This is a summarized version of an article that appeared in Ear & Hearing{{1}}[[1]] Dobie, R.A. (2011). The AMA Method of Estimation of Hearing Disability: A Validation Study. Ear & Hearing, 32:6, 732-740.[[1]], Dec. 2011.
Robert Dobie, M.D.
Robert A. Dobie, M.D.
First, thanks to Wayne Staab for his kind invitation to add this note to the blog.
Audiologists and otolaryngologists know that hearing loss is a multidimensional phenomenon: sensorineural vs conductive, sensitivity loss vs distortion, high vs low frequencies (not forgetting those essential middle frequencies!), peripheral vs central, child vs adult, bilaterally symmetrical vs unilateral/asymmetric. While no single descriptor can completely capture all these dimensions, there are times when a single-number scale is useful and in fact necessary. Epidemiologists may need to count individuals whose hearing loss is worse than some arbitrary degree of severity; clinical researchers may need to summarize the benefits of an intervention (or the harms of a hazardous exposure) using a yardstick that can be applied to all subjects; and of course workers compensation systems need a way to estimate the severity of a worker’s hearing loss on a scale that can then be translated to a financial benefit.
AMA Binaural Hearing Impairment (BHI) Estimate of Hearing Disability
The most widely used single-number descriptor of hearing loss for American adults is the AMA’s “binaural hearing impairment” (BHI). The AMA first promulgated a method of estimating what was then called “hearing handicap” in 1942 {{2}}[[2]] Dobie, R. A. (2001). Medical-Legal Evaluation of Hearing Loss (2nd ed.). San Diego, CA: Singular Thomson Learning.[[2]]; after revisions in 1947, 1959, and 1979, the method has been unchanged for the past 32 years. Briefly, the AMA method considers air-conduction thresholds at 500, 1000, 2000, and 3000 Hz; assumes a “low fence” of 25 dB HL and a “high fence” of 92 dB HL; and gives five times more weight to the better ear than to the worse ear. BHI is calculated from these eight thresholds, and ranges from 0% (for people whose pure-tone averages are both 25 dB HL or less) to 100% (both pure-tone averages > 92 dB HL). There are many assumptions implicit in this (or any other) method of severity estimation, as pointed out by Dix Ward in his 1983 article in Audiology {{3}}[[3]] Ward, W. D. (1983). The American Medical Association/American Academy of Otolaryngology formula for determination of hearing handicap. Audiology, 22, 313–324.[[3]].
Testing the AMA BHI Assumptions
Many studies have compared self-report of hearing problems to various audiometric tests, but none have directly tried to test the assumptions of the AMA method, as Dix described them, against self-report. I was pleased to find in the literature a large set of audiometric and self-report data (over 1000 patients) from five geographically-diverse audiology centers {{4}}[[4]] Erdman, S. A., & Demorest, M. E. (1998a). Adjustment to hearing impairment I: Description of a heterogeneous clinical population. J Speech Lang Hear Res, 41, 107–122.[[4]] {{5}}[[5]] Erdman, S. A., & Demorest, M. E. (1998b). Adjustment to hearing impairment II: Audiological and demographic correlates. J Speech Lang Hear Res, 41, 123–136.[[5]]. Marilyn Demorest and Sue Ann Erdman were kind enough to share these data with me.
My “gold standard” for self-assessed hearing disability was the mean score on the 18 communication performance (CP) questions in Demorest and Erdman’s Communication Profile for the Hearing Impaired. As expected, CP correlated better with better-ear thresholds than with worse-ear thresholds; CP also correlated better with pure-tone thresholds than with word recognition scores. These findings are consistent with most previous studies. Better-ear to worse-ear weights ranging from 3:1 to 9:1 performed very similarly. CP correlated better with pure-tone averages that included 500, 1000, and 2000 Hz than with averages that omitted any of these three frequencies, and no pure-tone average outperformed the AMA average of 500, 1000, 2000, and 3000 Hz. Finally, the relationship between better-ear pure-tone average and CP was similar to that of the AMA method: a roughly linear reduction in CP for average thresholds above 25 dB HL.
Combining pure-tone averages and word recognition scores could improve prediction of CP, but only very slightly. In medical-legal settings where there are no good tools to detect and manage exaggeration of word recognition difficulties, the potential benefits of combining these two dimensions would seem to be outweighed by the risk of greater inaccuracy. In other settings, such as clinical research on non-claimants, a method that combined these two audiometric dimensions might prove worthwhile.
Summary
The bottom line was that this study did not suggest any change that could improve the accuracy of the AMA method, at least in the medical-legal context.
