Question
For a long time we have used pure tone sweeps for real ear measurements, but now the signal of choice is speech, or modulated speech-like noise. Theoretically the use of a pure tone sweep could result in softer aspects of speech to be missed if we use this signal to generate a fitting target. Please can you tell me how big of a deal it is if you use a pure tone versus modulated speech signal? Is it absolutely wrong or could it be "good enough"? Also, from a functional perspective, do we have any evidence to suggest that hearing aid outcomes are different just due to using these different stimulus types?
Answer
First you should be congratulated for using real ear measurements. You are definitely in the minority. In spite of all the evidence that supports probe microphone verification, audiologists and dispensers have not embraced the only truly objective way to verify what a hearing aid does in the ear canal. To answer your first question, it is a great big deal! The frequency, intensity, duration and modulation characteristics of speech have very little in common with pure tone sweeps. Modern nonlinear digital hearing aids will not respond to different signals in the same way. To illustrate, see the figure below. Using two speech like signals (ANSI S3.42 vs. ICRA), you can see that there is at least a 10dB difference in output at all frequencies above approximately 1.5 KHZ. If two speech like signals result in such different curves, you should expect that curves comparing speech and pure tones would be even more divergent. So, using pure tones instead of speech or speech like signals to verify nonlinear digital hearing aids is not good enough. 
(Figure reproduced with kind permission of Sallie Frye, Frye Electronics)
Your second question is not easy to answer directly. I am not aware of any published data that suggest people are more satisfied when their hearing aid fitting is verified with probe microphone measures. Dillon (2006, 2003, 1999) has reported that patients do better when the fitting approximates the target and that some prescriptions result in higher speech recognition than others. If 'doing better' means improved speech recognition then the answer to your second question is also yes. Using speech or a speech-like signal will result in a better approximation of the prescribed target and should therefore result in better outcomes.
Anecdotally, Mayo Clinic Arizona is the winter destination for people from all over the world. Working there provided the opportunity to see many winter visitors who had been fit with hearing aids elsewhere and who came to the clinic for a second opinion. Since probe microphone measures were standard practice at Mayo, we often found that people were wearing hearing aids that were frankly not fit well. A simple adjustment in frequency/gain response based on probe microphone findings often resulted in a delighted user who very often asked, "Why didn't they just do that on me when they fit the hearing aid?" That is a great question! Please continue to use probe microphone verification and encourage your colleagues to do so as well. Keep up the good work and good luck.
Suggested readings:
https://frye.com/manuals/7000.pdf
https://www.audiology.org/resources/documentlibrary/Documents/haguidelines.pdf
Dillon, H. (1999). NAL-NL1: A new procedure for fitting non-linear hearing aids. Hearing Journal, 52(4), 10-16.
Dillon, H. (2006). What's new from NAL in hearing aid prescriptions? Hearing Journal,50(10),10-16.
Dillon, H, Keidser, G. (2006). Is probe-mic measurement of HA gain-frequency response best practice? Hearing Journal, 56(4), 28-30.
Dr. Steve Huart has over 20 years experience fitting hearing aids and is an outspoken advocate for objective verification of hearing aid fittings using probe microphone measures. He was a clinical and dispensing audiologist at Mayo Clinic in Arizona before joining Cochlear Americas in 2007. He can be reached at shuart@cochlear.com.
