Editor's Note: This text course is an edited transcript of a live seminar on AudiologyOnline, presented as part of Siemens Hearing Instruments Expert Series 2012. Please download supplemental course materials. To view other courses in this series, please visit the Siemens course library on AudiologyOnline.
Today we are going to be talking about implementing a gain learning feature, as well as whether timing matters. The concept of adaptation and how people adapt to hearing aids over time has been an interest in my lab for the last 20 years. Currently, along with Min Zhang and Reem Mulla, who are both Ph.D. students in my lab, we are looking at some of the technology that has been developed to address adaptation. Before we begin, I would like to thank my colleagues for the work that they have done as well as acknowledge Siemens for their funding assistance and for providing hearing aids that we used in some of the research I will be presenting today. Thanks also to Gus Mueller for his input. Now, on to our topic.
I think we can all agree that the basic goals for a hearing aid fitting are to return audibility, with comfort, and to eventually restore normal loudness perception.
Regarding audibility, most of you know that I often talk about verification of audibility using real-ear measures. Whether you prescribe to that line of thinking or not, you probably agree that the goal of fitting hearing aids is to return audibility to the listener, regardless of how we get there. Patients may talk about how they want to hear and perform in communication situations, but underlying that is the knowledge that audibility is important. We have to bring the soft, moderate and loud signals back for them to do those things that are meaningful in their lives, while keeping the signal comfortable.
Audiologists also tend to be concerned about signal comfort, especially for loud sounds. There have been data published that indicate that people do not really adapt to the comfort of loud sounds. So before we send the patient out with a new pair of hearing aids, we want to make sure that loud sounds are comfortable for them. Soft and moderate sounds are a different issue because we know there is adaptation that goes along with those sounds. That same concept of comfort has to do with physical comfort, and certainly we do not send patients out of the clinic if something is not physically comfortable. Much of our time in the first hearing aid fitting sessions can be spent on ensuring comfort. Especially for soft and moderate signals, our time may even be spent convincing people that it is okay to be a little uncomfortable. Either way, we are spending time with this concept of comfort.
In addition, most audiologists would like patients to end up with a normal loudness perception while wearing hearing aids. This means that while the patient is wearing hearing aids, soft sounds should sound soft, moderate sounds should sound moderate, and loud sounds should sound loud. When a patient gets upset that a particular sound is loud while he or she is wearing hearing aids, and we can determine that indeed it was a loud sound, we can tell the patient that he or she perceived that sound the same way that others would have. We can convince the patient that this is actually a good result. Initially after being fit with new hearing aids, patients will likely experience some abnormal loudness perception, especially for soft and moderate sounds. We certainly all have had patients experience that and have dealt with that counseling issue.
Managing New Users' Perceptions of Sounds
Most new hearing aid users find that soft sounds are too loud, or louder than they prefer at the time of the initial hearing aid fitting. We have changed their audibility. If we are programming based on speech targets, we know we have achieved audibility. When we fit a hearing aid in a quiet, carpeted clinic room we may say, "When you leave here, you may be less comfortable, and that is okay." There are different ways we may handle this. We can use the tactic where we tell patients that they will adapt over time and that their brain will get used to the new sound. They will be coming back for a checkup in a few weeks so any issues can be resolved at the follow-up. Or, we may offer, "Get used to it, and then what you cannot get used to, we will deal with in your next appointment." The downfall with this approach is it that too much time could lapse before the next appointment, and the patient is not going to be ready for full-time hearing aid use because they really cannot tolerate the sound.
Some of us will turn the gain down at the first fitting. We make sure the hearing aid is doing what it is supposed to do, but then we will dial it down and tell the patient, "This is going to be a little bit overwhelming. Once you get used to this level, we will turn it up a little bit more in a few weeks at your next appointment." The problem with this approach is that we know it means multiple visits, which can be a problem for you in the clinic, and it can be a problem for patients who are busy or who are more fragile and have a difficult time coming in to see you.
Another way audiologists approach the initial hearing aid fitting is to just click "first fit". Manufacturer hearing aid programming software generally under amplifies at first fit. I think the reason for this is because manufacturers are worried that patients will reject or not like the hearing aids if the clinician does not perform measurements and make adjustments. Indeed, hearing aids are sometimes rejected for a variety reasons, one of which is because they are under-fit. This approach is probably not our best way of going about a hearing aid fitting, either.
You could use an adaptation feature. Several hearing aids now have adaptation features, which means you program the hearing aid to the desired fitting, and the hearing aid will slowly add more gain over a short period of time to reach the final fitting you prescribed. Often the adaptation feature will allow you dictate the time for the adaptation manager to increase the gain, whether that is over a period of weeks or months. All manufacturers are different in how they gradually apply gain, also. It tends to be softer inputs for the high frequencies. These are sounds that most everyone believes are the most problematic in the beginning. There has not been a lot of research on how successful the adaptation feature is, but it is certainly fairly common in hearing aids today.
You can always just give patients a volume control and tell them to adjust it as they want. Likewise, we do not have good data to see if users adjust the hearing aid to whatever setting you prescribed originally, but a volume control is also not an uncommon feature. However, once you give someone a volume control, they can do whatever they want with it. Or, you can set the program button like a volume control or you could use a gain learning or training feature, which is what we are focusing on today.
This is a bit different from adaptation. Adaptation means that you believe you have an adequate first fit based on matching targets or using a prescriptive algorithm. That is ultimately the goal. All you are doing with adaptation is backing off the gain at first and getting the person there slowly so they have time to adjust gradually, rather than all at once.
A gain learning feature has the premise that you, as the clinician, have a starting point. You have to define the starting point based on individual patient characteristics and professional knowledge. The concept is that you let the patient use their hearing aid in the real world, and they are going to train the hearing aid on some of their individual preferences. You can think of this as getting to someone's preferred gain. The data from people that developed the Desired Sensation Level (DSL) or the National Acoustic Laboratories (NAL) programs show that there is some variability of how you might fit someone.
It is not that you have to match the target exactly. You can have some leeway, as people have different preferences. Indeed, the newest versions of DSL and NAL are derived partially from research on patient preferences. There have been changes in both programs from their previous version, with DSL reducing some gain and NAL increasing some gain. DSL and NAL are now more similar than ever to each other as a result.
Here is how a gain learning or training feature works. You determine a hearing aid fitting, and send the patient out to wear the hearing aids and make adjustments. The hearing aid is sampling sound, and it has a way to define if the sounds are quiet, noise, music, or speech. Depending on the environment as determined by the hearing aid, when the user makes adjustments, those adjustments apply only to the settings in that specific environment, which makes this even more interesting and complicated.
So, the person may make no adjustments in quiet, but in noise they want things tuned down. The tuning could be for the low frequencies or for the high frequencies. It could be for soft input sounds. It could be for moderate or loud level sounds. You can picture the complicated matrix of possible changes. If these changes are done consistently enough, the hearing aid then implements them, and that is the new starting point. This way, the user does not have to turn the hearing aid down every time they are in that same situation. That is the new point at which the person would start. That is what gain learning or a training feature does.
In my lab, we are trying to back up and think about what we would expect, in terms of the auditory system. We almost get ahead of ourselves before we back up and say, "Given what we know about the ear, what would we really expect to occur?" Why does this matter to us as audiologists? We could just fit the hearing aid to the final prescribed settings right from the start, but the user might reject the hearing aid, and then we are done. Again, we could alternately fit the hearing aid to a lower than ideal target, but they might reject it for the opposite reason of being too soft or not providing enough benefit. We could gradually increase the gain over time, but this may not be convenient or practical if the patient has to physically come in to our office. Or, we could use the automatic gain adaptation feature.
The question that drives our discussion today is, "What is the efficacy behind this feature?" Would we expect new hearing aid users to adapt differently to changed audibility? We assume there is quite a bit of variability between individuals. This means that some people could be set at their ideal hearing aid settings from the beginning. They deal with the changed audibility, and they are fine. Conversely, there are other people who have continual follow-up and may not ever be able to use the hearing aid settings you would like them to have. This supports the idea that it for a "preferred gain" fitting for individuals.
Levels of Variables
There are three variables that impact how we adapt and deal with changed audibility and they are physiologic, psychoacoustic and psychological. The physiologic level, or physical level, in this context can refer to the cochlea and disorders of the cochlea. We know a lot is happening physically to the cochlea if there is sensorineural hearing loss. There is elevation of the absolute threshold. There is hearing cell damage. We know there is a growth in audibility that is often described as very fast because sounds go from being audible to suddenly being too loud. In a sense, it is not really that the growth is that rapid; it is that there is a reduced dynamic range. There is less audible room for soft and comfortable sounds before becoming uncomfortable. That is what we refer to as recruitment. We know from research that there is definitely a change in perception of what is loud (Chen, Hu, Glasberg, & Moore, 2011).
There are also changes in loudness summation. When we are hearing through two ears, we have an advantage of not needing things as loud. There is quite a bit of data to show that people with hearing loss have a different loudness summation pattern than normal hearing individuals, both by intensity and by frequency. That is a complicated aspect that I do not think we have completely accounted for in hearing aid fittings.
The central control of the physiological level contains quite a bit of conflicting literature, so I cannot give you a definitive answer here. The literature is less suggestive of central auditory components to loudness perception (Philibert, Collet, Vesson, & Veuillet, 2002), but is more suggestive of peripheral excitation patterns explained by models of loudness (Heinz, Issa, & Young, 2005). I think this will be an interesting area to keep examining, and another reason why we probably see individual differences.
The next level is the psychological level. How we react has to do with personality, compliance, satisfaction and success. Robin Cox has studied personality, and her research (Cox, Alexander, & Gray, 1999) shows that personality can predict certain things about the success of hearing aid fittings. She has not specifically looked at loudness perception, but that is part of it. Personality may not change a person's perception, but it can influence their reaction and how they are going to cope with this altered perception of loudness. All these things point to why there are individual differences and why it makes sense to have the user have some control of hearing aid settings after the audiologist has made the first fitting.
The psychoacoustic level mixes both the psychology and the acoustic levels. There are other pieces that might also fit into this including hours of use, and that may go back to personality and compliance. We do have some evidence documenting that people change their perception of soft sounds over time. There was a time where we could not make soft sounds audible, so it just did not matter. What mattered was people still did not hear well. Then, with the use of compression and expansion, it was possible for people wearing hearing aids to hear soft sounds very well. Interestingly, many patients were not grateful for that. Their reaction was, "Now I hear the refrigerator humming, the air conditioning and other annoying low-level sounds that I do not care to hear." Again, many of us said, "You will get used to that." Then we saw more use of expansion in hearing aids so that the very soft sounds were not amplified, while maintaining more gain for "regular" soft sounds.
A study by Mueller and Powers (2001) examined new users and previous users who were fit with new hearing aids. The subjects rated their perception of loudness to sounds they heard through the hearing aids. The did this in different time intervals, from day one out to a week, four weeks, and then eight weeks. On day one, the new users rated what was evidently a soft sound as quite a bit louder than comfortable. It was not uncomfortably loud, but louder than the category you would expect for a sound of that absolute intensity. Previous users had a larger range, and gave responses to sounds more in the normal range. You would expect that because they should be more acclimated to amplification.
In week one, the new users were still rating soft sounds as comfortable, but much louder than "soft." The previous users already adapted quite a bit. In week four, the outcome was identical to week one, but by week eight, the previous users were in the range of a normal soft perception, and the new users were just about there. This tells us that over a couple of months, patients may, in fact "get used to it," just as we tell them.
One way to put individuals in control is with trainable hearing aids or those that have a gain learning feature. Two studies (Mueller, Hornsby, & Weber, 2008; Keidser, O'Brien, Carter, McLelland, & Yeend, 2008) specifically tested patient outcomes with trainable hearing aids. In the Mueller et al. (2008) study, the subjects were fit at a starting point, and then they were given a remote control to tune the treble, bass and volume. The hearing aid was sampling the environment including intensity and type of environment, such as noise or music. They found that the initial starting point of the fitting impacts the ending point. If you start the listener louder, they are going to end up louder. If you start them softer, they are going to end up softer. The subjects' manual changes appeared to be relative to the starting point of the fitting.
Keidser and colleagues (2008) conducted a similar study. The final analysis was also similar, and that is, it matters where you start. You need an appropriately prescribed starting point before you let people make changes. If the patient is under-fit from the start, they are essentially going to be even more under-fit after they make changes. Dreschler's group also looked at this and also came up with the same conclusion about the starting point (Dreschler, Keidser, Convery & Dillon, 2008). They also looked at people's reliability in making these kinds of changes, which also matters. They had people in control of the gain learning feature on more than one occasion. Partway through, the hearing aid was reset and the patients had to train the hearing aid again. They were within 2 dB between the trials. That is encouraging to know that the gain changes they make would be relatively consistent over time.
Zakis' and colleagues investigated the issue of trainability and found that the majority of people in their study preferred the final trained settings in environments like noise and music (Zakis, Dillon, & McDermott, 2007). They allowed the subjects to train the hearing aid, but they also had the original settings as a program in the hearing aid. The people preferred the trained gain, even though they were blinded at the time when they were making those ratings.
University of Pittsburgh Study: Does Timing Matter?
In my group, we were concerned with knowing when it was appropriate to turn this feature on. In a Siemens' hearing aid, the default setting is for gain learning is on. We knew that several studies demonstrated that the gain learning depends on where the fitting starts. So then we wondered if it made any difference as to when the gain learning feature was turned on. On the first day, at least in our experience, people think many sounds are too loud. Would they be apt to turn things down quite a bit more than someone who wore the hearing aids at the optimum setting for a few weeks or a month and then had the learning feature turned on? Does the timing of this feature impact the outcome measures? Those were the questions we were pursuing in the following study.
We decided to look at "return to normal loudness" and "satisfaction with loudness" which we measured by the Profile of Aided Loudness (PAL). We had people rate soft, moderate and loud sounds, and we looked to see if their overall satisfaction with the hearing aids and their audibility for soft sounds were in line with normative data. By looking at the articulation index, we can examine how audibility is impacted when users turn the volume down. We also wanted to see if the timing of the gain learning had an effect on the performance of speech understanding in noise.
In this investigation, the experimental group was given the gain learning feature from the first day of the fitting and the control group had the feature turned on later. There were 18 individuals in each group. The users had moderately-severe sloping sensorineural hearing loss so we could fit them with hearing aids and return audibility at least through 4000 Hz. They did have hearing loss in the low frequencies as well, because quite a bit of the gain learning feature has to do with low frequencies that people are annoyed by, so we wanted to make sure we included subjects that would require gain in the low frequencies. The groups had no previous hearing aid experience, and this was different than the other studies. The other studies were done with people that had hearing aid experience and were used to amplified sound.
We fit each listener to achieve audibility for soft, moderate, and loud sounds, while maintaining comfort. It was a double-blinded study, in that the audiologists doing all the outcome measures did not know anything about the fittings and the subjects did not know anything about what they were listening for. We used the Siemens Pure 701 hearing aids because it has input-specific and situation-specific gain learning, and the users can operate the hearing aid by remote control.
For the first four weeks, the gain learning feature was turned on for the experimental group. It was off for the control group. Then for four more weeks, until the eight week mark, the feature was on in both groups. They came back again and we repeated the outcome measures. Then we sent them out with two programs (the original gain and the "trained" gain) for ten days with a journal.
The control group only had the training feature on for one month. The experimental group had it on for two months. We did that on purpose. We also realize that to make a direct comparison in terms of how the groups might finally end up, you would need to have the control group have the feature on for another month so it would be matched. However, that was not our goal for this particular study.
We were more interested in the direct time comparisons, as opposed to controlling for how long the feature had been on. At two months out we wanted to know how people were doing depending on when we turned the feature on, not where they were when they had been given a certain amount of time with the feature.
Audibility was assessed using the calculation from the Speech Intelligibility. As we predicted, the experimental group that had the learning feature on from the beginning had less audibility by the end of the study. That means they turned it down more. They had the full fitting on the first day, and from the first day they had the ability to train the hearing aid. They turned it down more than the control group. That does not mean things became inaudible, just less audible.
There were no differences between groups on any of the other outcome measures (IOI-HA, the PAL, or the HINT. In terms of final preferences based on the journal entries, 22 people out of the 36 wanted the preferred gain setting. One thing that jumps out when we started to glance at these individual data was that there were definitely a couple of people who tuned themselves right out of audibility. We are continuing to analyze the individual data at this time.
As an audiologist, it would be worth your while to take a look at the data logging in your patients' hearing aids. Every now and then you are going to have someone who probably does more than they meant to do on any given situation, and you may have to reset that a little bit. In the Siemens hearing aid, there are some built-in constraints to prevent any wild changes that may not make sense.
In conclusion, we know that the timing of turning on the gain learning feature does have an impact on preferred gain for soft sounds at the end of the eighth week. Clearly there is a difference. The subjects who we had the feature activated on from the beginning created less audibility than a group that had become acclimated to the signal for a month before we turned the feature on. With that said, however, there was no impact on their speech performance or any of the self-report outcomes. To reiterate, this was not an equal-amount-of-time measure, which is what we wanted to do in order to simulate clinic life. Clinically, this matters more because these are often time-based decisions for people to either keep or reject hearing aids. As practicing clinicians, we are sometimes constrained by time-based decisions.
We did note individual differences, so if you are using any of these features I think it is worth looking at the data logging and the responses to make sure the patient has not overhauled their hearing aid fitting. From these data we can say that, on average, having this feature does not appear to have a negative impact on how the person is going to do, objectively or subjectively. We think it is still appropriate from a clinical perspective to have that default setting turned on at the first fitting.
Chen, Z., Hu, G., Glasberg, B.R., & Moore, B.C. (2011). A new model for calculating auditory excitation patterns and loudness for cases of cochlear hearing loss. Hearing Research, 282(1-2), 69-80.
Cox, R.M., Alexander, G.C., & Gray, G. (1999). Personality and the subjective assessment of hearing aids. Journal of the American Academy of Audiology, 10, 1-13.
Dreschler, W.A., Keidser, G., Convery, E., & Dillon, H. (2008). Client-based adjustments of hearing aid gain: the effect of different control configurations. Ear & Hearing, 29(2), 214-227.
Heinz, M.G., Issa, J.B., & Young, E.D. (2005). Auditory-nerve rate responses are inconsistent with common hypotheses for the neural correlates of loudness recruitment. Journal of the Association for Research in Otolaryngology, 6(2), 91-105.
Keidser, G., O'Brien, A., Carter, L., McLelland, M., & Yeend, I. (2008). Variation in preferred gain with experience of hearing-aid users. International Journal of Audiology, 47(10), 621-635.
Mueller, H.G., Hornsby, B., & Weber, J. (2008). Preferred hearing aid gain using trainable hearing aids. Journal of the American Academy of Audiology, 19(10), 758-773.
Mueller, H.G., & Powers, T. (2001). Consideration of auditory acclimatization in the prescriptive fitting of hearing aids. Seminars in Hearing, 22(2), 103-124.
Philibert, B., Collet, L., Vesson, J.F., & Veuillet, E. (2002). Intensity-related performances are modified by long-term hearing aid use: a functional plasticity? Hearing Research, 165(1-2), 132-151.
Zakis, J.A., Dillon, H., & McDermott, H.J. (2007). The design and evaluation of a hearing aid with trainable amplification parameters. Ear & Hearing, 28(6), 812-830.
