increasing amounts of variability in patient outcomes. Clearly the number of individuals with significant difficulties on both tasks increased as the degree of hearing loss increased. However, the observation that some individuals continue to perform well, even in the presence of significant hearing loss, suggests that it may be an oversimplification to assume that hearing loss always results in a deterioration in performance on tests of speech understanding. Rather, it seems to increase the likelihood that someone will score more poorly on such a test.

QuickSIN Performance in Individuals With Normal Hearing

Another noteworthy aspect of our present data is the number of individuals with normal HFPTA values who had abnormal QuickSIN SNR losses. Here, 23.8% of individuals with HFPTA values <15 dB HL had QuickSIN SNR losses >3 dB SNR, indicating a mild SNR deficit according to the QuickSIN manual (Etymotic Research 2006). Even if we adopt a 4 dB SNR loss criteria to account for any binaural benefit, 13.2% of patients with HFPTA values <15 dB HL still had abnormally high QuickSIN SNR losses. These percentages are even higher for individuals with HFPTA values between 16 and 25 dB HL, with 44% having QuickSIN losses >3 and 30% having QuickSIN losses >4 dB SNR. Taken together, these results are consistent with data demonstrating that individuals with normal pure-tone thresholds who self-report difficulties understanding SIN often have a measurable deficit in SIN abilities (Saunders & Haggard 1989; Middelweerd et al. 1990; Vermiglio et al. 2018). With regard to the present data, there are a number of possibilities which could contribute to the high percentage of individuals with poor SNR abilities despite having normal HFPTA values.

One possibility is that some of these patients may have had auditory pathology which is not fully encapsulated by the normal HFPTA. For example, patients may have had a sloping hearing loss such that their thresholds at 4 kHz were abnormal, but not sufficiently so to push the HFPTA into the range of a mild or moderate loss. Alternatively, some of these patients may have had low-frequency hearing loss which would not be flagged by using HFPTA as a proxy for normal hearing thresholds. Other patients may have had a retrocochlear hearing loss such as a vestibular schwannoma which hindered their ability to understand SIN. Alternatively, some patients may have had an air-bone gap despite the normal HFPTA, whether from some sort of eustachian tube dysfunction, perforations of the tympanic membrane, superior semicircular canal dehiscence, or other auditory pathology. The patient base from which these data were drawn consisted of individuals referred for audiometric testing in a tertiary medical center because of suspected auditory and/or vestibular pathology. Thus, it is likely that a subset of these patients may not have had completely normal hearing despite the normal HFPTA, and that these other pathologies contributed to an abnormal QuickSIN SNR loss. However, it is important to note that any deficit in understanding SIN is not due to audibility. Here, the lowest presentation level used was 70 dB HL, and this level was adjusted depending on the configuration and degree of hearing loss. Moreover, WRQ scores in these patients were almost universally excellent in these patients. By this logic, these patients do have some challenges understanding SIN even when the signal is fully audible, but these challenges reflect some sort of auditory pathology that is not captured by the HFPTA.

A closely related possibility is that these individuals have with normal HFPTA but abnormal QuickSIN SNR losses have a deficit in peripheral encoding or central function that inhibits their ability to understand speech in background noise. For example, one possibility that has been put forth to explain normal audiometric thresholds with poor SIN understanding is the concept of a “hidden hearing loss” or cochlear synaptopy (Kujawa & Liberman 2009, 2015; Liberman et al. 2016; Liberman & Kujawa 2017). In this scenario, some individuals have a synaptic deafferentation in the cochlea which may be due to noise exposure, aging, or other processes. This results in auditory damage which is not visible in the audiometric thresholds or medical examination, but may hinder the ability to understand speech in the presence of background noise. An alternative possibility is that some individuals have small deficits in working memory or other forms of executive function which inhibit their ability to understand SIN despite the presence of normal hearing (Akeroyd 2008; Janse & Jesse 2014; Moore et al. 2014; Souza & Arehart 2015; Nagaraj 2017; Vermeire et al. 2019; Yeend et al. 2019).

In individuals with normal hearing, some reports suggest that the influence of central factors is influenced by age, such that working memory is related to SIN abilities in older individuals, but not younger (Füllgrabe & Rosen 2016). Given the design of this study, we cannot say whether some of the deficits observed here in individuals with normal hearing reflect cochlear synaptopy, executive function, or some other factor. Nonetheless, a high number of patients in this data set not only had normal HFPTA values, but normal audiometric thresholds from 250 to 8 kHz. The poor SIN abilities in these individuals is consistent with the concept that SIN difficulties may be present in a number of individuals with normal hearing thresholds (Saunders & Haggard 1989; Middelweerd et al. 1990; Vermiglio et al. 2018). These deficits are not captured by the traditional audiometric test battery of WRQ as performance in quiet was universally excellent in these patients.

Considerations for SIN Measurement in Clinical Practice

When considering the widespread implementation of speech in noise testing for clinical use, it is worth noting that there is a general agreement that the ability to understand SIN is affected by the audibility of the signal, with contributions from degradations in peripheral encoding as well as central auditory/cognitive abilities (see Humes & Dubno 2010 for a review). Consistent with this general concept, we observed mean decreases in QuickSIN SNR losses with increasing degrees of hearing loss. However, this relationship was not linear, and considerable variance in SIN abilities was observed within all degrees of loss, suggesting other factors beyond audibility contribute to performance on these measures. This is consistent with other large-scale investigations into SIN performance using clinically available measures. For example, Vermiglio et al. (2012) demonstrated little to no relationship between audiometric thresholds and the ability to repeat HINT sentences in the presence of steady-state background noise after applying age as a covariate. Wilson et al (2011) observed a largely linear relationship between HFPTA and WIN scores in over 3000 patients seen within the VA system. However, consistent with the present data, the authors also observed considerable variance in their WIN scores for a given degree of hearing loss, which greatly limited the predictive power of using the audiogram to estimate SIN abilities.

One implication from our data and these previous investigations is that, for clinical practice to reflect the ability of patients to communicate in their daily life, audiologists need to either (1) measure SIN abilities directly as with the present study (see also Wilson 2011 and Vermiglio et al. 2012), or (2) another measure should be added which assesses some degree of central processing. This approach is embodied by the recently developed “Word Auditory Recognition and Recall Measure” (Smith et al. 2016), which integrates measurement of monosyllabic word-recognition and working memory into a single test. Notably, the working-memory portion of this measure was able to predict a small but significant portion (4.3%) of variance on the WIN test in older adults with hearing loss (Smith et al. 2016). However, a separate report indicated no relationship between performance on the WARRM and SIN abilities as measured by the Multi-Modal Lexical Sentence Test for adults (Miller et al. 2017). Thus, it is uncertain the extent to which this measure can replace direct measurement of SIN abilities in clinical practice. Moreover, other measures which may predict a higher proportion of variance in SIN abilities are likely to either be too unwieldy or time-consuming for clinical use, and would likely fall prey to the same issues hindering the use of SIN measurement in current clinical practice (e.g., limited time, and uncertainty of how to use the additional information). The simplest solution would be to replace WRQ testing with measurement of SIN abilities in the standard audiologic test battery. The present data suggest that this approach is feasible because of the high accuracy in identifying patients with excellent or good WRQ scores from QuickSIN and HFPTA values.

Although the present investigation used the QuickSIN, there are a number of other measures which could be used to assess speech in noise abilities in clinical practice. As noted in the introduction, the WIN (Wilson 2003, 2011) has been widely studied, and has been implemented widely with veterans receiving care through the VA system. Moreover, like the QuickSIN, the WIN has a large separation in average performance between individuals with normal hearing and those with hearing loss, which Wilson and his team to suggest that the WIN or the QuickSIN were optimal for clinical use relative to high-context sentences such as those in the BKB-SIN or the HINT (Wilson et al. 2007). Of the latter measures, the HINT has been examined in large numbers of patients, and these data are largely consistent with the conclusions of Wilson et al., 2007. For example, the HINT scores observed in over 200 patients in Vermiglio et al (2012) appeared to vary less with increasing hearing loss than that observed in the present study, or in Wilson (2011). This led Vermiglio et al (2012) to suggest that performance on the HINT varied little with degree of hearing loss once the variance associated with age was partialed out in their statistical analyses.

Other investigators have attempted to develop tests of speech understanding in noise which could be used clinically to better depict patient performance. Since 1992, the speech recognition in noise test has been used to assess fitness of duty for soldiers in the military. In this measure, monosyllabic words are presented at a +9 dB SNR in the presence of six-person multitalker babble (Brungart et al. 2017). Subsequent investigations have built on this approach by utilizing tests which rely on binaural hearing for optimal performance (Phatak et al. 2018, 2019). These authors note that a potential limitation of tests such as the QuickSIN fail to account for or assess binaural function, which is critical for speech understanding in complex environments (Phatak et al. 2018). In these reports, Phatak et al (2018) compared performance on several tests to that on the QuickSIN, and suggested that the individuals with SIN deficits who score normally on traditional measures of speech perception in quiet could be best identified through a combination of measures that assess understanding of time-compressed and reverberant speech, as well as speech in which the signal and noise are out of phase. In a subsequent investigation, these authors created a composite score of functional hearing ability by measuring performance on measures including audiometric thresholds, traditional measures of WRQ, and several measures of speech understanding in noise. They then suggested that 80% to 90% of the variance on this composite score could be accounted for with the PTA, age, and speech understanding ability with binaural NU-6 lists presented at +9 dB SNR (Brungart et al. 2017) with a single female talker in the left ear (TKL), and then again in the right ear (TKR). They concluded that combining the NU-6 measure with TKL and TKR conditions would facilitate the ability of clinicians to identify SIN deficits resulting from hearing loss, and therefore make clinical evaluations more sensitive to the complaints of patients (Phatak et al. 2019).

There are several key differences between the approach utilized in previous investigations, and in the approach utilized to acquire the current data. First, as Phatak et al note, several of their measures require binaural hearing for optimal performance, and this may provide a more realistic measure of real-world performance. In contrast, monaural performance on the QuickSIN was our primary outcome measure. This is due in large part to the goal of our study, which was to predict classifications of monaural WRQ scores in an effort to make SIN measures replace WRQ measures in routine audiologic testing. Perhaps more relevant, the data here were drawn from individuals seen in a tertiary medical center, many of whom were also being followed by neurotologists for medical treatment. Thus, it is likely that the proportion of individuals with asymmetric hearing is considerably higher in the present population, and historically asymmetries in speech understanding have been of considerable interest in the evaluation process. The likely higher incidence of asymmetric hearing in medical environments may limit performance on these measures as well as complicate interpretation, thereby limiting the utility of tests that require binaural hearing for optimal performance. Thus, to adopt the approach recommended by Phatak et al (2019), it is possible that audiologists will need to balance the likelihood of symmetric hearing and/ or auditory pathology with the desire to include measures of SIN performance that rely on binaural hearing for optimal performance.

Clinical Implications

Regardless of the mechanism behind deficits in understanding SIN, our observation that classification of WRQ scores can be predicted by HFPTA thresholds and QuickSIN SNR losses has clear clinical implications. Perhaps most relevant for clinical practice, these data can be used to generate clinical recommendations in which SIN measures are the default test of speech perception in routine audiometric testing, and WRQ is only performed when it is likely to be suboptimal. This approach is analogous to the decision-making processes used by physicians when ordering MRI, or other diagnostic tests. For example, one application of these data would be to only perform WRQ when the HFPTA is ≥40 dB HL, and the QuickSIN is ≥7 or 8 dB SNR. Our data suggest that these criteria yield high sensitivity and specificity values for identifying patients at risk for suboptimal WRQ scores (see Figs ​6 and ​7). In this way, audiologists and physicians could readily integrate SIN measures into busy clinical practices, because WRQ would only be performed on a small subset of patients. Another application of these data would be to perform WRQ only when either the HFPTA is >40, or when the QuickSIN is >7 or 8 dB SNR. Doing so would result in a higher sensitivity than the “both classification” approach described previously, but at the loss of some specificity (see Tables ​5 and ​6). Thus, this approach would be more likely to identify patients with WRQ scores <88 or 76%, albeit with the cost of a higher number of individuals completing both WRQ and SIN measures. A more conservative application of these data would be for audiologists to forego WRQ measures if the HFPTA is <40 dB HL and the QuickSIN SNR loss is <7 dB SNR. Our data suggest that patients who have lower scores on both of these criteria have over a 99% likelihood of having WRQ scores between 88 and 100%. With this approach, audiologists could ensure that WRQ scores would not be conducted on a subgroup of individuals who are almost certain to have excellent performance. However, use of this approach would result in a considerably higher number of patients completing both QuickSIN and WRQ measures than only performing WRQ when HFPTA or QuickSIN exceeds a specific value. Taken together, the present data suggest that replacement of WRQ with measurement of SIN abilities is a viable application for clinical practice that would address patient concerns and provide information that is not well captured through the standard audiologic test battery.

CONCLUSIONS

It has been suggested that auditory function could be better represented by examining the audiometric thresholds and SIN abilities than by the current clinical practice of relying on auditory thresholds and measuring WRQ (Vermiglio et al. 2012). Here, we provide data from a clinical population consistent with this view, by demonstrating that many patients have abnormal SIN abilities despite having normal WRQ scores. More important for clinical feasibility, we demonstrate that patients with excellent or good WRQ scores can be predicted with high accuracy by the HFPTA and QuickSIN SNR losses. The predictive ability of our statistical model suggests that widespread measurement of SIN abilities can be feasibly achieved in a time-sensitive manner, as a single QuickSIN requires 57 seconds to complete (Wilson et al. 2007). More important, the predictive power of our model suggests that SIN can replace measures of WRQ in most patients with no loss of information. Making this subtle, but profound shift to clinical practice would better position the routine hearing test to be more sensitive to patient concerns, and has the potential for a number of benefits for both clinical practice and research studies.

ACKNOWLEDGMENTS

We thank Christian Bourdon, Brianne Davis, Cory Hillis, Rebecca Howard, Angela Huang, Lauren Jacobs, Amanda Murphy, Mateel Musallam, Sarah Pirko, Goutham Telukuntla, Sunny Yoon, Justin Cha, Rachael Jocewicz, Veronica Koo, Grace Nance, Devon Palumbo, Michael Smith, Soumya Venkitakrishan, and Madison Wageck for collecting these data as part of their normal clinical activities. We thank Gerald Popelka for his role in creating the database used to store these data. We would also like to thank Robert Jackler. Finally, we would like to thank Mona Taliaferro for her generous support of this project.

The odds ratio is a statistic which describes the strength of an association between two factors. It is defined as the ratio of the odds between factor 1 in the presence of factor 2, and vice versa. Odds ratios greater than 1 indicates that the factors are related, and that the presence of one factor raises the odds of another.

M.B.F. designed and implemented the study, assisted in data collection, analysis, preparation of figures, interpretation of data and played the lead role in article preparation. S.P.G. assisted in data collection, analysis, preparation of figures, and assisted with article preparation. Z.J.Q. assisted in data analysis. S.L. assisted in study design and preparation of figures. A.C.S. assisted in study design and implementation, and with data collection.

REFERENCES

• Akeroyd M. A. (2008). Are individual differences in speech reception related to individual differences in cognitive ability? A survey of twenty experimental studies with normal and hearing-impaired adults. Int J Audiol, 47, S53–S71. [PubMed] [Google Scholar]
• Arlinger S. (2003). Negative consequences of uncorrected hearing loss—a review. Int J Audiol, 42(Suppl 2), 2S17–2S20. [PubMed] [Google Scholar]
• Bossuyt P. M., Reitsma J. B., Bruns D. E., Gatsonis C. A., Glasziou P. P., Irwig L. M., Moher D., Rennie D., de Vet H. C. W., Lijmer J. G.; Standards for Reporting of Diagnostic Accuracy Group. (2003). The STARD statement for reporting studies of diagnostic accuracy: Explanation and elaboration. The Standards for Reporting of Diagnostic Accuracy Group. Croat Med J, 44, 639–650. [PubMed] [Google Scholar]
• Brungart D. S., Walden B., Cord M., Phatak S., Theodoroff S. M., Griest S., Grant K. W. (2017). Development and validation of the Speech Reception in Noise (SPRINT) Test. Hear Res, 349, 90–97. [PubMed] [Google Scholar]
• Carhart R., & Jerger J. F. (1959). Preferred method for clinical determination of pure-tone thresholds. J Speech Hear Disord, 24, 330–345. [Google Scholar]
• Carhart R., & Tillman T. (1970). Interaction of competing speech signals with hearing losses. Arch Otolaryngol, 91, 273–279. [PubMed] [Google Scholar]
• Carney E., & Schlauch R. S. (2007). Critical difference table for word recognition testing derived using computer simulation. J Speech Lang Hear Res, 50, 1203–1209. [PubMed] [Google Scholar]
• Cox R. M., & Alexander G. C. (1995). The abbreviated profile of hearing aid benefit. Ear Hear, 16, 176–186. [PubMed] [Google Scholar]
• Cox R. M., & Alexander G. C. (2002). The International Outcome Inventory for Hearing Aids (IOI-HA): Psychometric properties of the English version. Int J Audiol, 41, 30–35. [PubMed] [Google Scholar]
• Davidson A., Marrone N., Wong B., Musiek F. (2021). Predicting hearing aid satisfaction in adults: A systematic review of speech-in-noise tests and other behavioral measures. Ear Hear, 42, 1485–1498. [PubMed] [Google Scholar]
• Davis H. (1948). The articulation area and the social adequacy index for hearing. Laryngoscope, 58, 761–778. [PubMed] [Google Scholar]
• Dawes P., Emsley R., Cruickshanks K. J., Moore D. R., Fortnum H., Edmondson-Jones M., McCormack A., Munro K. J. (2015). Hearing loss and cognition: The role of hearing AIDS, social isolation and depression. PLoS One, 10, e0119616. [PMC free article] [PubMed] [Google Scholar]
• Dirks D. D., Kamm C., Bower D., Betsworth A. (1977). Use of performance-intensity functions for diagnosis. J Speech Hear Disord, 42, 408–415. [PubMed] [Google Scholar]
• Duncan K. R., Aarts N. (2006). A comparison of the HINT and QuickSIN test. J Speech-Lang Pathol Audiol, 30. [Google Scholar]
• Egan J. P. (1948). Articulation testing methods. Laryngoscope, 58, 955–991. [PubMed] [Google Scholar]
• Etymotic Research (2006). QuickSIN: Speech in noise test.
• Fitzgerald M. B., Ward K. M., Gianakas S. P., Smith M. L., Blevins N. H., Swanson A. C. (in revision). Speech in noise in the routine audiology test battery: Relationship to perceived auditory disability. Ear Hear. [Google Scholar]
• Fitzgerald M. B. & Ward K. M.. (in revision). Speech understanding in quiet and noise as a function of age and degree of hearing loss. Ear Hear. [Google Scholar]
• Füllgrabe C., & Rosen S. (2016). Investigating the role of working memory in speech-in-noise identification for listeners with normal hearing. Adv Exp Med Biol, 894, 29–36. [PMC free article] [PubMed] [Google Scholar]
• Gatehouse S., & Noble W. (2004). The Speech, Spatial and Qualities of Hearing Scale (SSQ). Int J Audiol, 43, 85–99. [PMC free article] [PubMed] [Google Scholar]
• Giolas T. G., Owens E., Lamb S. H., Schubert E. D. (1979). Hearing performance inventory. J Speech Hear Disord, 44, 169–195. [PubMed] [Google Scholar]
• Goman A. M., & Lin F. R. (2016). Prevalence of hearing loss by severity in the United States. Am J Public Health, 106, 1820–1822. [PMC free article] [PubMed] [Google Scholar]
• Halpin C., & Rauch S. D. (2009). Clinical implications of a damaged cochlea: Pure tone thresholds vs information-carrying capacity. Otolaryngol Head Neck Surg, 140, 473–476. [PubMed] [Google Scholar]
• High W. S., Fairbanks G., Glorig A. (1964). Scale for self-assessment of hearing handicap. J Speech Hear Disord, 29, 215–230. [PubMed] [Google Scholar]
• Hirsh I. J., Davis H., Silverman S. R., Reynolds E. G., Eldert E., Benson R. W. (1952). Development of materials for speech audiometry. J Speech Hear Disord, 17, 321–337. [PubMed] [Google Scholar]
• Holder J. T., Sheffield S. W., Gifford R. H. (2016). Speech Understanding in Children With Normal Hearing: Sound Field Normative Data for BabyBio, BKB-SIN, and QuickSIN. Otol. Neurotol, 37, e50–e55. [PubMed] [Google Scholar]
• Hosmer D. W., & Lemeshow S. (2000). Applied logistic regression (2nd ed., pp. 156–164). Wiley. [Google Scholar]
• Humes L. E. (2003). Modeling and predicting hearing aid outcome. Trends Amplif, 7, 41–75. [PMC free article] [PubMed] [Google Scholar]
• Humes L.E., Dubno J.R. (2010). Factors affecting speech understanding in older adults. In The Aging Auditory System. (pp. 211–257). New York: Springer. [Google Scholar]
• Hurley R. M., & Sells J. P. (2003). An abbreviated word recognition protocol based on item difficulty. Ear Hear, 24, 111–118. [PubMed] [Google Scholar]
• Janse E., & Jesse A. (2014). Working memory affects older adults’ use of context in spoken-word recognition. Q J Exp Psychol (Hove), 2006, 1842–1862. [PubMed] [Google Scholar]
• Jerger J., & Jerger S. (1971). Diagnostic significance of PB word functions. Arch Otolaryngol, 93, 573–580. [PubMed] [Google Scholar]
• Jr D.W.H., Lemeshow S., Sturdivant R.X. (2013). Applied Logistic Regression, John Wiley & Sons. [Google Scholar]
• Killion M. C., & Gudmundsen G. I. (2005). Fitting hearing aids using clinical prefitting speech measures: An evidence-based review. J Am Acad Audiol, 16, 439–447. [PubMed] [Google Scholar]
• Killion M. C., Niquette P. A., Gudmundsen G. I., Revit L. J., Banerjee S. (2004). Development of a quick speech-in-noise test for measuring signal-to-noise ratio loss in normal-hearing and hearing-impaired listeners. J Acoust Soc Am, 116, 2395–2405. [PubMed] [Google Scholar]
• Kujawa S. G., & Liberman M. C. (2009). Adding insult to injury: Cochlear nerve degeneration after “temporary” noise-induced hearing loss. J Neurosci, 29, 14077–14085. [PMC free article] [PubMed] [Google Scholar]
• Kujawa S. G., & Liberman M. C. (2015). Synaptopathy in the noise-exposed and aging cochlea: Primary neural degeneration in acquired sensorineural hearing loss. Hear Res, 330, 191–199. [PMC free article] [PubMed] [Google Scholar]
• Lawson G. D., Peterson M. E. (2011). Speech Audiometry, Plural Publishing. [Google Scholar]
• Le Prell C. G., & Clavier O. H. (2017). Effects of noise on speech recognition: Challenges for communication by service members. Hear Res, 349, 76–89. [PubMed] [Google Scholar]
• Liberman M. C., & Kujawa S. G. (2017). Cochlear synaptopathy in acquired sensorineural hearing loss: Manifestations and mechanisms. Hear Res, 349, 138–147. [PMC free article] [PubMed] [Google Scholar]
• Liberman M. C., Epstein M. J., Cleveland S. S., Wang H., Maison S.F. (2016). Toward a differential diagnosis of hidden hearing loss in humans. PLoS One, 11, e0162726. [PMC free article] [PubMed] [Google Scholar]
• Lin F. R. (2011). Hearing loss and cognition among older adults in the United States. J Gerontol A Biol Sci Med Sci, 66, 1131–1136. [PMC free article] [PubMed] [Google Scholar]
• Lin F. R., Ferrucci L., Metter E. J., An Y., Zonderman A. B., Resnick S. M. (2011). Hearing loss and cognition in the Baltimore Longitudinal Study of Aging. Neuropsychology, 25, 763–770. [PMC free article] [PubMed] [Google Scholar]
• McArdle R. A., Wilson R. H., Burks C. A. (2005). Speech recognition in multitalker babble using digits, words, and sentences. J Am Acad Audiol, 16, 726–39; quiz 763. [PubMed] [Google Scholar]
• Middelweerd M. J., Festen J. M., Plomp R. (1990). Difficulties with speech intelligibility in noise in spite of a normal pure-tone audiogram. Audiology, 29, 1–7. [PubMed] [Google Scholar]
• Miller C. W., Stewart E. K., Wu Y. -H., Bentler R.A., Tremblay K. (2017). Working memory and speech recognition in noise under ecologically relevant listening conditions: Effects of visual cues and noise type among adults with hearing loss. J Speech Lang Hear Res, 60, 2310–2320. [PMC free article] [PubMed] [Google Scholar]
• Moore D. R., Edmondson-Jones M., Dawes P., Fortnum H., McCormack A., Pierzycki R. H., Munro K. J. (2014). Relation between speech-in-noise threshold, hearing loss and cognition from 40-69 years of age. PLoS One, 9, e107720. [PMC free article] [PubMed] [Google Scholar]
• Nagaraj N. K. (2017). Working memory and speech comprehension in older adults with hearing impairment. J Speech Lang Hear Res, 60, 2949–2964. [PubMed] [Google Scholar]
• Newman C.W, Weinstein B.E, Jacobson G.P, Hug G.A. (1990). The hearing handicap inventory for adults: Psychometric adequacy and audiometric correlates. Ear Hear, 11, 430–433. [PubMed] [Google Scholar]
• Newman C.W, Weinstein B.E, Jacobson G.P, Hug G.A. (1991). Test-retest reliability of the hearing handicap inventory for adults. Ear Hear, 12, 355–357. [PubMed] [Google Scholar]
• Newman C.W, Hug G.A, Wharton J.A, Jacobson G.P. (1993). The influence of hearing aid cost on perceived benefit in older adults. Ear Hear, 14, 285–289. [PubMed] [Google Scholar]
• Nilsson M., Soli S. D., Sullivan J. A. (1994). Development of the hearing in noise test for the measurement of speech reception thresholds in quiet and in noise. J Acoust Soc Am, 95, 1085–1099. [PubMed] [Google Scholar]
• Noble W., Jensen N. S., Naylor G., Bhullar N., Akeroyd M. A. (2013). A short form of the Speech, Spatial and Qualities of Hearing scale suitable for clinical use: The SSQ12. Int J Audiol, 52, 409–412. [PMC free article] [PubMed] [Google Scholar]
• Peterson G. E., & Lehiste I. (1962). Revised CNC lists for auditory tests. J Speech Hear Disord, 27, 62–70. [PubMed] [Google Scholar]
• Phatak S. A., Sheffield B. M., Brungart D. S., Grant K. W. (2018). Development of a test battery for evaluating speech perception in complex listening environments: Effects of sensorineural hearing loss. Ear Hear, 39, 449–456. [PubMed] [Google Scholar]
• Phatak S. A., Brungart D. S., Zion D. J., Grant K. W. (2019). Clinical assessment of functional hearing deficits: Speech-in-noise performance. Ear Hear, 40, 426–436. [PubMed] [Google Scholar]
• Qian Z. J., Vaisbuch Y., Gianakas S. P., Tran E., Ali N., Blevins N. H., Fitzgerald M. B. (2023). Evaluation of speech in noise asymmetries in audiologic screening for vestibular schwannoma. Ear Hear Aug. 22. [Epub ahead of print]. [PMC free article] [PubMed] [Google Scholar]
• Saunders G. H., & Forsline A. (2006). The performance-perceptual test (PPT) and its relationship to aided reported handicap and hearing aid satisfaction. Ear Hear, 27, 229–242. [PubMed] [Google Scholar]
• Saunders G. H., & Haggard M. P. (1989). The clinical assessment of obscure auditory dysfunction-1. Auditory and psychological factors. Ear Hear, 10, 200–208. [PubMed] [Google Scholar]
• Smith M. L., Winn M. B., Fitzgerald M. B. (in revision). A large-scale study of the relationship between degree & type of hearing loss and recognition of speech in quiet and noise. Ear Hear. [Google Scholar]
• Smith S. L., Pichora-Fuller M. K., Alexander G. (2016). Development of the word auditory recognition and recall measure: A working memory test for use in rehabilitative audiology. Ear Hear, 37, e360–e376. [PubMed] [Google Scholar]
• Sobon K. A., Taleb N. M., Buss E., Grose J. H., Calandruccio L. (2019). Psychometric function slope for speech-in-noise and speech-in-speech: Effects of development and aging. J Acoust Soc Am, 145, EL284. [PMC free article] [PubMed] [Google Scholar]
• Souza P., & Arehart K. (2015). Robust relationship between reading span and speech recognition in noise. Int J Audiol, 54, 705–713. [PMC free article] [PubMed] [Google Scholar]
• Tillman T. W., & Carhart R. (1966). An expanded test for speech discrimination utilizing CNC monosyllabic words. Northwestern University Auditory Test No. 6. SAM-TR-66-55. Tech Rep SAM-TR:1–12. [PubMed] [Google Scholar]
• U.S. Census Bureau (2010). Bay Area Census. Available at: https://doi.org/http://www.bayareacensus.ca.gov/bayarea.htm.
• Vermeire K., Knoop A., De Sloovere M., Bosch P., van den Noort M. (2019). Relationship between working memory and speech-in-noise recognition in young and older adult listeners with age-appropriate hearing. J Speech Lang Hear Res, 62, 3545–3553. [PubMed] [Google Scholar]
• Vermiglio A. J., Soli S. D., Freed D. J., Fisher L. M. (2012). The relationship between high-frequency pure-tone hearing loss, hearing in noise test (HINT) thresholds, and the articulation index. J Am Acad Audiol, 23, 779–788. [PubMed] [Google Scholar]
• Vermiglio A. J., Soli S. D., Fang X. (2018). An argument for self-report as a reference standard in audiology. J Am Acad Audiol, 29, 206–222. [PubMed] [Google Scholar]
• Walden T. C., & Walden B. E. (2004). Predicting success with hearing aids in everyday living. J Am Acad Audiol, 15, 342–352. [PubMed] [Google Scholar]
• Wilson R. H. (2003). Development of a speech-in-multitalker-babble paradigm to assess word-recognition performance. J Am Acad Audiol, 14, 453–470. [PubMed] [Google Scholar]
• Wilson R. H. (2011). Clinical experience with the words-in-noise test on 3430 veterans: Comparisons with pure-tone thresholds and word recognition in quiet. J Am Acad Audiol, 22, 405–423. [PubMed] [Google Scholar]
• Wilson R. H., McArdle R. A., Smith S. L. (2007). An evaluation of the BKB-SIN, HINT, QuickSIN, and WIN materials on listeners with normal hearing and listeners with hearing loss. J Speech Lang Hear Res, 50, 844–856. [PubMed] [Google Scholar]
• Yeend I., Beach E. F., Sharma M. (2019). Working memory and extended high-frequency hearing in adults: Diagnostic predictors of speech-in-noise perception. Ear Hear, 40, 458–467. [PubMed] [Google Scholar]
• Zwolan T. A., Schvartz-Leyzac K. C., Pleasant T. (2020). Development of a 60/60 guideline for referring adults for a traditional cochlear implant candidacy evaluation. Otol Neurotol, 41, 895–900. [PubMed] [Google Scholar]

---

Articles from Ear and Hearing are provided here courtesy of Wolters Kluwer Health

---