An existing tool that can help guide more consumers to hearing help
According to a national survey by the Better Hearing Institute (BHI),1 50% of adults with hearing loss do not use hearing aids because they have not had their hearing professionally tested. Offering free hearing tests in hearing care professional offices may be effective in bringing some people into hearing care practices. However, for the majority of people with hearing loss, being better informed regarding the status of their hearing loss may be the first step to better hearing.
Offering people quick and easy methods for assessing their hearing loss can provide them with more information about their hearing loss and move them closer to seeking a solution. To this end, the Better Hearing Institute offers a Quick Hearing Check in written form (see sidebar below), as well as an online automatic-scoring version of the test at www.hearingcheck.org. The test provides the consumer with hearing loss norms based on the National Council on Aging (NCOA) study2 and also descriptively evaluates how the individual’s significant other views the individual’s hearing loss. Additionally, the BHI Quick Hearing Check is used in the national MarkeTrak surveys as part of the battery to quantify and segment people on subjective hearing loss.
We first became aware of this simple test during the construction of the NCOA survey.2 The BHI Quick Hearing Check is a 15 item, 5-point Likert-scaled hearing loss inventory based on the revised American Academy of Otolaryngology-Head and Neck Surgery (AAO-HNS) five-minute hearing test. It has excellent “face validity” in that the items assess common problems of people with hearing loss that are generally accepted as “signs” of hearing loss by hearing care practitioners. It has been shown to be correlated with objective measures of hearing loss by Koike and her colleagues.3 In a sample of 74 adults, the revised inventory was shown to be correlated with speech reception threshold scores (r=.49), speech discrimination scores (r=-.33), speech puretone averages in the better ear (r=.60), and high frequency puretone averages (r=.61).
The revised AAO-HNS five-minute hearing test is scaled on a 4-point categorical scale in which the respondent indicates the frequency with which they experience the problem stated in each item: “Almost always,” “Half the time,” “Occasionally,” and “Never.” The author of the NCOA survey changed the scaling to a 5-point interval scale anchored at each end of the scale with the terms “Strongly agree” to “Strongly disagree.”4 We are in agreement with this change in scaling due to its superior psychometric properties. BHI has embraced this scaling in all post-NCOA studies.
To our knowledge, the revised AAO-HNS five-minute hearing check has not been explored other than in Koike’s 74-subject study in one clinic, nor is it currently in use on the AAO-HNS Web site (the Web site offers the original AAO-HNS five-minute check). Koike has shown that both the reliability of the measure and correlations with other audiological data are lower with the original version than the revised inventory.
Given the revised scaling and its current wide-scale use on the BHI Web site and by practitioners in the field, we thought it would be useful to explore both its audiological validity among a large sample of subjects across a wide number of practices, as well as its reliability, unidimensionality, and its concurrent validity against other subjective measures of hearing loss or quality of life factors. Four databases were utilized for this study:
- Audiological data (gathered in 2009) across 64 clinics for 987 patients;
- MarkeTrak VIII5 (2009) database comprised of a nationally representative sample of 7,201 individuals with hearing loss;
- NCOA (1998) database comprised of a nationally representative sample of 2,304 adults ages 55 or higher with hearing loss; and
- NCOA (1998) spousal perception database comprised of 2,439 spouses or significant others of individuals with hearing loss.
The MarkeTrak VIII survey, NCOA survey, and BHI Quick Hearing Check are all available for download at www.betterhearing.org/professionals/tools.cfm under the topic “Instruments.”
Audiological database. A total of 64 audiologists and hearing instrument specialists were recruited to participate in this study in 2008-2009. An online data entry screen was designed for the BHI Web site. Practitioners administered the BHI Quick Hearing Check to 987 patients and collected puretone threshold scores for both the right and left ears at 500, 1000, 2000, 3000, and 4000 Hz. Additionally, speech recognition scores for the patient’s better ear were collected using various speech understanding tests including: W22 (68%), NU6 (23%), and others (8%).
All data, including age and gender, were entered into the BHI Web database. Subject identities were kept confidential by the practitioners.
Descriptive statistics for this sample are presented in Table 1. The mean age for this sample was 72 years, split nearly equally between males and females.
|TABLE 1. Descriptive statistics and audiological measures (n=987).|
MarkeTrak database.6 In November and December 2008, a short screening survey was mailed to 80,000 members of the National Family Opinion (NFO) panel. The NFO panel consists of households that are balanced to the latest US census information with respect to market size, age of household, size of household, and income within each of the nine census regions, as well as by family versus non-family households, state (with the exception of Hawaii and Alaska), and the nation’s top 25 metropolitan statistical areas.
This short screening survey was completed by 46,843 households, and it helped identify 14,623 people with hearing loss and provided detailed demographics on those individuals and their households. The response rate to the screening survey was 59%. In January 2009, an extensive 7-page legal-size survey was sent to the total universe of hearing aid owners in the panel database (3,789); 3,174 completed surveys were returned representing an 84% response rate. In February 2009, an extensive 7-page legal-size survey was sent to a random sample of 5,500 people with hearing loss who had not yet adopted hearing aids. The response rate for the non-adopter survey was 79%. Both hearing aid owners and non-adopters were given a $1 incentive to complete and return their surveys.
The data presented in this article refer only to households as defined by the US Bureau of Census—that is, people living in a single-family home, duplex, apartment, condominium, mobile home, etc. Not surveyed were people living in institutions, including residents of nursing homes, retirement homes, mental hospitals, prisons, college dormitories, and the military. The reader should keep in mind that the demographics to follow refer only to those who are aware of and admit to their hearing loss.
NCOA database. Since the MarkeTrak V survey7 had already identified close to 15,000 households with at least one person with a hearing loss, in November 1997, a sample of 3,000 individuals with a self-admitted hearing loss ages 50 and over—1,500 hearing aid owners and 1,500 non-owners—were randomly drawn from the identified panel of hearing-impaired households.
Utilizing information from previously developed industry surveys (ie, Knowles MarkeTrak and Knowles/Northwestern University Pygmalion survey), interviews with industry researchers and academia, and a review of the literature on the psychosocial and physical aspects of hearing loss, Seniors Research Group4 designed an 8-page legal-size questionnaire with 300+ questions for the individual with hearing loss and a 4-page legal-size questionnaire with 150 questions for the spouse or significant other of the identified respondent. The comprehensive survey covered a myriad of topics including: self and family assessment of hearing loss, psychological well-being, social impact of hearing loss, quality of relationships, life satisfaction, general health, self and family perceptions of benefit of hearing aid (users only), reasons for purchasing hearing aids (users only), reasons for not purchasing hearing aids (non-users only), and attitudes toward hearing health and hearing aids. Additionally, a number of personality scales deemed relevant to this study were included in the survey.8
The National Family Opinion Panel, which conducted the field work for this study in the spring and summer of 1998, sent respondents one questionnaire for themselves and one for their spouse, family member, or close friend to fill out. The hearing-impaired respondent who received the survey packet was asked to give the survey to the family member or friend of their choice who was most familiar with them. While hearing aid owner and non-owner samples were matched on important census demographics, NFO was unable to match them on severity of hearing loss, since hearing loss was not measured in the MarkeTrak screening panel survey.
Response rates were impressively high among both the hearing-impaired individuals and their family or friends, 79% and 71%, respectively (sample sizes of 2,364 and 2,132, respectively).
Seniors Research Group conducted extensive analysis but never published the study except in a press statement and summary on the NCOA Web site. A detailed executive summary of the findings was published in the January 2000 edition of HR by Kochkin and Rogin2 with the goal of focusing on the positive effects of hearing aids on quality of life.
Unidimensionality and reliability. In order for us to proceed with the demonstration of the validity and utility of the BHI Quick Hearing Check, it is important to show that the scale is measuring one underlying dimension and that the survey is reliable and has internal consistency.
The responses to the 15 items in the BHI Quick Hearing Check were subjected to a factor analysis for two very large samples: The MarkeTrak VIII subjects (n=7,201) and NCOA subjects (n=2,282). The factor analysis shown in Table 2 indicates there is one dimension to the items explaining 54% and 58% of variance, respectively. The correlations of each item with the underlying factor (subjective self-evaluation of hearing loss) ranged from a high of .84 to a low of .63. Being unidimensional with interval scaling, it is permissible to add the items to form a total score ranging from 0 to 60.
|TABLE 2. Factor analysis and reliability of BHI Quick Hearing Check survey.|
Reliability was measured using Cronbach’s coefficient alpha,9 in this case an estimate of reliability based on the average correlation among the 15 items within the BHI Quick Hearing Check. Coefficient alpha provides a good estimate of reliability (eg, test-retest correlations) in most situations since the major source of measurement error is because of the sampling of content. Coefficient alpha for the two samples is .94 (MarkeTrak) and .95 (NCOA).
The results of the factor analysis and the reliability analysis indicate that the items sampled to form the BHI Quick Hearing Check are very high in terms of internal consistency and that the content occupies a single domain of the human experience.
Objective validity. By objective validity, we mean that the BHI Quick Hearing Check is correlated with objective measures of hearing loss. First we will examine the relationship between the BHI Quick Hearing Check and objective measures of hearing loss, and then we will assess its utility.
Table 3 shows the Pearson correlations between the BHI Quick Hearing Check and total threshold averages (500, 1000, 2000, 3000, 4000 Hz), speech threshold averages (500, 1000, 2000 Hz), and high frequency threshold averages (2000, 3000, 4000 Hz) for both ears, the better ear, and the poorer ear. Correlations ranged from a low of r=.43 (speech threshold better ear) to a high of r=.55 (a five-frequency average, or 5PTA, for both ears). Speech recognition scores in the better ear showed a weaker correlation (r=-.33) with this measure of subjective hearing loss. This was the same finding as was reported in the smaller Koike study; the weaker correlation may be due to variability in tests used between clinics, as well as differences in test administration. There is also the possibility that speech recognition scores obtained in a binaural/bilateral manner may have yielded a high correlation.
|TABLE 3. Correlations between the BHI Quick Hearing Check and audiological measures (n=987).|
Next, the average scores at each of the 61 scale points of the BHI Quick Hearing Check were plotted and modeled. Figure 1 shows the relationship between speech threshold averages (best ear) and the BHI survey (R2=.82). Traditionally, in audiology, better-ear validity is reported; however, for this large sample, the mean threshold scores (5PTA) for both ears yielded higher predictability (Figure 2, R2=.84).
|FIGURE 1. Relationship between the BHI Quick Hearing Check and Average Speech Threshold Scores. Model = Better Ear.|
|FIGURE 2. Relationship between the BHI Quick Hearing Check and Average Threshold Scores. Model = 5PTA Both Ears.|
There is some variability in the data as shown by the spikes and valleys; for example, at scale point 6, the sample size is less than 5. The speech threshold average (better ear) model would appear to be slightly polynomial while the 5PTA scores (both ears) would appear to be linear in structure.
Objective utility. The predictive accuracy within BHI Quick Check scoring ranges are shown in Table 4 for poorer ear, better ear, and both ears. Contingency coefficients ranged from a low of .41 (better ear) to .47 (poorer and both ears); all models were significant at the P<.0001 level. In Figure 3, we have plotted the probability of a hearing loss of 40 dB or higher in both ears by level of BHI score (in 5-point increments). The underlying model is a second-order polynomial with an R2 of .98.
|TABLE 4. Predictive accuracy of BHI Quick Hearing Check with a 40 dB cut-off.|
|FIGURE 3. Probability of Hearing Loss of 40 dB (both ear average) or higher based on BHI Quick Hearing Check Scores.|
Subjective validity. We can now ask the question: “Is there a correlation between the BHI scale and other measures of subjective hearing loss and quality of life factors?”
In Figure 4, average BHI Quick Hearing Check scores are plotted against the Gallaudet scale.10 The Gallaudet scale is an 8-point scale in which respondents indicate whether they can understand speech under the following conditions: “whisper across a quiet room,” “normal voices across a quiet room,” “shouts across a quiet room,” “loud speech spoken into their better ear,” “not able to understand loud speech in their better ear,” as well as “tell noises from each other,” “hear loud noises at all,” and “hear any sound or any noise.” An individual’s score can range from 1 to 8. Typically, they are classified into one of five groups (1=hear whisper; 2=hear normal voice; 3=hear shouts; 4=hear speech in loud ear; 5=can’t hear speech). What makes the Gallaudet scale of particular value is it has been validated against clinical information (dB loss better ear). The Gallaudet scale has historically been used by the Centers for Disease Control and Prevention (CDC) in their quantification of the hearing-impaired population.
|FIGURE 4. Average BHI Quick Hearing Check Score by level of performance on the Gallaudet scale.|
The correlation between the Gallaudet scale (with scales points 4-8 collapsed) is .49. We did not find that the BHI scale was sensitive to levels of profound hearing loss as stated above.
In Figure 5, we have shown the average BHI Quick Hearing Check for both self-ratings (r=.50) and ratings of the spouse or significant other (r=.51) while assessing their family member’s hearing loss.
In Figure 6, the correlation between Plomp’s11 difficulty of hearing in noise scale was shown to be highly correlated (r=.64) with performance on the BHI Quick Hearing Check on more than 7,000 subjects from the MarkeTrak VIII database.
|FIGURE 5. Average BHI Quick Hearing Check Score by level of Self and Family Member perception of hearing loss.|
|FIGURE 6. Average BHI Quick Hearing Check Score by level of Plomp’s Difficulty of Hearing in Noise Scale.11|
From the NCOA quality of life database, a number of scales were correlated with the BHI Quick Hearing Check for both self-ratings of unaided subjects (n=994) and the spouse/significant others of these subjects (n=810). First, all scales from the NCOA database were standardized to a mean of 5 and standard deviation of 2 with scores taking the range of 1 to 9 (called stanines). In Table 5, the average standardized score by BHI score range (in increments of 10) is shown along with the ANOVA results (R2, correlation, and statistical significance level). Scores on the BHI Quick Hearing Check in the eyes of family members that were significantly correlated with quality of life factors in rank order are:
|TABLE 5. Average mean scores of quality of life issues (mean=5, std=2) by level of BHI Quick Hearing Check score. Subjects are unaided individuals with hearing loss from the NCOA database.|
- Concerns with safety;
- Family has to accommodate the individual’s hearing loss;
- The person with hearing loss experiences rejection socially;
- Tends to withdraw;
- Experiences great difficulty in communicating;
- Is less likely to be independent;
- Has a negative impact on the family’s quality of life;
- Tends to compensate for their hearing loss;
- Is perceived to have poor cognitive functioning by others;
- Lacks self-confidence;
- Is discriminated against by others;
- Displays anger and frustration;
- Shows signs of emotional instability and is introverted.
In the second half of Table 5, we have shown self-perceptions on quality of life factors and their correlations with performance on the BHI Quick Hearing Check. In rank order, higher scores on the BHI Quick Hearing Check are associated with:
- Higher levels of anger and frustration;
- Higher levels of withdrawal;
- Belief that hearing loss affects their health;
- Paranoid, phobia, anxiety, and depressive symptoms; and
- More problems in relationships.
Thus, hearing loss as measured by the BHI Quick Hearing Check is clearly associated with the reduction of social, emotional, psychological, and physical well-being of people with hearing loss and reductions in the quality of life of the family.
Norms and Predictions
With such a large database, we have published BHI Quick Hearing Check norms for more than 7,200 people in the United States from the latest MarkeTrak VIII database5 (2009) in Table 6. In addition, at each scale point, we have added estimated speech threshold scores in the better ear (Figure 1 model), threshold scores both ears (Figure 2 model), and the probability of a 40 dB hearing loss or worse in both ears (Figure 3 model) based on a sample size of 987 from our audiological database. Finally, we have documented and color coded perceptions of hearing loss for nearly 11,000 subjects (self and family member perceptions combined) and compared these findings to standard audiological classifications of hearing loss.12 Perceived hearing loss classifications at each scale point are based on the modal frequency of “mild,” “moderate,” “severe,” and “profound” classifications.
|TABLE 6. Norms (n=7,201) for the BHI Quick Hearing Check, predictions of hearing loss (n=987), and consumer/family perceptions of hearing loss (n=10,708) as a function of scores on the BHI Quick Hearing Check.|
The most striking thing about Table 6 is the disparity between self-perceptions and audiological classifications. For instance, the general public is more likely to begin classifying their hearing loss as “moderate” with speech threshold scores in their better ear at 31 dB while standard classifications rate hearing loss as mild up to 40 dB. The general public begins rating their hearing loss as “severe” with speech threshold scores in their better ear at 51 dB compared to 65 dB based on the audiological classification.
Utilizing both objective and subjective information across four databases involving nearly 11,000 subjects, the BHI Quick Hearing Check is shown to have validity and reliability:
- It is a unidimensional scale with very low content sampling error as well as face validity of items sampled. The items comprising this scale are common sense “signs of hearing loss.”
- Its reliability was established in two studies demonstrating very high internal consistency (Cronbach’s alpha = .94 and .95).
- It is correlated with puretone threshold scores, with correlations as high as .55 across 64 clinics and nearly 1,000 subjects. Correlations with speech recognition were lower (r=-.33).
- Models of puretone threshold averages at each BHI Quick Hearing Check scale point were as high as R2 = .84 (both ear average-5PTA).
- Predictive models of hearing losses of 40 dB or higher using the BHI Quick Hearing Check were very high (R2 = .98).
- In terms of concurrent validity, the BHI Quick Hearing Check is shown to be correlated with the Gallaudet scale (r=.49), Plomp’s scale of difficulty of hearing in noise (.64), self-perceptions of hearing loss (r=.50), and spousal perceptions of significant other’s hearing loss (r=.51).
- The BHI Quick Hearing Check is also highly correlated with a number of quality of life issues: social, emotional, cognitive, and physical well-being.
Given its objective and subjective validity and internal consistency, we have published representative norms for the United States on more than 7,000 people with hearing loss and integrated its known relationship with both objective and subjective information as supplied by the general public.
This study was made possible by a special grant from Knowles Electronics Inc, Itasca, Ill.
- Kochkin S. MarkeTrak VII: Obstacles to adult non-user adoption of hearing aids. Hear Jour. 2007;60(4):27-43.
- Kochkin S, Rogin C. Quantifying the obvious: the impact of hearing aids on quality of life. Hearing Review. 2000;7(1):8-34.
- Koike J, Hurst MK, Wetmore SJ. Correlation between the American Academy of Otolaryngology-Head & Neck Surgery (AAO-HNS) five minute hearing test and standard audiological data. Otolaryngol-Head Neck Surg. 1985;111(5):625-632.
- National Council on Aging (NCOA). The Impact of Untreated Hearing Loss in Older Americans. Conducted by the Seniors Research Group. Supported through a grant from the Hearing Industries Association. Preliminary report, December 28, 1998.
- Kochkin S. MarkeTrak VIII: 25 year trends in the hearing health market. Hearing Review. 2009;16(11):12-31. Accessed October 20, 2010.
- Better Hearing Institute. MarkeTrak VIII surveys from the NFO panel. Available at: www.betterhearing.org/professionals/MarkeTrak_8_survey.zip. Accessed October 20, 2010.
- Kochkin S. MarkeTrak V: “Baby Boomers” spur growth in potential market, but penetration rate declines. Hear Jour. 1999;52(1):33-48.
- Better Hearing Institute. Surveys used in the NCOA study in 1998-1999 and published in the January 2000 Hearing Review. Available at: www.betterhearing.org/professionals/tools.cfm. Accessed October 20, 2010.
- Nunnally JC. Psychometric Theory. New York: McGraw Hill Book Co; 1967.
- Schein JD, Gentile A, Haase KW. Development and evaluation of an expanded hearing loss scale questionnaire. National Center for Health Statistics. Vital Health Stat 2. 1970 Apr;(37):1-42.
- Plomp R. Auditory handicap of hearing impairment and the limited benefit of hearing aids. J Acoust Soc Am. 1978;63:533-549.
- Roeser RJ. Audiology Desk Reference. New York: Thieme Medical Publishers; 1996:167.
Correspondence can be addressed to Sergei Kochkin, PhD, at .