Dr. Ross on Hearing Loss
Hearing Aid Research: Reporting both Positive and Negative Findings
by Mark Ross, Ph.D.
According to the New York Times, the pharmaceutical industry now funds the majority of biomedical research being conducted in this country, either directly (by employees) or indirectly (through grants to outside investigators). As a result of this trend, medical journals are now devoting a great deal of editorial space on the potential conflict of interest inherent in these financial ties. It is rather a sticky problem; without support, some research will simply not get done. While the article does not question the basic integrity of the majority of the researchers, still the fact remains that as human beings they are nonetheless subject to unconscious biases. Not biting the hand that feeds you is a response true for academic researchers as well as the rest of humanity. The basic question raised by the article was whether this financial support translated into the selective public dissemination of the results, either by overemphasizing positive findings and/or de-emphasizing or ignoring negative findings.
According to the Times, this does indeed happen. One example given occurred in l998 with a study that investigated dozens of articles conducted on a controversial family of blood pressure medicines. It seems that "authors who championed the drug's safety were more likely to have financial relationships with the manufacturers than did the critics." A similar example was found with studies concerning the potential problems of passive smoking: authors who discounted the dangers were more often found to have financial ties to the tobacco industry than those investigators who reported potential problems. Finally, there is the fact that some research never makes it to publication at all. It seems that research that demonstrates that some drug is no improvement over older ones is less likely to be published than one that demonstrates a drug's efficacy.
While reading this item in the paper, I thought of the hearing aid industry. Here, too, the majority of published articles are at least in part supported by hearing aid manufacturers, many by staff scientists and some through grants to outside researchers. Of course, the stakes are not quite so high with hearing aids as with medication: bad medicine can prove fatal, an event hardly likely to happen with an ill-chosen hearing aid! Still, as it affects an individual, how a hearing aid performs is nevertheless very significant. Any hearing aid that provides superior functional hearing can improve a person's quality of life, hardly something that we can dismiss as irrelevant.
As I read the hearing aid literature, my impression is that research that fails to demonstrate the efficacy of some new feature (as opposed to a speech processing strategy or fitting "prescription") rarely makes it to publication. This is not a trivial point. Researchers will agree that what is not found is often as important as what is. That is, demonstrating that something doesn't work is not a waste of time or a research failure, but often, rather, a necessary precursor for arriving at what does work. Many of us older folks remember Spencer Tracy (acting the part of Thomas Edison) going through hundreds of possible filaments in light bulbs before he finally arrived at the one filament that proved to be acceptable. Without all the failures preceding the "eureka" moment he couldn't have arrived at the one successful ingredient. First-class researchers always need to know what didn't work so that they can build on the past and not simply repeat it.
Not just researchers, but consumers too have a legitimate interest in desiring that both positive and negative results be published. They have a right to know whether some hyped hearing feature actually does work as promised. Is there clinical evidence to support the very appealing technical descriptions and marketing terminology that appear in consumer and professional publications? If only positive results are published, then only half the story is being told. Later I will discuss several excellent studies in which the investigated features were not supported by the research. This does not mean these studies are unimportant; on the contrary, they are very important for reasons I'll attempt to clarify below.
As I look through the hearing aid research literature (appearing in two trade and three professional journals) in the last several years, it does appear that there are few published reports on negative findings. The overwhelming majority of the studies, particularly in the trade journals, found that the hearing aid developments they investigated were beneficial. Perhaps this reflects the expertise of the engineers and hearing scientists responsible for the new developments; that is, their theoretical predictions all proved to be valid when tested on human subjects.
But this is not how the process works. "Eureka" moments are built upon a great deal of experimentation and trials. What happens is that engineers usually tinker, trying this and that, before arriving at an end product. They get an idea from their own work, from the basic research of hearing scientists, and from their own previous efforts and go on from there. But how do they then prove the value of the end product? In the hearing aid industry, the final step in any new development has to be some validating clinical research project. In other words, does the new feature really do what it is purported to do when used by human subjects? This is what we can't be sure of unless all research reports, both positive and negative, are published. Ideally, this step should precede or be concurrent with the marketing of some new and presumably improved hearing aid. Testimonials, no matter how heart-felt and sincere, are not enough. What is required is objective, and preferably peer-reviewed, published research. Consumers are not helped by improved marketing slogans, but by the actual performance of the hearing aids they wear. It is not what a hearing aid theoretically should be doing that is germane, but what it actually is doing.
I know of one sponsored multi-clinic hearing aid study that was never published. The grant contract specified that the results were the property of the hearing aid company and could not be independently released by clinic researchers. In this case, even though the research results were not released, the hearing aid itself is still being marketed. What was the result of this research? We don't know. I have no problem with the hearing aid employed in the study; it is a good one. It's just that neither professionals nor consumers know how the hearing aid fared in this investigation. Evidently, the results were squelched ("proprietary" information, you know). One colleague tells me of other studies with negative findings that are often put in the drawer.
The "accentuating the positive" orientation can be seen in a little survey that I conducted. By my count, 14 of the 17 hearing aid clinical research studies published in the two trade journals in the last several years were funded either directly or indirectly by the hearing aid industry. Sixteen studies reported positive results for the hearing aid feature being investigated (the last one had mixed results). While none of these studies were peer-reviewed (a requirement of professional journals), they are nonetheless published studies and each can be evaluated and stand on its own merits. This is what research is supposed to do. But, and I'm not being facetious, we can't evaluate and be guided by research that is not published.
While fewer hearing aid articles were found in the three professional journals I looked at, more of them reported negative findings. In the past two years, I could find a total of only eight articles in which some new hearing aid feature was tested on human subjects (three others were basically derivatives of one major study). Half of these received some support from a hearing aid company (perhaps just the loan of several hearing aids). Three of the eight reported negative results. While this is a very small sample, it accords with our common knowledge: independent researchers do appear more likely to report that a new hearing aid or hearing aid feature does not work as predicted.
One of the more interesting, and valuable, hearing aid features developed in recent years has been the inclusion of directional microphones. We now have an abundance of research that demonstrates the effectiveness of this feature. We also have a better understanding of the environmental and social conditions in which the maximum benefit can be achieved. These include the fact that directional microphones offer less help in highly reverberant enclosures or when the source is distant from the listener, and the necessity to face the sound source while interfering noise is at the rear.
To its credit, the hearing aid industry is constantly attempting to refine and improve its products. One recent advancement is the adaptive directional microphone. In an adaptive directional microphone system, the directional characteristics are constantly changing in order to place the "null" (the point of maximum sound attenuation) at the most intense noise source. The classic example of this phenomenon is the moving truck behind a listener. As the truck moves from one position to another behind the listener's back, the point of maximum attenuation follows the truck movements. Thus the hearing aid will continue to deliver the highest speech-to-noise ratio regardless of the changing position of the noise. At least this is the theory.
In real life, the acoustic environment is always more complex and more individualized than can be simulated in the clinic or laboratory. According to the research, adaptive directional microphones will work as predicted when confronted with a single, moving noise source. As such, they have performed slightly better than conventional directional microphones (with fixed directional characteristics). The question posed by Dr. Ruth Bentler and her colleagues in a recent issue of the American Journal of Audiology was whether adaptive directional microphones retain their superiority over fixed directional microphones when the dominant moving noise source was combined with a number of lower level noises. That is, the moving noise source (think of the moving truck) was then combined with other noise sources (think of other cars on the road at the same time), all somewhere to the rear and sides of a listener.
Ten experienced hearing aid subjects participated in this study. They were tested under two conditions of noise (moving and fixed) and in two types of environments (anechoic and reverberant). In the moving noise condition, the dominant noise was periodically switched from one of five speakers (located behind and to the sides of the listener) to another. The difference between this procedure and others that have tested adaptive microphones is that the other speakers were not turned off. They continued to deliver competing noises but at a level 8 dB lower than the dominant speaker. This seems to be a more plausible simulation of a realistic environmental noise condition than a single moving source of interfering noise.
The research question posed was whether the adaptive directional microphones would "follow" the dominant noise as it moved around the back of the head. If so, the result should be a stable increase in the speech-to-noise ratio regardless of the position of the noise. This should then produce higher speech perception scores than would occur in the fixed directional mode. This latter condition would, theoretically, provide the highest speech-to-noise ratio only when the noise source was directly to the rear. In addition to objective measures of speech perception, the subjects were required to complete three different self-report measures that assessed their subjective responses to the different microphone conditions. The question here was whether the participants would subjectively be aware of the differences between the two microphone conditions and, if so, which they preferred.
Most significant, the study was "double-blinded". Neither the subjects nor the examiners were aware of what microphone system the subjects were wearing (the same aid was used, but programmed differently by one set of audiologists and tested by other audiologists). This procedure effectively precludes any subtle biases, deriving from the appearance of the hearing aids, or expectations of benefit, from influencing the results. For subjects to be told, or to gain the impression, that one aid represents the newest and most "modern" technology while the other aid is older and more "conventional" can't help but influence a person's perceptions. (This did happen in a study that looked at how the labels used to describe hearing aids influence a user's preferences.)
The result can easily be summed up: There was no difference in objective performance or subjective ratings between the adaptive and the fixed directional microphone conditions. (Both, however, would be superior to hearing aids with omnidirectional microphones.) For me, reading this study, this result was unexpected. From all I had read I expected the adaptive directional system to be superior. But should this study with negative findings be classified as a failure? Absolutely not: it is a very valuable study. It has provided us with significant insights into the performance of one of the new developments in hearing aid design, insights that can be incorporated in newer instruments as they are developed.
In fact, it was the simultaneous noise occurring at different positions around the head that was probably responsible for the negative result. Evidently, the adaptive system could not distinguish the 8 dB difference between the dominant and secondary noise sources and treated the situation as a diffuse noise field. This may not have occurred if, instead of only 8 dB separating the two types of noises, the difference was increased to perhaps 15 dB (just guessing). Adaptive microphones do work, but evidently they require a larger difference between the moving, dominant noise and the other noise sources than 8 dB. But the larger this difference is, the more unlikely it is to occur in real-world situations. I mean, how often are we exposed to a single, moving noise source? Herein, lies the quandary.
Should some consumer, then, purchase hearing aids with adaptive directional microphones as opposed to aids with fixed characteristics? I would say it depends. It depends upon the difference in cost and whether the aid with the adaptive system also includes other features that may be useful for a listener (like one that automatically switches to the omni-directional mode in relative quiet or in a diffuse noise environment). It is also important to realize that, in the real world, we are free to move our heads, something not encouraged in a controlled research project. And when we move our heads, we can try to keep the dominant noise directly to our rear thus maximizing the directional effect. The cost/benefit ratio is something that each person would have to work out for him or herself, after a serious trial period.
In another article appearing in the same issue of the American Journal of Audiology as the preceding one, Dr. Larry Humes and his colleagues examined a topic that I thought had basically been put to rest. They looked at the performance differences between single channel in-the-ear (ITE) linear hearing aids and two channel in-the-canal aids (ITC) that incorporated a wide dynamic range compression (WDRC) circuit. (These aids automatically adjust the amount of amplification depending upon the intensity of the input sounds.) My understanding of the previous research was that generally two-channel amplification was preferable to a single channel and that WDRC circuits were better than linear aids. The results of this research require us to reexamine these assumptions.
The background of the current study is interesting. In a previous study, the authors had administered a very comprehensive set of objective speech perception scores, plus six types of surveys, questionnaires, and scales designed to provide subjective measures of benefit, satisfaction, hearing aid usage, and sound quality. The purpose of this previous study, in which 173 subjects were tested with linear ITE hearing aids, was to examine various kinds of outcome measures and relate them to hearing aid performance data. This study, which was concerned with modeling (or predicting) hearing aid outcomes, will not be reviewed here.
However, given the existence of this valuable data, collected under very rigorous and controlled circumstances, the authors then devised a second study in which they could compare a group of subjects using two-channel WDRC aids to the single-channel linear aids employed in the previous study. The availability of a large group of subjects in the first study (173) permitted the authors to find 53 of them who closely matched the 53 subjects in the second study. To review, then, two groups of 53 subjects each were used in the current study, very closely matched on demographic and audiological variables, all given the same extensive battery of objective and subjective tests, with one group using linear ITE hearing aids and the other two-channel WDRC hearing aids.
The results indicate that there was no difference in scores between these two types of hearing aids. The results of all the speech perception tests (in quiet and noise and at different intensity levels) were not statistically different. This was true of the measures performed at both one- and six-month post-fit intervals. These results contradict previous findings on two counts. First, a number of studies have found that speech perception scores using the WDRC hearing aids (which do not use a volume control) are better than those obtained with a linear circuit, particularly at low input levels. However, when people are given access and freedom to use the volume control in a linear aid, this benefit dissipates. (But we should recognize that, for many people, not requiring a volume control is in itself a benefit.) The second count is the fact that the two-channel instrument did not perform better than the single-channel hearing aid. In the past, I have reviewed research that demonstrates the opposite, i.e. that two channels were superior.
The scores on all the subjective measures also did not show any statistically significant differences between the two types of hearing aids. These measures include the Hearing Aid Performance Inventory, Hearing Handicap Inventory for the Elderly, Hearing Aid Satisfaction Survey, Hearing Aid Benefit Profile, and, finally, the Hearing Disability and Aid Benefit Interview. (I can't recall any study that administered more such subjective rating scales to their subjects!) All of them require people to subjectively rate their personal experiences with the hearing aids. On none of them did the subjects judge one type of aid as being superior to the other. It turns out that both technologies provided significant help to the subjects, all of whom had mild-to-moderate bilateral hearing losses.
Although the authors did not actually predict the superiority of one type of hearing aid over the other, given the differences between them, the logical prediction would have been higher objective scores and subjective preferences with the two-channel WDRC aid. So one could define these results as essentially negative, but of course they're really not. They provide future investigators on these topics with a bit of an enigma and a bit of a challenge. But this is how the process is supposed to work. Out of it all we can hope to see a better approximation of the "truth."
What it says for consumers, as in the previous study, is that maybe the higher priced, more sophisticated hearing aids represent a bit of overkill. Maybe once we get beyond increasing the speech-to-noise ratio, we begin to see diminishing returns with increasingly complex (and expensive) hearing aids. But, and this is a big but, making predictions in an individual case based on general research is always a bit chancy. This is why a supervised and rigorous trial period is often so necessary.
I do have one bone to pick with the authors of this study. In the full-shell ITE hearing aids they selected for their subjects, they requested only one telecoil (for the ear selected by the subject). As we've discussed in these pages many times, a telecoil can also be used as an assistive listening device receiver. Including a telecoil in just one ear precludes utilizing the hearing aids for this purpose. And evidently there were no telecoils in any of the ITC aids. While there is less room in these aids for a telecoil, surely some could have accommodated a telecoil. This study was conducted a few years ago; hopefully, the situation would be different now.