Article Access Statistics   Viewed  4211   Printed  163   Emailed  3   PDF Downloaded  18   Comments  [Add]   Cited by others  2  




Year : 2012
 Volume
: 14  Issue : 56  Page
: 1320 

Validity of hearing impairment calculation methods for prediction of selfreported hearing handicap 

Andrew B John^{1}, Brian M Kreisman^{2}, Stephen Pallett^{3}
^{1} Department of Communication Sciences and Disorders, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma, USA ^{2} Department of Audiology, SpeechLanguage Pathology and Deaf Studies, Towson University, Towson, Maryland, USA ^{3} Department of Audiology, SpeechLanguage Pathology and Deaf Studies, Towson University, Towson, Maryland; ENTAA Care, Annapolis, Maryland, USA
Click here for correspondence address
and email
Date of Web Publication  29Feb2012 




Worker's compensation for hearing loss caused by occupational noise exposure is calculated by varying methods, from state to state within the United States (US), with many employing arithmetic formulas based on the puretone audiogram, to quantify hearing loss. Several assumptions unsupported or weakly supported by empirical data underlie these formulas. The present study evaluated the ability of various arithmetic hearing impairment calculations to predict a selfreported hearing handicap in a sample of presenting with sensorineural hearing loss. 204 adults (127 male, 77 female) ranging in age from 18 to 94 served as participants. The sample was selected to exclude patients who had been referred for hearing testing for a medicolegal examination or a hearing conservation appointment. A hearing handicap was measured by the Hearing Handicap Inventory for Adults/for the Elderly (HHIA/E). The covariance analysis of linear structural equations was used to assess the relative strength of correlation with the HHIA/E score among the six formulas and various forms of puretone average. The results revealed that all the hearing impairment calculations examined were significantly, but weakly, correlated with the selfreported hearing impairment scores. No significant differences among the predictive abilities of the impairment calculations were evident; however, the average binaural impairment assigned differed significantly among the six calculations examined. Individuals who demonstrated 0% impairment had significantly lower (i.e., better) HHIA/E scores compared to those with nonzero impairment for each formula. These results supported the idea that audiometric data provided an insufficient explanation for realworld hearing difficulties. Keywords: Hearing handicap, hearing impairment, sensorineural hearing loss, worker′s compensation
How to cite this article: John AB, Kreisman BM, Pallett S. Validity of hearing impairment calculation methods for prediction of selfreported hearing handicap. Noise Health 2012;14:1320 
How to cite this URL: John AB, Kreisman BM, Pallett S. Validity of hearing impairment calculation methods for prediction of selfreported hearing handicap. Noise Health [serial online] 2012 [cited 2015 Mar 4];14:1320. Available from: http://www.noiseandhealth.org/text.asp?2012/14/56/13/93321 
Introduction   
Over the last several decades, a number of formulas have been proposed within the US to quantify adultonset hearing loss, using audiometric data. These include calculations developed by the American Medical Association, Council on Physical Medicine; ^{[1]} the American Academy of Ophthalmology, and Otolaryngology, Committee on Conservation of Hearing; ^{[2]} the National Institute for Occupational Safety and Health; ^{[3]} the National Academy of the Science Committee on Hearing, Bioacoustics, and Biomechanics; ^{[4]} the American Medical Association/American Academy of Otolaryngology, Committee on Hearing and Equilibrium; ^{[5]} and the American SpeechLanguageHearing Association, Task Force, on the definition of a hearing handicap, duplicated by the 1997 formula revision by the National Institute for Occupational Safety and Health. ^{[6],[7]} These formulas vary in three main dimensions: (1) audiometric frequencies included in the calculation; (2) low/high fence, from which percentageimpairmentperdB calculation is derived; and (3) better ear/worse ear weighting for the calculation of binaural impairment, or %BI [Table 1].
High fence refers to the amount of hearing loss in the decibel hearing level (dB HL) that constitutes a maximum (100%) monaural impairment for everyday speech understanding; additional losses in dB HL above the high fence do not accrue an additional handicap. Conversely, low fence refers to the maximum amount of hearing loss in dB HL that is considered to contribute no handicap (0%) to the understanding of speech. All thresholds below the low fence are considered to contribute a 0% handicap. Binaural impairment (%BI) refers to the total percentage handicap to speech understanding across both ears realized by an individual with a measurable hearing loss. All formulas evaluated here calculate %BI using unequal weighting of the two monaural impairment figures calculated for an individual, with a bias (most commonly five to one) toward the better ear. To be precise, the better ear monaural impairment is multiplied by five, added to the worse ear impairment, and the sum is divided by six to produce %BI. Notably, while a fivetoone weighting is common among hearing impairment calculations, there is no research basis for this particular proportion. ^{[8]}
The motivation for development of mathematical calculations of workplace hearing loss in the US dates back to the early twentieth century. In 1908, the Federal Employees' Compensation Act (FECA) provided America's first federallymandated worker's insurance for occupational injury or disease. Among the conditions considered by the FECA to be a workplace injury was traumatic loss of hearing, such as tympanic membrane perforation following an explosion or head injury. ^{[9]} However, for the next 40 years, nontraumatic hearing loss due to workplace noise exposure was not compensable under United State's worker's compensation laws. In 1948, the New York Supreme Court ruled in Slawinski versus J.H. Williams and Co. that partial loss of hearing was eligible for compensation even if no wage loss could be demonstrated. That ruling led to the development of occupational hearing loss statutes, to assist in the determination of the direct effect of a hearing loss during workplace production. Later rulings by courts in Wisconsin, Maryland, and Georgia, as well as the Federal Judiciary, during the 1950s and 1960s, refined and upheld the concept of hearing loss as a compensable occupational injury even without demonstrable wage loss. ^{[8]}
Currently, all fifty states in the US and the federal government compensate workers for occupational hearing loss, either through specific hearing loss compensation statutes or the general worker's compensation law. Methods for calculating hearing impairment for the purpose of worker's compensation vary from state to state. According to recent data from the United States Department of Labor's Office on Worker's Compensation Programs, four different arithmetic calculations of hearing impairment are currently in use by state worker's compensation offices [Table 1]. In addition, 22 states report the primary use of medical evidence as a determinant of award. The state of Montana currently does not have a statute in place specifically covering compensation for hearing loss, instead it treats it under statutes covering the general occupational injury or disease. ^{[10]}
Audiometricallyderived measures of hearing impairment such as those used for Worker's Compensation awards are likely to be insensitive to individual variability in the impact of hearing loss on daily functioning. Indeed, selfreport measures of hearing handicap consistently show low correlations with audiometric sensitivity. ^{[11],[12],[13],[14],[15]} The relation between hearing sensitivity and a selfreported hearing handicap is believed to be influenced by a number of nonauditory factors, including, age, gender, personality, and socioeconomic status. ^{[16],[17],[18],[19],[20],[21]} For example, Lutman and colleagues ^{[16]} noted that age, gender, and socioeconomic status are factors in hearing selfreport, with males reporting more disability than females and older individuals reporting less disability than younger individuals. In another study, Lutman ^{[17]} concluded that older individuals tend to overrate disability, consistent with the findings of GordonSalant and colleagues, ^{[19]} who demonstrated that increased age correlated with increased disability as reported on the Hearing Handicap Inventory for the Elderly (HHIE). Gatehouse ^{[21]} also found substantial variability in the correlation between selfreported hearing handicap and hearing loss, with significant effects of age and personality.
Furthermore, several studies have noted that the specific methodology used in most audiometricallybased measures of hearing impairment limit their correlation to speech recognition. In one such investigation, Suter ^{[22]} evaluated the AAOO1959 calculation and found that the failure of that formula to include thresholds above 2000 Hz, made it insensitive to sloping hearing losses that caused significant deterioration of speech understanding, both in quiet and with noise. This is not surprising given the critical importance of the highfrequency spectrum, in the recognition of speech sounds. ^{[23],[24]}
A subjectivelyreported hearing handicap, measured using instruments such as the Hearing Handicap Inventory for Adults (HHIA; ^{[25]} ) or for the Elderly (HHIE; ^{[26]} ), may help to determine the social and communicative impact of hearing loss on daily functioning. However, the use of subjective measures of hearing impairment is clearly contraindicated for patients in cases with financial stakes, such as worker's compensation claims. Even as audiometric sensitivity measures have little face validity for the explanation of everyday listening performance and difficulty, they contain elements critical to medicolegal evaluation: They are objective, reproducible, and if performed correctly, resistant to exaggeration and opinion. ^{[27]}
Schow and Gatehouse ^{[28]} have suggested that objective hearing impairment measures may be validated by studies comparing the selfreport of hearing handicap in individuals who are not seeking financial or other compensation. In one such study, Brainerd and Frankel ^{[11]} evaluated the correlation among six hearing impairment calculations, the Denver Scale of Communication Function, and the English version of the Social Hearing Handicap Index. The authors determined that no audiometricallyderived formula was a strong predictor of selfreported hearing function. In fact, the better ear puretone average (500, 1 k, and 2 kHz) was a stronger predictor of selfreported handicap than any formula examined. In a second study, Stewart and colleagues ^{[29]} evaluated four audiometric calculations for hearing impairment, as predictors of performance, on the screening version of the HHIA in a group of firearm users. The correlations ranged from r=.55 to r=.66, with the highest correlation seen in the formulas omitting the 500 Hz threshold as a formula element.
Few studies have found evidence for any of the several arithmetic hearing loss calculations in current or recent use in the US, as an effective measure of realworld hearing difficulty. More significantly, a literature review was unable to identify any study that has used appropriate statistical methods to evaluate the relative strength of association between these hearing impairment calculations and selfreport measures. Statetostate inconsistency in the technique used to quantify compensable hearing loss suggests the need for an evidence basis in an impairment calculation. In particular there is a need to determine whether the differences in formula predictive power, in the studies cited herewith and others, are statistically significant. If so, this might suggest a real benefit for the use of one formula (or more) over others currently in use, for appropriately evaluating hearing loss as a limitation on daily activity. With these considerations in mind, we have examined the relative predictive value of seven published hearing impairment formulas for a commonlyemployed selfreport of hearing difficulty, the Hearing Handicap Inventory, in a compensationineligible clinical population.
Methods
Participants
Data were collected from adult patients seen at the speech and hearing clinics of the University of Florida, Towson University, and the ENTAA Care of Annapolis, Maryland, between 1997 and 2009. All participants were diagnosed with adultonset sensorineural hearing loss (SNHL) and were administered the Hearing Handicap Inventory for Adults (HHIA) or Hearing Handicap Inventory for the Elderly (HHIE) at the diagnosis appointment. Patients presenting with conductive or fluctuating hearing loss were excluded from the analysis. Consistent with the recommendations of Schow and Gatehouse, ^{[28]} patients who had been referred for hearing testing for a medicolegal examination or hearing conservation appointment were not recruited. In order to minimize the varying effect of amplification use on the HHIA/E score, individuals using hearing aids or other assistive listening devices at the time of testing were excluded. A total of 204 participants (127 male, 77 female) met the criteria for inclusion. Subjects ranged from 18 to 94 years in age, with a mean of 65.6 years (SD=13.8). See [Figure 1] for a summary audiogram of this sample. For the purpose of analysis, thresholds marked as 'no response' were coded at 5 dB above the limits of the test equipment (typically 100  115 dBHL).  Figure 1: Mean audiogram (n=204). Bars indicate one standard deviation. Note that a standard octaveinterval audiogram format has been expanded to include thresholds at the interoctave frequencies of 3000 and 6000 Hz, as thresholds at thesefrequenciesare considered in certain impairment calculations
Click here to view 
Hearing impairment calculations
Six different arithmetic calculations of hearing impairment were included in our analysis. Four were selected on the basis of the current or recent (since 2006) use in worker's compensation award calculation in the US. These four calculations were designated AAOO1959, CHABA1975, AMA1979, and Oregon, referring to a fourth formula, unique to the Oregon statutes. ^{[10]} In addition, we evaluated two versions of a hearing impairment calculation developed by NIOSH: NIOSH1972 and NIOSH1997 [Table 1]. In light of the findings of Brainerd and Frankel, ^{[11]} on the strength of the pure tone average (PTA), as a predictor of hearing impairment, we also evaluated the simple averages of monaural hearing thresholds at (1) 500, 1 k, and 2 kHz, (2) 1 k, 2 k, and 4 kHz, (3) 500, 1 k, 2 k, and 3 kHz, and (4) 500, 1 k, 2 k, and 4 kHz. Both betterear and worseear PTA were included in the analysis.  Table1: Worker's compensation hearing loss statutes by state. Low fence refers to the average threshold below which 0% impairment is assigned. High fence refers to the average threshold above which 100% impairment is assigned. Weighting refers to relative contributions of better ear impairment and worse ear impairment to binaural impairment (i.e., 5:1 indicates that the better ear is weighted five times as much as the worse ear). This weighting is applied after calculation of the monaural impairment, for each ear. Source: Department of Labor (2006)
Click here to view 
Selfreported hearing impairment
All participants were administered the HHIA or HHIE at the appointment during which the hearing thresholds were obtained. The Hearing Handicap Inventories were the widelyused clinical measures of communicative function, containing a series of questions that attempted to quantify the social and emotional impact of hearing loss on daily communication and functioning. The HHIA and HHIE are similar, differing in three questions (out of 25) to reflect situations appropriate to individuals under age 65 (HHIA) or 65 and older (HHIE). For the purpose of analysis, we collapsed the scores across the two versions of the HHI.
Statistical analysis
A repeatedmeasures ANOVA was conducted, to determine whether the significant differences in the average percent binaural impairment (%BI) produced by the six calculations were evident. Post hoc paired comparisons were conducted using a Bonferroni correction for paired samples, to determine which calculations differed significantly in the assigned %BI. The Pearson product moment correlations were calculated between the HHIA/E score and %BI, calculated using the six formulas mentioned earlier, and the four variations of a puretone average, in the worse and better ears. As the hearing impairment calculations were used principally in the worker's compensation claims, these analyses were repeated for the subgroup of participants under age 65 (n=91).
A comparison of the strength of paired correlations among each of the impairment formulas and the HHIA/E score was conducted using a familywise covariance analysis of linear structural equations (CALIS) in an SAS software, as described by Cheung and Chan. ^{[30]} The CALIS technique allowed for significance testing among the bivariate correlations between multiple dependent predictor variables (%BI values for the six impairment calculations, derived from the same audiometric data) and the one criterion variable (HHI score). Post hoc testing was conducted using a Holm correction. ^{[31]}
Results   
Average %BI as well as the maximum and minimum %BI for the six hearing impairment calculations can be seen in [Table 2]. Results of a repeatedmeasures ANOVA for the effect of calculation suggested that %BI was significantly different among the calculations, F(5,1015)=358.75, P<.001. After applying a Bonferroni correction (.05/6=.008), all paired differences among %BI of the six calculations were significant (P<.001). That is, the six calculations could be ranked discretely from the greatest to the least average %BI as follows: NIOSH1997 (40.3%), Oregon, NIOSH1972, AMA1979, CHABA1975, AAOO1959 (15.8%) [Table 2].  Table 2: Mean percentage of binaural impairment (%BI) by formula, range of %BI, and correlation between %BI and HHIA/E total score (rHHI) (BE=better ear; WE=worse ear)
Click here to view 
A repeatedmeasures ANOVA for the effect of calculation on the subgroup of participants under age 65 also found a significant difference, F(5, 205)=86.26, P<.001. After a Bonferroni correction, all paired differences were significant (P<.001) except for the comparison between AAOO1959 and CHABA1975 (P=.03). For this subgroup, the six calculations could be ranked from the greatest to the least average %BI, as follows: NIOSH1997 (35.7%), Oregon, NIOSH1972, AMA1979, CHABA1975 (17.5%)/AAOO1959 (15.7%).
Scores on the HHIA/E showed considerable variance, ranging from zero to 96 with a mean of 37.1 and standard deviation of 23.1. No significant effect of age (P=.19) or gender (P=.89) was seen for the HHIA/E score. No significant difference in inventory score was seen between subjects who were administered the HHIE (age 65+, n=113, mean=35.7, SD=23.0) and those subjects who were administered the HHIA (age 1864, n=91, mean=38.8, SD=23.2, P=.963).
Pearson product moment correlations between the HHIA/E score and %BI, calculated by using the six formulas mentioned earlier and the four variations of a puretone average in the worse and better ears can be seen in [Table 2]. These coefficients were denoted as r_{HHI} for each impairment calculation. Correlation coefficients within the whole sample ranged from .462 to .502 for the impairment formulas and from .480 to .521 for the puretone average variations. All correlations were highly significant (P<.001). These findings indicate that the impairment calculations predicted approximately 21% to 25% of the variance in the HHI score, while various forms of the PTA in the better and worse ear predicted approximately 23% to 27% of the variance.
Correlation coefficients for HHIA/E score and %BI in the subgroup of participants under age 65 ranged from .492 to .551 for the impairment formulas and from .548 to .596 for the puretone average variations. All correlations were highly significant (P<.001). These findings indicated that, for the subgroup of participants under age 65, the impairment calculations predicted approximately 24 to 30% of the variance in the HHI score, while various forms of the PTA in the better and worse ear predicted approximately 30 to 36% of the variance.
Next, a comparison of the strength of the paired correlations among each of the impairment formulas and HHIA/E score was conducted using CALIS, as described earlier. After applying a Holm correction for multiple comparisons to the P values generated, no significant differences among the predictive abilities of the impairment calculations were evident.
Finally, the likelihood that each of the six impairment calculations would return a compensable (nonzero) impairment for the individuals in the sample under study was calculated. Individuals were most likely to demonstrate impairment greater than zero under the Oregon calculation (185/204 impaired, or 90.7% in the full sample, 74/91 impaired or 81.3% in the subsample under age 65), and least likely under the AAOO1959 calculation (147/204 impaired, or 72.1% in the full sample; 62/91 impaired or 68.1% in the subsample). As expected, for each of the six impairment calculations, individuals who demonstrated 0% impairment had significantly lower HHIA/E scores (P<.001) compared to those with nonzero impairment in both the full sample and subsample under age 65 [Table 3].  Table 3: Number of individuals classified as having zero impairment (%BI=0) versus nonzero binaural impairment (%BI>0) as determined by six arithmetic calculations with mean HHIA/E scores and independent samples ttest results
Click here to view 
Large differences in calculated %BI were evident among many individuals in the sample, particularly those with steeply sloping hearing losses above 20003000 Hz. Calculated impairment and HHIA/E score from the selected individuals with widely varying %BI, depending on the calculation employed, are shown in [Table 4]. These data are presented to illustrate the high variance in %BI within individuals as well as the wide range in HHIA/E score across individuals with similar %BI.  Table 4: Selected participants with betterear (BE) and worseear puretone averages (PTA), percent binaural impairment as calculated by six impairment calculations, and HHIA/E score
Click here to view 
Discussion   
The present study evaluated differences among the relations between six arithmetic calculations of hearing impairment and the score on a common selfreport measure of hearing, the HHIA or HHIE. All the impairment calculations examined were derived from the audiometric puretone threshold data, but differed in the audiometric frequencies included in the calculation, low/high fence, from which the percentageimpairmentperdB calculation was derived, as also better ear/worse ear weighting. The goals of the analysis were, (1) to determine statistically whether one or more of the calculations was significantly better in predicting selfreported hearing difficulty than others, and (2) to evaluate whether these calculations differed in the likelihood of assigning nonzero impairment to an individual with SNHL and the percentage of impairment assigned.
The results of this analysis suggest that selfreported hearing handicap as measured by the HHIA and HHIE can be estimated by calculations based on puretone thresholds. However, the statistical relations found were weak overall, ranging from r=.46 to r=.50 for the six impairment calculations and from r=.48 to r=.52 for forms of simple monaural puretone average. In the subsample of participants under age 65, these values ranged from r=.49 to r=.55 for the six impairment calculations and from r=.55 to r=.60 for forms of simple monaural puretone average. The strongest predictor of a handicap was the worseear standard PTA (500, 1000, and 2000 Hz), which had 27.1% shared variance with the HHIA/E score, in the full sample and the betterear standard PTA, which had 35.5% shared variance with HHIA/E score, in the subsample. Our results agreed in part with Brainerd and Frankel's finding that betterear PTA was a stronger predictor of the standard summarized handicap (r=.38) than was any impairment calculation studied (1985). It should be noted that Brainerd and Frankel did not include isolated worseear PTAs in their analysis. The current findings suggest that greater attention to worseear status may improve the understanding of the handicap in individuals with asymmetric hearing impairment.
Of the calculations currently in used in the US for Worker's Compensation claims, the strongest predictive value was seen for the AMA1979 formula (r=.50 in the full sample and r=.55 in the subsample). The weakest predictor was the CHABA1975 formula (r=.46 in the full sample and r=.49 in the subsample). Correlations were significant for all the formulas examined, with these impairment calculations accounting for between 21% and 25% of the variance in HHIA/E score in the full sample and between 24 and 30% of the variance in the subsample. These correlation coefficients were lower than those reported by Stewart and colleagues, ^{[29]} who reported that four audiometric formulas were more highly correlated with the HHIA screener score (r=.55 to .66) between four audiometric formulas and the screening HHIA. However, the present data are more similar to values reported by Stewart in a subset of only those individuals with a nonzero hearing handicap. For this set, the r values for the HHIA screener varied between .45 and .59 among the formulas examined. The finding that impairment of the subgroup of younger participants was better predicted by the calculations studied than the sample as a whole can likely be attributed to the greater degree of highfrequency hearing loss (to which the formulas are insensitive) resulting from presbyacusis in participants over 65.
A familywise comparison of the strength of the correlations among the impairment formulas and HHI score found no significant differences among the calculations, suggesting that all six calculations performed equally well (or equally poorly) in predicting the HHIA/E score. This finding was the most compelling of the present study. Previously, investigators had conducted informal comparisons of these impairment calculations, but the fact that the calculations were structurally correlated (incorporating similar or identical elements) made comparisons of the strength of correlation problematic. The CALIS technique described here accounts for these structural similarities and allows for a comparison of correlation strength with the HHIA/E score among the calculations under study.
Large differences were evident in both the number of individuals in the sample classified as impaired by the six calculations and in the degree of impairment assigned by the calculations. In particular, individuals with steeplysloping hearing losses above 2000  3000 Hz demonstrated much greater impairment as calculated by the NIOSH1997 formula, which incorporated thresholds at 3000 and 4000 Hz, and the Oregon formula, which incorporated thresholds at 3000, 4000, and 6000 Hz.
Several conclusions can be drawn from these data. First, the four formulas currently in use in Worker's Compensation statutes in the US have performed equally well. No strong evidence can be found in the present analysis to support the use of one of these impairment calculations over another. Notably, each of the calculations examined predicted the HHIA/E score almost as well as the simple puretone average in the individual's better ear.
However, a review of the average %BI for each of the six formulas [Table 2] showed that while these formulas may have performed equally well in predicting the handicap, they differed greatly in the binaural impairment percentage assigned. The most 'conservative' of these formulas was the AAOO1959 formula, which found an average of 15.8% binaural impairment in the full sample and 15.7% in the subsample. The NIOSH1997 formula, by contrast, produced an average of 40.3% impairment in the full sample and 35.7% in the subsample. Although the mean %BI differed significantly among each of the six calculations examined, the NIOSH1997 formula produced considerably higher impairment than the other five, for most of the participant samples. Two reasons for this finding were evident. First, the NIOSH1997 formula was one of the two (along with Oregon) to consider thresholds above 3000 Hz when calculating an impairment, increasing the sensitivity of the calculation to the highfrequencybiased hearing losses consistent with noise exposure and presbycusis. Second, and probably more importantly, the NIOSH1997 formula employed a 75 dB HL 'high fence' compared to a 92 dB HL fence for the other five formulas. This had the effect of increasing the percentage impairment per dB to 2% (compared to 1.51.8) and to assign 100% impairment to any individual ear with an average threshold of above 75 dB HL in the range of 1000  4000 Hz.
The six impairment calculations examined here also differed considerably in the number of subjects in the sample who were assigned a compensable hearing impairment (%BI>0). Of the 204 participants, 147 (72.1%) would be considered as having a hearing impairment under the AAOO1959 calculation, 148 (72.5%) under CHABA1975, 172 (84.3%) under NIOSH1972, 163 (79.9%) under AMA1979, 181 (88.7%) under NIOSH1997, and 185 (90.7%) under the Oregon state calculation. In the subsample of 91 adults under age 65, 62 (68.1%) would be considered to have a hearing impairment under the AAOO1959 calculation, 62 (68.1%) under CHABA1975, 67 (73.6%) under AMA1979, 70 (76.9%) under NIOSH1972, 73 (80.2%) under NIOSH1997, and 74 (81.3%) under the Oregon state calculation. By contrast, Brainerd and Frankel (1985) found that 27% of their sample was impaired under the AAOO1959 calculation, 30% under AMA1979, 18% under CHABA1975, and 40% under NIOSH1997, suggesting that the present sample had significantly greater hearing loss overall than in the previous study.
Many of these betweenformula differences were substantial at the individual level as well. [Table 4] displays data from selected individuals with widely varying impairment depending on the calculation employed. Also notable here is the wide variance in the HHIA/E score among the individuals in [Table 4], illustrating the low correlation between arithmeticallyderived impairment and selfreported hearing handicap. Of the six calculations examined, only the ASHA1981 formula produced a 100% binaural impairment score, and did so for eleven individuals in the sample (5.3%). In the other six calculations, binaural impairment scores for those individuals ranged from 27.5 to 96.7%.
Ward ^{[32]} noted that each of the hearing impairment formulas in current use, in the US, is predicated on a series of arbitrary assumptions; among them are,
 that some combination of threshold values at specific frequencies are related monotonically to the intelligibility of (some form of) speech in (some form of) noise
 that the rate of change in impairment is linear, and equals 1.5, 1.8 or 2% per dB, and rises between the absolute cutoffs of 25 or 35 dB at the low end and 75 or 92 dB at the high end
 that each frequency employed is equally important, such that a loss above the low fence at one frequency can be offset by a threshold below the low fence at another frequency
 that betterear thresholds contribute precisely five or seven times as much to overall impairment as do worseear thresholds.
There is little empirical data to support these assumptions. The US state and federal organizations who promulgate the calculations resulting from them offer little or no evidence basis for the use of the six formulas considered in the present study. Some criteria appear to have a pragmatic basis. For example, the apparently arbitrary choice of a 92dB HL high fence for maximum impairment is probably a result of desire for an easy byhand calculation of impairment (1.5% per dB) of above 25 dB HL, a reasonable low fence for compensable hearing loss. The authors are currently conducting a study to statistically evaluate the assumptions identified by Ward, and also to study whether the predictive value of impairment calculations for hearing and communication difficulty might be improved by alterations in the formula structure.
It should be noted that the hearing impairment formulas examined here are used primarily to quantify hearing loss in individuals exposed to occupational noise rather than presbyacusis, the predominant etiology of hearing loss for at least some of our sample. Noise exposure alone results in a characteristic pattern of binaural hearing loss with a significant 'notch' around 4000  6000 Hz. The sample used in this study includes individuals with a wide variation in the degree and configuration of hearing loss. It is possible that a sample more representative of the noiseexposed workforce might demonstrate different relations between audiometric sensitivity and selfreported hearing handicap. In particular, a more homogenous sample with this characteristic noise notch might produce different results when analyzing the role of midhigh frequency thresholds in the prediction of the hearing handicap score.
However, the aim of this investigation was to make comparisons between audiometricallyderived impairment and a selfreported handicap, in a population with no motivation to exaggerate responses on the selfreport inventory. Data collection using these methods with noiseexposed workers is difficult to accomplish without introducing some bias. Participant files have been selected to exclude individuals seen for hearing conservation audiograms, Worker's Compensation evaluations, or any other medicolegal appointment. Not surprisingly, this process has resulted in an older sample whose hearing loss was not limited to noise exposure. Analysis of the subsample of participants under age 65 did reveal somewhat stronger correlations between the impairment calculations and the HHIA/E score than the sample at large; however, these correlations were still low.
It is likely that the low correlation between the objective and subjective measures of the hearing handicap seen in this study and others is related to nonauditory factors. Several studies have noted that the relation between hearing sensitivity and selfreported handicap is related to individual characteristics including age, gender, and personality, ^{[16],[17],[18],[19],[20],[21]} which may have affected the results of the present study. Considering the substantial literature demonstrating that hearing handicap by selfreport is influenced by factors outside of the purview of the audiologist or physician, it is likely that any audiometricallyderived calculations would have only a limited correlation with the scores on an instrument such as the HHIA or HHIE.
It should also be noted that the present study is limited to calculations used to estimate the hearing handicap primarily in the US. Methods for quantifying hearing loss as a compensable injury vary substantially across nations, although many do employ arithmetic formulas similar to those described here. Even as the present study is limited to a sample of American adults and compares across calculations presently or recently in use within the US, examination of differences across nations may be warranted.
Conclusions   
The principal finding of the present study is that arithmetic hearing impairment calculations currently in use predict selfreported hearing loss in a population of individuals with SNHL significantly, but weakly. No formula was able to account for more than about 25% of the variance in the HHIA/E score in the full sample or 30% in the younger subsample, and none was a more effective predictor than a simple threefrequency, puretone average in the better or worse ear. A CALIS technique revealed no significant differences in the correlations between the HHIA/E score and the six impairment calculations. Further research into the theoretical basis of hearing impairment calculations is needed, as few empirical data are available to support the methods employed for the allocation of impairment and for financial awards for hearing loss suffered as a result of workplace noise exposure.
Acknowledgments   
The authors acknowledge the assistance of Valerie Skaggs, PhD, for her assistance in the statistical analysis. Portions of this article were presented at the Seventeenth Annual American Academy of Audiology Convention, Washington, DC, April, 2005, and at the New Zealand Audiological Society Conference, Christchurch, NZ, June, 2005.
References   
1.  American Medical Association (AMA), Council on Physical Medicine. Tentative standard procedure for evaluating the percentage loss of hearing in medicolegal cases. J Am Med Assoc 1947;133:39697. 
2.  American Academy of Ophthalmology and Otolaryngology (AAOO), Committee on Conservation of Hearing. Guide for the evaluation of hearing impairment. Trans Amer Acad Acad Ophthalmol Otolaryngol 1959;63:2368. 
3.  National Institute for Occupational Safety and Health (NIOSH). Publication no. 7311001: NIOSH criteria for a recommended standard: Occupational exposure to noise. Cincinnati, OH: U.S. Department of Health, Education, and Welfare, Health Services and Mental Health Administration, National Institute for Occupational Safety and Health; 1972. 
4.  Committee on Hearing, Bioacoustics, and Biomechanics (CHABA). Compensation formula for hearing loss. Report of Working Group 77. Washington, DC: National Academy of Science; 1975 
5.  Catlin FI. Guide for the evaluation of hearing handicap. American Academy of Otolaryngology Committee on Hearing and Equilibrium. Otolaryngol Clin North Am 1979;12:65563. 
6.  American SpeechLanguageHearing Association (ASHA). On the definition of hearing handicap. ASHA 1981;23:29397. 
7.  National Institute for Occupational Safety and Health (NIOSH). Publication no. 98126: NIOSH criteria for a recommended standard: Occupational noise exposure. Cincinnati, OH: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health; 1997. 
8.  Dobie RA, Megerson SC. Worker's compensation. In Berger EH, Royster LH, Royster JD, Driscoll DP, Layne M, editors. The Noise Manual. 5 ^{th} ed. Fairfax, VA: American Industrial Hygiene Association; 2000. p. 689710. 
9.  Melnick W, Morgan W. Hearing compensation evaluation. Otolaryngol Clin North Am 1991;24:391402. 
10.  Department of Labor. State Worker's Compensation Laws, Table 20: Occupational Hearing Loss Statutes. Washington, DC: US Department of Labor, Employment Standards Administration, Office of Worker's Compensation Programs, Branch of Planning, Policy, and Review; 2006. 
11.  Brainerd SH, Frankel BG. The relationship between audiometric and selfreport measures of hearing handicap. Ear Hear 1985;6:8992. 
12.  Matthews LJ, Lee FS, Mills JH, Schum DJ. Audiometric and subjective assessment of hearing handicap. Arch Otolaryngol Head Neck Surg 1990;116:132530. 
13.  McKenna, L. Some psychological aspects of deafness. In: Ballantyne J, Martin MC, Martin A, editors. Deafness. 5 ^{th} ed. London: Whurr; 1993. p. 2378. 
14.  Newman CW, Jacobson GP, Hug GA, Sandridge SA. Perceived hearing handicap of patients with unilateral or mild hearing loss. Ann Oto Rhinol Laryngol 1997;106:21014. 
15.  Stewart M, Scherer J, Lehman ME. Perceived effects of high frequency hearing loss in a farming population. J Am Acad Audiol 2003;14:1008. 
16.  Lutman ME, Brown EJ, Coles RR. Selfreported disability and handicap in the population in relation to puretone threshold, age, sex and type of hearing loss. Br J Audiol 1987;21:4558. 
17.  Lutman ME. Hearing disability in the elderly. Acta Otolaryngol Suppl 1990;476:23948. 
18.  Stephens D, Hétu R. Impairment, disability, and handicap in audiology: Towards a consensus. Audiology 1991;30:185200. 
19.  GordonSalant S, Lantz J, Fitzgibbons P. Age effects on measures of hearing disability. Ear Hear 1994;15:2625. 
20.  Gatehouse S. Components and determinants of hearing aid benefit. Ear Hear 1994;15:3049. 
21.  Gatehouse S. Speech tests as measures of outcome. Scand Audiol Suppl 1998;49:5460. 
22.  Suter AH. Speech recognition in noise by individuals with mild hearing impairments. J Acoust Soc Am 1985;78:887900. 
23.  French NR, Steinberg JC. Factors governing the intelligibility of speech sounds. J Acoust Soc Am 1947;22:89151. 
24.  Humes LE. Understanding the speechunderstanding problems of the hearing impaired. J Am Acad Audiol 1991;2:5969. 
25.  Newman CW, Weinstein BE, Jacobson GP, Hug GA. The Hearing Handicap Inventory for Adults: Psychometric adequacy and audiometric correlates. Ear Hear 1990;11:4303. 
26.  Ventry IM, Weinstein BE. The hearing handicap inventory for the elderly: A new tool. Ear Hear 1982;3:12834. 
27.  Lutman ME, Robinson DW. Quantification of hearing disability for medicolegal purposes based on selfrating. Br J Audiol 1992;26:297306. 
28.  Schow RL, Gatehouse S. Fundamental issues in selfassessment of hearing. Ear Hear 1990;11(5 Suppl):6S16S. 
29.  Stewart M, Pankiw R, Lehman ME, Simpson TH. Hearing loss and hearing handicap in users of recreational firearms. J Am Acad Audiol 2002;13:1608. 
30.  Chueng MWL, Chan W. Testing dependent correlation coefficients via structural equation modeling. Organ Res Methods 2004;7:20623. 
31.  Holm S. A simple sequentially rejective multiple test procedure. Scand J Statist 1979:6:6570. 
32.  Ward WD. The American Medical Association/American Academy of Otolaryngology formula for determination of hearing handicap. Audiology 1983;22:31324. 
Correspondence Address: Andrew B John 1200 N. Stonewall Ave., Room 3080, University of Oklahoma Health Sciences Center, Oklahoma City, OK, 73117 USA
 2 
DOI: 10.4103/14631741.93321 PMID: 22387708
[Figure 1]
[Table 1], [Table 2], [Table 3], [Table 4] 

This article has been cited by  1 
Interventions to prevent occupational noiseinduced hearing loss: A Cochrane systematic review 

 Verbeek, J.H., Kateman, E., Morata, T.C., Dreschler, W.A., Mischke, C.   International Journal of Audiology. 2014; 53(2)   [Pubmed]   2 
Interventions to prevent occupational noiseinduced hearing loss: A Cochrane systematic review 

 Jos H. Verbeek,Erik Kateman,Thais C. Morata,Wouter A. Dreschler,Christina Mischke   International Journal of Audiology. 2014; 53(S2): S84   [Pubmed]  






