Can uHear Me Now? Validation of an iPod-Based ...

ARTICLE

Can uHear Me Now? Validation of an iPod-Based Hearing Loss Screening Test Jacek Szudek, MD, PhD, Amberley Ostevik, BEng, MSc, RAud, Aud(C), Peter Dziegielewski, MD, Jason RobinsonAnagor, Nahla Gomaa, MD, Bill Hodgetts, PhD, and Allan Ho, MBBS, MSc, FRCS (ORL-HNS) ABSTRACT Objective: To evaluate the uHear iPod-based application as a test for hearing loss. Methods: We recruited 100 adult participants through a single otology practice. Patients with otorrhea and cognitive impairment were excluded. All patients completed the uHear test in the clinic and in the sound booth and underwent a standard audiogram by the same audiologist. We compared the results of the uHear test to the standard audiogram. Results: The uHear was able to correctly diagnose the presence of hearing loss (pure-tone average [PTA] . 40 dB) with a sensitivity of 98% (95% CI 5 89–100), a specificity of 82% (95% CI 5 75–88), and a positive likelihood ratio of 9 (95% CI 5 6.0–16). Compared to the audiogram, the uHear overestimated the PTA among all ears by 14 dB in the clinic and by 8 dB in the sound booth (p , .0001). Compared to the audiogram, the uHear overestimated the PTA among ears with hearing loss by 6 dB in the clinic and by 4 dB in the sound booth. Conclusions: The uHear application is a reasonable screening test to rule out moderate hearing loss (PTA . 40 dB) and and is valid at quantifying the degree of hearing loss in patients known to have abnormal hearing. SOMMAIRE Objectif: L’e´tude visait a` e´valuer l’application uHear, sur iPod, comme test de de´pistage de perte auditive. Me´thode: Nous avons retenu 100 participants adultes dans un seul centre d’otologie. Les patients pre´sentant de l’otorrheé ou atteints d’une de´ficience cognitive ont e´te´ ećarte´s. Tous les sujets ont reálise´ le test uHear au centre et en cabine et ont e´te´ soumis a` un audiogramme ordinaire, effectue´ par le meˆme audiologiste. Nous avons ensuite compare´ les re´sultats du test uHear avec ceux de l’audiogramme. Re´sultats: Le test uHear pouvait de´tecter correctement la pre´sence d’une perte auditive (moyenne des sons purs [MSP] . 40 dB); il avait une sensibilite´ de 98% (IC a` 95% 5 89–100), une spećificite´ de 82% (IC a` 95% 5 75–88), et un rapport de vraisemblance de 9 (IC a` 95% 5 6.0–16). Comparativement a` l’audiogramme, le test uHear s’est solde´ par une surestimation de la MSP, de 14 dB au centre et de 8 dB en cabine (p , .0001), dans l’ensemble des oreilles, et par une surestimation de la MSP, de 6 dB au centre et de 4 dB en cabine, pour ce qui est des oreilles affecteés d’une perte auditive. Conclusions: L’application uHear est un test de de´pistage relativement bon, qui permet d’ećarter une perte auditive mode´reé (MSP] . 40 dB), et valide en ce qui concerne la quantification du degre´ de perte auditive chez les patients connus pour avoir des troubles auditifs. Key words: audiogram, hearing loss, screening

earing impairment is the most frequent sensory deficit in humans, affecting more than 250 million people worldwide.1 The number of Americans with

H

Jacek Szudek, Peter Dziegielewski, Nahla Gomaa, and Allan Ho: Department of Otolaryngology-Head and Neck Surgery, Amberley Ostevik and Bill Hodgetts: Department of Audiology, and Jason Robinson-Anagor: School of Medicine, University of Alberta, Edmonton, AB. Address reprint requests to: Jacek Szudek, MD, PhD, 1E4.33 WMC, University of Alberta Hospital, 8440-122th St, Edmonton AB T6G 2B7.

DOI 10.2310/7070.2011.110089 # 2012 The Canadian Society of Otolaryngology-Head & Neck Surgery

S78

hearing loss has doubled during the past 30 years, to 30 million people. The onset of hearing loss varies across individuals but is common by middle age.2 By age 70, approximately 30% of the US population perceive themselves to be hearing impaired, and about 50% of adults age 80 and older report being hearing impaired.3 Furthermore, over the last 10 years, the prevalence of hearing loss among American adolescents has increased from 15 to 20%,4 and 12.5% of American children and young adults experience temporary noise-induced hearing threshold shifts.5 In Canada, as many as 3 million people suffer from some degree of hearing loss and 1 million describe their hearing problem as a disability.6

Journal of Otolaryngology-Head & Neck Surgery, Vol 41, No S1 (April), 2012: pp S78–S84

Szudek et al, Validation of an iPod-Based Hearing Test

The gold standard for diagnosing hearing loss is the audiogram. Unfortunately, there can be financial, geographic, and logistical obstacles to obtaining a timely audiogram.7–9 These obstacles could lead to delays in the diagnosis of hearing loss. This can be devastating when it comes to sudden sensorineural hearing loss because a delay in treatment may be associated with lower rates of hearing recovery.10 In school-age children, even slight hearing loss can create a need for speech therapy or auditory training.11 In elderly patients, the risk of dementia is three to five times higher in those with moderate or worse hearing loss.12 Moreover, based on estimates from foreign studies, the cost of hearing loss to the Canadian economy may be as high as $18 billion per year.13 Over the years, several strategies and devices have been developed to facilitate earlier diagnosis of hearing loss. The Hearing Handicap Inventory and other clinic-, postal-, telephone-, and Web-based questionnaires have been reported as reliable but do not offer objective results.14–16 Automated, patient-operated audiometers have been available for over 50 years.17 The latest automated audiometers can test air and bone conduction hearing thresholds and have been validated against the standard audiogram.8,18 However, access to these devices is still limited by cost and time. Recently, an iPod/iPhone-based application was developed for personal use as a screening tool for hearing loss. The uHear is freely available for download through iTunes and can test for air conduction hearing loss in under 6 minutes using standard earbud headphones.19 To date, no independent study has investigated the validity of the uHear application in screening for hearing loss. The desirable features of a safe and efficient screening test include a high sensitivity (ie, low rate of false-negative results) and a high specificity (ie, low rate of false-positive results). Therefore, the primary objective of this study was to evaluate the sensitivity and specificity of the uHear application in diagnosing hearing loss compared to the gold standard, the audiogram. If uHear proves accurate, it may be a useful clinical tool to screen for hearing loss. This, in turn, may help identify patients who need further assessment in the form of a standard audiogram.

S79

fluent in English and lacked active otorrhea and cognitive impairment. Informed consent was obtained from each study participant. Ethics approval was obtained from the University of Alberta Health Research Ethics Board. uHear Application The uHear application, version 1.0, was designed and developed by Don Hayes, PhD, director of audiology for Unitron, a Kitchener, Ontario–based, global manufacturer and innovator of hearing aids and other technologically-advanced instruments. The uHear runs on iOS devices such as the iPod, iPhone, and iPad and was released on iTunes in May 2009.19 The application enables users to test their pure-tone air conduction hearing sensitivity as well as speech in noise. For the purposes of this validation study, participants completed only the hearing sensitivity portion. The uHear hearing sensitivity test determines the quietest air-conducted sound the subject can hear at 250, 500, 1000, 2000, 4000, and 6000 Hz in both the right and left ears. Before each test begins, the uHear prompts the participant to (1) go to a very quiet environment, (2) place the headphones or earbuds in the correct ears, (3) set the device volume to 50% (with a slider tool to help complete this calibration step), and (4) select the listening device type being used to complete the test (earbuds or headphones). After the test begins, the participant presses a large button on the touch screen to indicate when a sound is heard. The uHear uses a 267 ms pulse duration and employs a simple ‘‘10 dB down and 5 dB up’’ approach. The lowest threshold with two positive responses of three excursions is recorded as the hearing sensitivity. The delay time between tone presentations is randomized to prevent anticipation. At the end of the evaluation, hearing sensitivity is displayed in a typical audiogram format, indicating pure-tone air conduction hearing sensitivities. Study Design All subjects underwent three hearing evaluations:

Methods Participant Selection One hundred adult participants were recruited through a single otology practice in January 2011 at the Grey Nuns and University of Alberta Hospitals in Edmonton, Alberta. All included participants were over 18 years of age and

1. Standard air conduction threshold screening assessment by a single audiologist (A.O.) in a sound-treated booth according to the Hughson-Westlake algorithm20,21 2. Self-administered uHear testing in a sound-treated booth 3. Self-administered uHear testing in a clinical setting, such as a waiting room or office, with an average

S80

Journal of Otolaryngology-Head & Neck Surgery, Volume 41, Number S1, 2012

ambient noise level of less than 50 dBA-weighted sound pressure level The three tests were completed for each subject in randomized order during a single visit. All clinical audiometers and accessory equipment were calibrated per ANSI S3.6, 1996.21 All uHear tests used the same iPod Touch (iOS 4.2) and the same pair of standard earbud headphones. A clinical investigator or research support staff member gave verbal instructions for the uHear application and was present to give supplementary or complementary instructions if questions arose. External auditory canals were examined and debrided, if necessary, prior to audiologic testing. The audiologist was blinded to former audiometric results. The time elapsed to complete the audiologist’s assessment was also measured. Data Analysis The pure-tone average (PTA) was calculated as the mean of hearing thresholds at 500, 1000, 2000, and 4000 Hz. The presence or absence of moderate or worse hearing loss (PTA . 40 dB HL) in each of the tested ears was determined by the (1) uHear application in the clinic, (2) uHear application in the sound booth, and (3) audiogram. The data were entered into 2 3 2 tables, which were then used to calculate sensitivity, specificity, and the likelihood ratio. In addition, we use paired t-tests with a Bonferroni correction to compare the air conduction PTAs obtained by the uHear (in the sound booth and in the clinic) to those obtained from a proper audiogram. Microsoft Excel 2008 for Mac version 12.2.8 and SPSS 16.0.1 (SPSS Inc, Chicago, IL) were used to perform all statistical analyses.

Results Participant Demographics One hundred participants were recruited and tested. Table 1 summarizes their demographics. Is uHear an Accurate Screening Test?

Table 1. Participant Demographics Number Total number Age (yr) Mean Range Number according to age (yr) 20–40 41–60 61–80 $ 81 Gender Female Male Number with perceived hearing loss Number with audiometric hearing loss (PTA . 40 dB) Unilateral Bilateral Number of ears with PTA (dB) # 25 26–40 41–55 56–70 71–90 $ 91 uHear test time (mean no. of minutes)

46 20–91 40 41 17 2 67 33 64 41 29 12 122 27 25 8 12 6 5 min, 1 s

PTA 5 pure-tone average.

89–100%). Of the 145 ears without moderate or worse hearing loss documented on the audiogram, 26 had moderate or worse hearing loss by uHear testing. This translates to a specificity of 82% (95% CI 5 75–88). The positive likelihood ratio was 5.5 (95% CI 5 4.2–12). uHear Testing in the Sound Booth Table 3 compares the results of the uHear measured in the sound booth to the standard audiogram. Of the 55 ears with moderate hearing loss (PTA . 40 dB HL) Table 2. Accuracy of uHear as a Screening Test Compared to a Standard Audiogram in the Clinic Audiogram

uHear Testing in the Clinic Table 2 compares the results of the uHear measured in the clinic to the standard audiogram. Of the 55 ears with moderate hearing loss (PTA . 40 dB HL) documented on the audiogram, only one had a PTA # 40 dB by uHear testing. This translates to a sensitivity of 98% (95% CI 5

100

uHear PTA # 40 dB PTA . 40 dB Total PTA 5 pure-tone average.

PTA # 40 dB

PTA . 40 dB

119 26 145

1 54 55

S81


Table 3. Accuracy of uHear as a Screening Test Compared to a Standard Audiogram in the Sound Booth Audiogram uHear PTA # 40 dB PTA . 40 dB Total

PTA # 40 dB

PTA . 40 dB

130 15 145

0 55 55

PTA 5 pure-tone average.

documented on the audiogram, none had PTA # 40 dB by uHear testing. This translates to a sensitivity of 100% (95% CI 5 92–100%). Of the 145 ears without moderate or worse hearing loss documented on the audiogram, 15 had moderate or worse hearing loss by uHear testing. This translates to a specificity of 90% (95% CI 5 83–94%). The positive likelihood ratio was 9.7 (95% CI 5 6.0–15). Are uHear Pure-Tone Hearing Thresholds Accurate? uHear Accuracy in All Ears Figure 1 shows the mean pure-tone air conduction thresholds for all ears tested by the uHear in the clinic, the uHear in the sound booth, and the standard audiogram. The mean PTA by the audiogram was 28 dB. The PTA by uHear in the booth was 8 dB greater than the PTA by the audiogram (p , .0001). The PTA by uHear in the clinic was 14 dB greater than the PTA by the audiogram (p , .0001).

uHear Accuracy in Ears with Hearing Loss Figure 2 shows the same outcome variables as Figure 1 but includes only the 78 ears that had mild or worse hearing loss (. 25 dB HL) according to the standard audiogram. The mean PTA by the audiogram was 54 dB. The PTA by the uHear in the sound booth was 4 dB greater than the PTA by the audiogram (p , .0001). The PTA by the uHear in the clinic was 6 dB greater than the PTA by the audiogram (p , .0001). Figure 3 compares the difference between the PTA by the audiogram and the PTA by the uHear in the sound booth. Among all ears tested by the uHear, 67% had a PTA within 10 dB of the audiogram. Among those with hearing loss (PTA . 25 dB HL), 86% had a PTA within 10 dB of the audiogram. Figure 4 compares the difference between the PTA by the audiogram and the PTA by the uHear in the clinic. Among all ears tested by the uHear, 41% had a PTA within 10 dB of the audiogram. Among those with hearing loss (PTA . 25 dB HL), 74% had a PTA within 10 dB of the audiogram. Effect of Ambient Noise on uHear Accuracy The effect of ambient noise was estimated by comparing uHear thresholds in the clinic to uHear in the sound booth. Among all ears, the difference in PTA was 6 dB (p , .0001) but approached 10 dB at lower frequencies (see Figure 1). Among those with hearing loss, the difference in PTA was 2 dB (p , .01) (see Figure 2).

Figure 1. Average air-conduction hearing thresholds among all ears as measured by a standard audiogram, uHear in the sound booth, and uHear in the clinic.

S82


Figure 2. Average hearing thresholds as measured by a standard audiogram, uHear in the sound booth, and uHear in the clinic. Only ears with a puretone average . 25 dB HL by audiogram are shown.

Discussion The audiogram is unquestionably the gold standard for measuring hearing sensitivity. Unfortunately, there can be geographic, financial, or logistical obstacles to obtaining an audiogram. These could potentially lead to delays in the diagnosis of hearing loss. This is associated with speech delay in children, lower rates of recovery from sudden sensorineural hearing loss, and increased risk of incident dementia in the elderly.10–12 The uHear is a freely available, iPod/iPhone-based application that quickly measures pure-tone air conduction hearing thresholds.19 Ours is the first independent study validating the uHear application in a clinical setting.

uHear Is a Valid Screening Test Because the uHear is 98 to 100% sensitive (see Table 2 and Table 3), a negative result can be used to confidently rule out moderate or worse hearing loss in a quiet clinical setting using standard-issue earbud headphones. In addition, a positive uHear test also has diagnostic utility, as demonstrated by its positive likelihood ratio of 9. The positive likelihood ratio is defined as the probability of a person who has the diagnosis testing positive divided by the probability of a person who does not have the diagnosis testing positive. In this case, the diagnosis of interest is hearing loss. A likelihood ratio of 9 is clinically relevant because a positive uHear test can

Figure 3. Difference in pure-tone average (PTA) between uHear in the sound booth and the audiogram, among all ears and among ears with a PTA . 25 dB HL.


S83

Figure 4. Difference in pure-tone average (PTA) between uHear in the sound booth and the audiogram, among all ears and among ears with a PTA . 25 dB HL.

be applied to the pretest probability of a patient having hearing loss to estimate a posttest probability that is 40% higher.22

uHear Pure-Tone Thresholds Are Accurate in Some Patients Although the uHear application can be used to estimate pure-tone hearing air conduction thresholds such as those measured by an audiogram, there are definite limitations in this capability. When it comes to quantifying the degree of hearing loss, the uHear is clearly more accurate in quieter testing environments—the PTA was 2 to 6 dB closer to the audiogram in the sound booth than in the clinic. The uHear is also more accurate in ears that are already known to have hearing loss (PTA . 25 dB)—the uHear PTA was only 4 to 6 dB greater than the audiogram (see Figure 1 and Figure 2). Therefore, the uHear may be clinically useful in watching for a change in hearing among patients with established hearing loss. For this reason, additional research into the test–retest reliability of the uHear warrants serious consideration.

Limitations In the study design, we chose 40 dB as the critical hearing threshold for screening. This was done for two reasons. Whereas the clinical significance of mild hearing loss may be questioned, moderate (or worse) hearing loss clearly warrants further investigation. Also, the 16-bit digital-toanalog converter in the iPod/iPhone limits the dynamic range of the uHear application to approximately 85 dB (15– 100 dB) (Don Hayes, personal communication, February 25, 2011). Given that the lower limit of the dynamic range falls within normal hearing thresholds (0–25 dB), the uHear may lack specificity at its lower range. Therefore, it is not surprising that the uHear overestimated thresholds in normal-hearing ears (see Figure 3 and Figure 4). The fundamental unit for statistical analysis was the ear rather than the person. Given that the outcome of interest was testability rather than the actual degree of hearing loss, the expected correlation in hearing level between a person’s ears would not violate statistical testing.23 One might assume, however, that the testability of a given person’s two ears would be correlated. In addition, this study and the uHear application considered air conduction

Key Points Hearng loss is a common problem There are serious consequences to delays in the diagnosis of hearing loss The uHear hearing test is easily-accessible, free and fast uHear is a reasonable screening test for hearing loss (sensitivity = 98%; specificity = 82%) uHear ovrestimates pure-tone hearing thresholds in normal-hearing ears, but is accurate in ears with established hearing loss

S84


hearing only and did not consider the potential confounding effect of crossover. 7.

Conclusion 8.

Despite the above limitations, this study shows that the uHear application is a reasonable screening test to rule out moderate hearing loss (PTA . 40 dB) and is accurate at quantifying the degree of hearing loss in patients known to have abnormal hearing. However, this diagnostic utility can be ensured only if it used under the supervision of a knowledgeable operator rather than by the patient alone. In future studies, we intend to test the real-world effectiveness (eg, in a family physician’s office or by otolaryngology residents on call) and the test–retest reliability of the uHear application in assessing hearing loss—particularly among the populations that may have difficulty accessing timely audiograms.

Acknowledgement

9.

10. 11. 12.

13.

14.

Financial disclosure of authors and reviewers: None reported.

15.

References

16.

1. Mathers C, Smith A, Concha M. Global burden of hearing loss in the year 2000. Geneva: World Health Organization; 2000. 2. Agrawal Y, Platz EA, Niparko JK. Prevalence of hearing loss and differences by demographic characteristics among US adults: data from the National Health and Nutrition Examination Survey, 19992004. Arch Intern Med 2008;168:1522–30, doi:10.1001/archinte. 168.14.1522. 3. Desai M, Pratt LA, Lentzner H, Robinson KN. Trends in vision and hearing among older Americans. Aging Trends 2001;2:1–8. 4. Shargorodsky J, Curhan SG, Curhan GC, Eavey R. Change in prevalence of hearing loss in US adolescents. JAMA 2010;304:772– 8, doi:10.1001/jama.2010.1124. 5. Niskar AS, Kieszak SM, Holmes AE, et al. Estimated prevalence of noise-induced hearing threshold shifts among children 6 to 19 years of age: the Third National Health and Nutrition Examination Survey, 1988-1994, United States. Pediatrics 2001;108:40–3, doi:10.1542/peds.108.1.40. 6. Statistics Canada. Prevalence of disability among adults aged 15 years and over, by type of disability, Canada, 2001. Statistics

17.

18. 19. 20.

21. 22. 23.

Canada, Participation and Activity Limitation Survey. Available at: http://www.statcan.gc.ca/pub/89-577-x/t/4065020-eng.htm (accessed April 1, 2011). Gates GA, Murphy M, Rees TS, Fraher A. Screening for handicapping hearing loss in the elderly. J Fam Pract 2003;52:56–62. Ho ATP, Hildreth AJ, Lindsey L. Computer-assisted audiometry versus manual audiometry. Otol Neurotol 2009;30:876–83, doi:10.1097/MAO.0b013e3181b120d0. Donahue A, Dubno JR, Beck L. Guest editorial: accessible and affordable hearing health care for adults with mild to moderate hearing loss. Ear Hear 2010;31:2–6, doi:10.1097/AUD.0b013e 3181cbc783. O’Malley MR, Haynes DS. Sudden hearing loss. Otolaryngol Clin North Am 2008;41:633–49, doi:10.1016/j.otc.2008.01.009. Northern JL, Downs MP. Hearing in children. 5th ed. Baltimore (MD): Williams & Wilkins; 2001. Lin FR, Metter EJ, O’Brien RJ, et al. Hearing loss and incident dementia. Arch Neurol 2011;68:214–20, doi:10.1001/archneurol. 2010.362. CRC Hear and the Victorian Deaf Society. Listen Hear! The Economic impact and cost of hearing loss in Australia. Available at: http://www.accesseconomics.com.au/publicationsreports/showreport. php?id571 (accessed April 1, 2011). Weinstein BE. Validity of a screening protocol for identifying elderly people with hearing problems. ASHA 1986;28:41–5. Sindhusake D, Mitchell P, Smith W, et al. Validation of selfreported hearing loss. The Blue Mountains Hearing Study. Int J Epidemiol 2001;30:1371–8, doi:10.1093/ije/30.6.1371. Koopman J, Davey E, Thomas N, et al. How should hearing screening tests be offered? Int J Audiol 2008;47:230–7, doi:10.1080/ 14992020801908236. McMurray RF, Rudmose W. An automatic audiometer for industrial medicine. Noise Control 1956;2:33–6, doi:10.1121/ 1.2369175. Wood TJ, Wittich WW, Mahaffey RB. Computerised pure tone audiometric procedures. J Speech Hear Res 1973;16:676–84. Unitron. uHear. Available at: http://itunes.apple.com/ca/app/ uhear/id309811822?mt58 (accessed April 1, 2011). American National Standards Institute. ANSI S3.21-1978 (R1997). In: Method for manual pure-tone threshold audiometry. New York: American National Standards Institute; 1997. American National Standards Institute. ANSI S3.6-1996. In: Specification for audiometers. New York: ANSI; 1996. McGee S. Simplifying likelihood ratios. J Gen Intern Med 2002;17: 646–9, doi:10.1046/j.1525-1497.2002.10750.x. Rosner B. Statistical methods in ophthalmology: an adjustment for the intraclass correlation between eyes. Biometrics 1982;38:105–14, doi:10.2307/2530293.

Copyright of Journal of Otolaryngology -- Head & Neck Surgery is the property of Decker Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Copyright of Journal of Otolaryngology -- Head & Neck Surgery is the property of Decker Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.