Dr. Ross on Hearing Loss
Hearing Aid Research
by Mark Ross, Ph.D.
Wandering through the aisles in the exhibit hall at the last AAA convention, I couldn't help but be impressed by the array of new and refined technical developments. Clearly, hearing aids have come a long way since I put my first one on 50 years ago. In particular, the advent of wearable digital signal processing hearing aids has made it technically feasible to package sounds in ways that could not even be conceptualized at that time. Their technical sophistication is truly awesome and, to be quite frank, more than a bit overwhelming and intimidating.
The current status of these developments was recently reviewed in an article appearing in The Hearing Journal (Mueller, 2000). In this article, representatives from seven major hearing aid manufacturers described the distinguishing features of their latest and most advanced offerings. Although they were all describing digital hearing aids, what struck me was how much the aids seemed to differ from each other, both in features, design, and the nature of the speech processing algorithms they employed. They included a varying number of bands (primarily for frequency response shaping) and channels (for more advanced signal processing). Each company has what appears to be a well-developed conceptual framework for its particular hearing aid design, including several based on the dynamic functioning of the cochlea using a critical band formulation.
All of the hearing aids include some form of wide dynamic range compression, but vary widely in the nature and rapidity of the time constants, threshold of compression (with and without expansion circuits) as well as other parameters relevant to compression amplification (input and/or output compression, compression ratio, etc.). The aids also differ in the types of noise suppression and anti-feedback circuits they include. While all of the hearing aids are programmable and thus can be extensively modified, other dimensions are built into the product (e.g., number of bands and channels, time constants, etc.). In brief, as described by the very knowledgeable representatives from each of the companies, these new products all appear to be the culmination of a great deal of research and intensive development efforts.
However, as I read through the article, I was reminded of a joke that seems to illustrate the quandary it presents us:
It seems this Rabbi is listening and counseling some woman complaining about a disagreement she is having with her husband. Meanwhile, the Rabbi's wife is listening from the other room. After listening to the woman, the Rabbi says, "You know, you're right. I agree with you."
The next day the woman's husband comes to tell his side of the story. The Rabbi's wife is again listening from the next room. After listening to the man for a while, the Rabbi says, "You know, you're right, I agree with you."
After the man leaves, the Rabbi's wife comes into the room and says to him. "Husband, when the woman was here yesterday, you said she was right. Today, you say the man is right: You know they can't both be right!"
"You know" replies the Rabbi, "You're right!"
This is exactly how I felt when I read the description of the different speech processing strategies and incorporated features of the hearing aids from these seven companies. They all can't be right! There are significant differences between them, differences that would remain to some degree regardless of any subsequent re-programming and fine-tuning. Are we to assume that these would be equally beneficial to people displaying a range of different degrees and configuration of hearing loss? I doubt that many of us would accept this assumption. Intuitively, we feel that some speech processing strategies must be better than others for particular persons.
The problem is that relatively few published research studies have directly compared different speech processing strategies with DSP hearing aids, particularly the kind of major differences illustrated in Mueller's article. Yes, we can point to the clearly demonstrated benefit of directional over omni-directional aids (something that all the contributors to this article agreed on), multi-band relative to single band instruments (with a bit less certainty), and wide dynamic range compression circuits compared to linear aids (but mainly at the lower input levels). Beyond these areas, however, the clinical research literature seems not quite so informative regarding the relative performance of one algorithm or feature over another. (While there are many studies that compare the loudness and audibility "targets" of various prescription methods, these studies are not germane to the main point of this paper.)
Instead of research that directly compares speech processing strategies, the superior performance of particular hearing aids is implied on the basis of their technical description and the engineering rationale of the various design features. There is an apparent assumption that technical sophistication per se translates into listening superiority. It is simply assumed that a more powerful microprocessor, greater electroacoustic flexibility, or some creative engineering feature will result in improved hearing performance. While this is a powerful and sometimes appealing assumption, and very effective for marketing purposes, it is not one that should be accepted by audiologists without direct verification by appropriate clinical research efforts. Marketing "hype" is not enough.
Personal experience by audiologists and testimonials by clients are also not enough. These subjective impressions do not negate the need for objective, preferably blinded, research efforts. Of course, we cannot ignore the testimony of our clients regarding their experiences with some new hearing aid or speech processing strategy. And, of course, an audiologist's clinical experience with a particular feature or speech processing strategy must be given due weight. But asserting the authority of "in my experience" is hardly a scientifically defensible method of claiming superiority of one set of electroacoustic characteristics (or one hearing aid) over another. As we well know, audiologists vary widely in their favorite hearing aids, those which they've had great "success" (however defined) in fitting to their clients. They all can't be "right."
Certainly, we all want to see our clients improve their functional hearing capacities because of the hearing aids we select for them. And, compared to the unaided condition, there's no doubt that most do. Hearing aids are unquestionably an effective way of reducing the handicap of hearing loss. Because of practical and logistical reasons, however, we have no way of knowing whether a client will do even better with some other speech processing strategy or hearing aid. We don't know, for example, whether an additional number of bands or modifications of the numerous parameters of a compression system would improve a client's performance even more. This is, however, precisely the kind of information that should be available from a corpus of clinical research studies.
I doubt if we'll soon arrive at the happy state where we can, on the basis of a comprehensive psychoacoustical evaluation, precisely predict an "optimum" system for a person with a specific type and configuration of hearing loss. But this does not mean that we still have to depend primarily on the "how does this sound?" strategy. As a discipline that claims to be scientifically based, we should be able to do better. For example, let us consider the research that has been and is being conducted with cochlear implants. These studies are impressive for numerous reasons: They are well controlled, comprehensive, longitudinal, and integrated into clinical practices. Every newly introduced coding strategy or electrode modification is compared to previous generations, using either the same subjects or comparable samples. And then the findings are published for the world to see. People who wear hearing aids deserve no less than this. If hearing aid research received this level of commitment, we would certainly have more answers to the many hearing aid fitting questions that now perplex us.
Client testimonials are also often used as a substitute for hard data in determining the "best" hearing aid. Personal testimonials, which speak to a hearing-impaired person's experiences with a particular hearing aid, are valuable information and should never be ignored. They also should not serve as a primary basis for selecting a hearing aid or speech processing strategy, either for the general population of people with hearing losses or even for the person who is making the testimonial. If you thumb through the trade journals of the past fifty years, you will find glowing testimonials associated with just about every newly introduced hearing aid or feature. People make such testimonials for a myriad of reasons, including clearly described superior listening experiences, comparison with the unaided condition or previous problem instruments, and an intense desire to believe they are really hearing better.
The well-known "hype" studies conducted several years ago by Ruth Bentler and her colleagues are a good example of how belief can influence both subjective judgment and objective performance with hearing aids. In one of these studies, the subjects alternately wore two digital hearing aids, one time while one was labeled "digital" and the other time "analog". The next trial, the situation was reversed, with the previously labeled "digital" aid now called "analog" and vice versa. The results showed that almost always the subjects preferred the "digital" instrument over the presumed "analog" hearing aid. In the other study, the subjects alternately used the exact same instrument, but one time while it was labeled "digital" and the other time "analog." Subjective preferences for the "digital" were overwhelming. Even more interestingly, the objective performance of the subjects was somewhat superior in the "digital" condition.
Audiologists often foster this "placebo" effect, albeit unconsciously. We understand the technology better than our clients and, in reading the manufacturer's material, it's hard not to be influenced by the engineering rationale underlying the inclusion of particular features and processing strategies. It seems to make so much sense (and it often does). We can't help but be impressed with the rapidity and flexibility of a digital signal processing circuit and the explanation given by bright hearing aid engineers as to why a system was designed the way it was. And of course we believe in the quality of the products we are dispensing, else we would not be dealing with them.
So we pass our convictions about the superiority of a particular device to our clients, hoping for and encouraging (perhaps unaware that we are doing this) positive reports from them. In turn, we take their positive responses as corroboration of our initial confidence in the product we have dispensed. While quite circular, all this is very understandable, very human, and will probably always be a factor in the dispensing process. I don't really object to this. We really do have to believe in our products and clients really do have to believe they are hearing better. What I do object to is when audiologists depend only upon manufacturer provided technical descriptions, their own personal impressions, and testimonials received from clients. This does not ensure a proper role for scientific objectivity in the hearing aid selection process.
The operation of the Songbird disposable hearing aid provides a perfect example why more objective hearing aid research is necessary. This hearing aid incorporates only nine individualized amplification choices, with the claim that one of them will be appropriate for approximately 80% of people with mild to moderate losses (McCandless, Sjursen & Preves, 2000; Preves & Dempsey, 2000; Staab & Preves, 2000). The company's research shows that the subjects did as well with the Songbird as they did with their own hearing aids, and that they achieved HINT scores comparable to those obtained in a number of other studies that employed the same test. (While hardly conclusive, since direct comparisons to advanced hearing aids were not made, at the least these results should serve as a challenge for either corroboration or refutation.) Given their results, does this mean that it is not necessary to further refine the electroacoustic characteristics of hearing aids in order to achieve maximal benefit for individual people? That picking just one of the nine response patterns offered will be as appropriate for someone with a mild to moderate hearing loss as that available with a modern digital hearing aid? What does this now say about the fantastic flexibility and amplification options incorporated in current hearing aids? Are they not necessary? Is it all overkill? Is an impaired auditory system unable to utilize all the electroacoustic flexibility and options that are now available?
I don't have an answer to these questions, but neither as far as I can see, does the hearing industry or the professional community. But these are not questions that can be swept under the rug. The advent of the Songbird is presenting the profession with a major conceptual challenge, one that is hitting us where we live. The issue I'm concerned with here is not the fact that the aid is disposable and very inexpensive, though this is significant in its own right. Rather, I'm thinking about the hearing aid's fitting rationale, which clearly implies that the current generation of modern digital hearing aids may be over-engineered. Does this mean fewer electroacoustic options would serve the vast majority of people with hearing loss as well as the almost infinite variety that is now possible?
We don't know the answers to these questions, which is precisely why we need more hearing aid research. Right now, we seem to be depending upon research funded or conducted by industry. This has a place, of course, but must not be the only source of information regarding the actual listening performance of hearing aids. An independent profession requires its own independent scientific base, developed through the efforts of its own objective researchers. Unfortunately, at the same time we require more people trained to conduct audiological research, the number of people trained to do such research is dropping (Bess, 2000), perhaps because the Au.D. is siphoning off prospective researchers. Whatever the reason for the relative paucity of hearing aid research, however, the many unanswered questions regarding hearing aid fitting still require answers. It's clear that "everybody can't be right!"
Bess, F. (2000). Professional Training: Past, Present, Future. Paper delivered at the Early Childhood Deafness Symposium, June 1-2, Emerson College, Boston, MA.
Mueller, G. (2000). What's the digital difference when it comes to patient benefit? The Hearing Journal, 53(3), 23-32.
McCandless, G., Sjursen, W. & Preves, D. (2000). Satisfying patient needs with 9 fixed prescription formats. The Hearing Journal, 53(5), 42-50.
Preves, D. & Dempsey, D. (2000). Speech recognition in noise results for a disposable hearing aid. The Hearing Review, 7(4), 34-42.
Staab, W. J. & Preves, D. (2000). Deep canal hearing instrument fittings: A new approach. The Hearing Review, 7(6), 50-53.
This article was supported, in part, by Grant #H133E980010 from the U.S. Department of Education, NIDRR, to the Lexington Center.