Probe-Mic Hearing Aid Verification: Can you afford NOT to do it? - - PDF document

probe mic hearing aid verification can you afford not to
SMART_READER_LITE
LIVE PREVIEW

Probe-Mic Hearing Aid Verification: Can you afford NOT to do it? - - PDF document

10/24/2016 Probe-Mic Hearing Aid Verification: Can you afford NOT to do it? H. Gustav Mueller Professor, Vanderbilt University, Nashville, TN. Consultant, Sivantos Group Contributing Editor, AudiologyOnline Mueller Disclosures:


slide-1
SLIDE 1

10/24/2016 1

  • H. Gustav Mueller

Professor, Vanderbilt University, Nashville, TN. Consultant, Sivantos Group Contributing Editor, AudiologyOnline

Probe-Mic Hearing Aid Verification: Can you afford NOT to do it?

Mueller Disclosures:

 Consultant, Sivantos Group  Consultant, AudiologyOnline  Author, Plural Publishing  Owner, www.EarTunes.com

Related reference material:

 http://www.audiologyonline.com/articles/20

q-today-s-use-14101

 http://www.audiologyonline.com/articles/20

q-probe-mic-measures-12410

 http://www.audiologyonline.com/audiology-

ceus/course/probe-mic-and-speech- mapping-27179

slide-2
SLIDE 2

10/24/2016 2 Probe-mic measures were in the news in back in 2009!

July, 2009: Consumer Reports talks about buying hearing aids and hearing aid fittings

What the article said:

“Audiologists made fewer serious fitting errors than did hearing-aid specialists, but in about two-thirds of all of the fittings, patients ended up with incorrect amplification.” “The provider should do several tests to verify that they [hearing aids] are working

  • ptimally. Of that battery of tests, one

stands out as a must-have: the real-ear test”

slide-3
SLIDE 3

10/24/2016 3

September, 2009: Catherine Palmer talks about ethics and the fitting of hearing aids

I f we talk about ethical practice, then we have to be comfortable saying that there are hearing health care providers who are not practicing ethically. There is not room for a statement like: I t’s okay that you aren’t following “Best Practices” by not measuring the output of the hearing aid you are fitting, and charging for your expertise but not using it. I f I ever, ever find out that one of you out there is not conducting probe-mic measures, you cannot

  • hide. I will find you, I will track you down, and . . .

Forgetting about ethics for a moment—do you know that every Best Practice Guideline written since 1990 (there are five of them) state that probe-mic measures should be used?

slide-4
SLIDE 4

10/24/2016 4

What I read on “AAA SoundOff”

“I’d like to talk to some of you

  • ut there who use probe to fit

hearing aids. I just haven’t found that it makes any difference, and I’m not sure I want to invest the money for the equipment.”

An important distinction:

Probe-mic measurements are NOT a method of fitting hearing aids; they are simply measures to verify the audiologist’s method

  • f fitting hearing aids.

Is the hearing aid fitting “good enough” to allow the patient to walk out the door (or for you to swipe the VISA card)?

verification: “substantiating or determining the truth or accuracy”

This means that we must have a gold standard—something to verify.

slide-5
SLIDE 5

10/24/2016 5 Selecting a gold standard for verification: Some considerations

 Maximize speech

understanding?

 Maximize audibility?  Maximize listening

comfort?

 Maximize sound

quality?

 Maximize patient

“first acceptance?” Today, there really are only two prescriptive fitting methods to consider, both have been systematically researched and validated:

 The Desired Sensation Level (DSL)

– Has been used since 1984, there have been several revisions, and the current version is DSL v5, available from most all manufacturers and on probe-mic equipment (but not as stand-alone software).

 The National Acoustic Laboratories (NAL)

– Has been used since 1976, many cut their teeth on the NAL-R (1986); then there was the non-linear version NAL-NL1, and today we use NAL-NL2, available from most all manufacturers, on probe-mic equipment, and there is a stand-alone version

Richard Seewald Harvey Dillon

See detailed comparison conducted by Earl Johnson, published in JAAA a couple years ago, or see “condensed version” at 20Q

slide-6
SLIDE 6

10/24/2016 6

Five audiograms used in analysis

20 40 60 80 100 100 1000 10000 Hearing Level (dB) Frequency (Hz) A-1 A-2 A-3 A-4 A-5 20 40 60 80 100 100 1000 10000 Hearing Level (dB) Frequency (Hz) A-1 A-2 A-3 A-4 A-5

NAL-NL2 versus DSL5.0 (adults) (average speech inputs)

More Gain DSL More Gain NAL

The bottom line:

 While somewhat different, for adults,

the two methods do not differ greatly, and both the NAL and the DSL are considered to be a reasonable “starting point” for the hearing aid

  • fitting. There is supporting evidence

for both (more on that later).

slide-7
SLIDE 7

10/24/2016 7

So, prescriptive methods seem to work pretty well . . . But to state the (hopefully) obvious:

The “prescription” is for the hearing aid output in earcanal SPL. Therefore, the earcanal SPL must be measured,

  • r otherwise you have no clue if you’re

even close to the prescribed gain!

Today’s probe-mic measures: What’s pretty much the same as 30 years ago?

 Verification is conducted for fitting

targets displayed on the screen.

 The probe, regulating mics and sound

delivery system is basically the same.

 The “technique” (placing the probe,

positioning the patient, etc.) is basically the same.

 This book was the first

time that all the probe- mic terminology was summarized in one source (although it has changed a little since then)

slide-8
SLIDE 8

10/24/2016 8

Today’s probe-mic measures: What’s changed

  • ver the past 30 years?

 We tend to verify using “output”

rather than “gain”

 We verify at several different input

levels

 The input signal is now speech (or

speech-like); the process is sometimes referred to as “speechmapping.”

 Today’s hearing aids have more

special features to verify

Patient’s Threshold (in earcanal SPL) Patient’s LDLs (in earcanal SPL) Average Of Amplified LTASS Fitting Targets Range Of Amplified LTASS

And, in case you’re curious, this would be considered a “good” fitting!

Fitting targets compared to patient thresholds for NAL-NL2 (left) and DSL V5 (right)

slide-9
SLIDE 9

10/24/2016 9

Reasons why the targets are below the thresholds in the high frequencies:

 Consider the complete speech spectrum, not

just the “average line.”

 You only have so much loudness to use—use

it effectively.

 There is a point where extra audibility does

not equate to increases speech understanding.

 The overall fitting must be reasonably

“comfortable” for patients, or they will turn down gain and reduce audibility for ALL frequencies.

This is not a good fitting This is an “okay fitting” (especially if patient has VC)

This verification process is conducted for several different inputs (e.g., 50, 65 and 80 dB SPL)

slide-10
SLIDE 10

10/24/2016 10 So, if probe-mic measures are:

 The intuitive thing to do . . .  The right thing to do . . .  The ethical thing to do . . .  The evidence-based thing to do . . .

Then everyone must be using them, right?

From Mueller and Picou, 2010

Three common choices used by audiologists when fitting hearing aids

?

slide-11
SLIDE 11

10/24/2016 11

The importance of audibility in successful amplification of hearing loss Ron Leavitt and Carol Flexer Hearing Review, 2012

What they did . . .

 Selected the premier product from each of

the “Big Six,” and programmed these hearing aids to each manufacturer's recommended fitting. All special features were activated.

 For benchmarking purposes, they added a

7th hearing aid—a circa 2002 single-channel analog instrument, which they programmed to NAL-NL1.

What they did . . .

 The subjects were all experienced hearing aid

users with typical downward sloping hearing losses.

 The subjects, fitted bilaterally, were tested

with all seven sets of instruments. The speech recognition test was the QuickSIN, presented at 57 dB SPL.

 Following the initial testing, all hearing aids

were re-programmed to NAL-NL1 and QuickSIN testing was repeated

slide-12
SLIDE 12

10/24/2016 12

HA-1 HA-2 HA-3 HA-4 HA-5 HA-6 OLD

Performance for the aided QuickSIN presented soundfield at 57 dB SPL. Bars indicate “SNR-Loss”: The average SNR disadvantage compared to individuals with normal hearing

So maybe things have gotten better? Or the problem is only with one

  • r two manufacturers?

Protocol for Sanders et al, 2015 (Data collected December, 2014)

  • Selected the premier hearing aid from the five

leading manufacturers.

  • Selected the manufacturer’s “default” fitting in the

software for typical downward sloping hearing loss; entered appropriate data for earmold plumbing, etc.

  • Matched all fitting and patient characteristics

between software and probe-mic equipment.

  • Conducted probe-mic measures using speech

mapping (male passage from the Verifit); 16 ears tested (8 male, 8 female)

slide-13
SLIDE 13

10/24/2016 13

Frequency (Hz) Output (dB)

Input=55 dB SPL: Real Speech of Verifit System

Output (dB) Frequency (Hz)

Input=65 dB SPL: Real Speech of Verifit System

Output (dB) Frequency (Hz)

Input=75 dB SPL: Real Speech of Verifit System

slide-14
SLIDE 14

10/24/2016 14

SIIs for Manufacturer Proprietary Fittings and NAL-NL2 targets

Speech Intelligibility Index

Output (dB) Frequency (Hz)

Input=75 dB SPL: Real Speech of Verifit System

Relating the SII (soft intputs) to speech recognition

slide-15
SLIDE 15

10/24/2016 15

Three common choices used by audiologists when fitting hearing aids

?

Probability of NAL fit when selecting “NAL” in the fitting software:

  • Aazh and Moore (2007): Programmed to the

manufacturer’s NAL using four different types of hearing aids on 42 ears. Only 36% of fittings were within +/-10 dB of NAL targets. After re- programming, 83% were within +/-10 dB (100% for hearing aids with four or more channels).

  • Aazh et al (2012): Of 51 fittings, after

programming to the manufacturer’s NAL, only 29% were within 10 dB of NAL targets; after re- programming, a match was obtained for 82% of the fittings.

So maybe things have gotten better? Or the problem is only with one or two manufacturers?

Data collected December 2014 (Sanders et al, 2015):

  • Selected the premier hearing aid from the five

leading manufacturers.

  • Selected “NAL-NL2” fit in the manufacturer’s

software; programmed for typical downward sloping hearing loss

  • Matched all fitting and patient characteristics

between software and probe-mic equipment.

  • Conducted verification using speech mapping (male

passage from the Verifit); 16 ears tested (8 male, 8 female)

slide-16
SLIDE 16

10/24/2016 16

Deviation from target (dB) Frequency (Hz)

Input=55 dB SPL: Real Speech of Verifit System

Deviation from target (dB) Frequency (Hz)

250 500 1000 1500 2000 3000 4000

Input=65 dB SPL: Real Speech of Verifit System

Deviation from target (dB) Frequency (Hz)

Input=75 dB SPL: Real Speech of Verifit System

slide-17
SLIDE 17

10/24/2016 17

Three common choices used by audiologists when fitting hearing aids

?

Some recent studies related to the NAL:

Palmer (2012) in a study with trainable hearing aids, found that when a group of new users (n=36) were fitted to the NAL-NL1, after as much of two months of training, on average, gain was only trained down a few dB (SII for soft speech reduced 2-4%; average was essentially unchaged).

Keidser and Karima (2013), in a study with trainable hearing aids, found that for a group of experienced users, on average, gain was only trained down 1-2 dB (for five different listening situations).

Abrams et al (2012) found that when the NAL-NL1 fit was compared to the manufacturer’s fit, there was significantly more benefit for the NAL fitting in real world use (based on the APHAB) and 15 of 22 users preferred the NAL fit.

Very recent (unpublished) data from Mike Valente study . . . NAL-NAL2 vs. Manufacturer’s First Fit:

 Laboratory performance for speech

recognition in noise significantly better for NAL-NL2 fitting.

 Real world self-assessment inventories

significantly better for NAL-NL2 fitting.

 After real-world trial, 19 of 24 preferred

the NAL-NL2 algorithm.

slide-18
SLIDE 18

10/24/2016 18

MarkeTrak VIII and hearing aid verification—and satisfaction?

A publication from Sergei Kochkin (2010, Hearing Review), “with a little help from his friends.”

Some of the primary purposes

  • f the survey:

 Determine overall satisfaction with

amplification

 Determine common fitting practices

(as reported by the patients)

 Determine if fitting practices influence

satisfaction

 Determine if specific aspects of

fitting/verification impact satisfaction more significantly than others. The effect of the overall protocol (# of tests administered) on patient satisfaction:

slide-19
SLIDE 19

10/24/2016 19

A Comparison of Consumer Satisfaction, Subjective Benefit, and Quality of Life Changes Associated with Traditional and Direct-mail Aid Use Sergei Kochkin Hearing Review, 21(1), 2014

Mail order hearing aids, patient satisfaction, and compliance with Best Practice protocols

What he did…

 MarkeTrak survey in 2009 of traditionally

fit hearing aids ≤ 3 years old (n = 1721).

 Survey in 2013 of customers of a very

large US direct-mail hearing aid firm (n= 2332)

 Compared traditionally fit aids to direct-

mail aids on a variety of measures including satisfaction and .

slide-20
SLIDE 20

10/24/2016 20

What he found…

Quality of Life changes (same for traditionally fit and direct-mail) vs. Best Practice

(14 items studied)

Data used by Kochkin:

The total national sample of hearing aid consumers (excluding direct mail customers) were asked to rate their HHP on seven factors using a 7-point Likert scale: “Very dissatisfied,” “Dissatisfied,” “Somewhat dissatisfied,” “Neutral” (equally satisfied and dissatisfied), “Somewhat satisfied,” “Satisfied,” and “Very satisfied.” Likert ratings on the following factors were captured:

  • Professionalism
  • Knowledge level
  • Explained care of the hearing aid
  • Explained hearing aid expectations
  • Quality of service during the hearing aid fitting
  • Quality of service post-fitting
  • Level of empathy
slide-21
SLIDE 21

10/24/2016 21

Recall from the previous slides that Kochkin also had data for:

  • What tests were administered to each

patient during the fitting process

  • The overall hearing aid satisfaction level

for each patient

Which then provides the data needed to create this chart: Some data for this practice:

  • They have been in business for 10 years.
  • Each year they sell hearing aids to around 100

patients.

  • Approximately 80% of fittings are bilateral: 180

hearing aids/year After 10 years of practice, many of their patients should be returning to replace their hearing aids.

  • But, about 20% have either died or moved away,
  • r are not in a position to obtain new hearing

aids.

  • What about the other 80%
slide-22
SLIDE 22

10/24/2016 22

Using the MarkeTrak VIII data: The impact of verification and validation:

 Verification and validation not

conducted: patient return rate=57% 57% of 80 x 1.8 = 81 hearing aids

 Verification and validation conducted:

patient return rate=84% 84% of 80 x 1.8 = 120 hearing aids Does probe-mic verification influence patient perception?

(Amlani, Pumford, and Gessling, 2016)

The authors collected data on three groups of 20 (n = 60) adult listeners with mild to moderately-severe sensorineural hearing loss: – Group 1: Experienced users of amplification (> 1 year of use, 8 hours of daily use) – Group 2: Owner of hearing aids who did not use their devices on a frequent basis – Group 3: Non-users who experienced hearing difficulties and were interested in a trial period.

Ten listeners in each group were fitted using the manufacturer’s First-Fit NAL-NL2/Quick-Fit procedure, while the remaining 10 listeners in each group were fitted using a standard probe- microphone verification protocol that included REARs fitted to multiple levels and MPO verification.

Does probe-mic verification influence patient perception?

(Amlani, Pumford, and Gessling, 2016)

During the fitting process with each group, listeners had an

  • pportunity to influence the final settings by rating the comfort and

clarity of the experimental devices until a rating of 8 on a 10 point scale (i.e., 1 = lowest 10 = highest) was achieved while listening to a passage of continuous discourse.

During and after the fitting process, each subject was counseled on the procedures being performed by the same experienced clinician.

The participants provided survey responses regarding (1) willingness-to-pay (WTP), price anchored at $250, for professional services, and (2) a Perceived Value Measurement designed to assess a respondent’s attitude and behavior toward professional services assessed in five dimensions: perceived quality, perceived value, behavioral intent, emotion, and price.

slide-23
SLIDE 23

10/24/2016 23

Willingness to pay findings: Satisfaction measures

Yet another reason to do probe- mic measures! How many reasons do people need?

Some questions from our workshops?

slide-24
SLIDE 24

10/24/2016 24

Questions from

  • ur workshops

“Whenever I fit frequency compression, the patients don’t seem to hear any difference. Why is that?”

Case #1

Compress this region

Case #2

slide-25
SLIDE 25

10/24/2016 25

Case #3

Conventional probe-mic measures tell you that something happened, but do not reveal if audibility is present for the lowered signal:

Example of specialized test signal (4000 Hz)

Green = Standard LTASS (carrot passage) Pink = 4000 Hz test signal

slide-26
SLIDE 26

10/24/2016 26

Example of specialized test signal (6300 Hz)

Green = Standard LTASS (carrot passage) Blue = 6300 Hz test signal

Illustrative Case #1

Compress this region 

Fit to NAL-NL2 targets for 65-dB real speech input

slide-27
SLIDE 27

10/24/2016 27

After some tweaking: Audibility of compressed signal

 Green = 6200 Hz FCo “Off” Pink = 6200 Hz FCo “On”

Some manufacturers have a calibrated recording

  • f the Ling 6-Sound Test on their equipment,

which can then be presented to the patient.

Why it’s critical that this feature is fitted using probe-mic verification!

Patient’s Thresholds Manufacturer’s Default Fitting After Tweaking

slide-28
SLIDE 28

10/24/2016 28

Questions from

  • ur workshops

“I’ve been told I’m should do something “special” when I test OC fittings, but I’m not entirely sure what to do or why I’m doing it?”

Potential equalization problem with OC fittings

 Sound leaking out of ear is picked up by

reference mic

 Sound leaking out of ear may be greater

than the input to reference mic from loudspeaker

 Reference mic thinks it is output from

loudspeaker, and so loudspeaker output to ear is then turned down

 The result will be less measured hearing aid

  • utput (and gain)

Magnitude of problem probably related to:

 Gain/output of hearing aid  Feedback reduction algorithm  Location of reference mic  Proximity of reference mic to hearing

aid mic

 Openness of the fit  Residual ear canal resonance of the

patient

slide-29
SLIDE 29

10/24/2016 29

Typical mistake expected for hearing aid gain ~20 dB—concurrent versus stored equalization Note what happens when you increase gain in the highs!

(input = real speech @ 65 dB SPL; hearing aid gain ~34 dB)

Questions from

  • ur workshops

“When using speechmapping for target match, how close of a match is necessary? 3 dB? 5 dB?”

slide-30
SLIDE 30

10/24/2016 30

When thinking about “how close is close enough,” three things to consider:

 The preciseness of the target

calculations

 The variability of individual loudness

preferences

 The overall shape of the amplified

speech spectrum For speech mapping, we need to convert from “dB HL” to “dB earcanal SPL”

Where did these numbers come from? Threshold in HL + RETSPL + AVERAGE RECD

So let’s look at adult RECDs—here are some data (n=1814 ears, 904 subjects, 69% male) from Saunders and Morgan (2003)

Mean=3.4 (s.d.=3.1) Mean=-0.3 (s.d.=5.4) Mean=6.3 (s.d.=3.5) Mean=10.3 (s.d.=8.9)

slide-31
SLIDE 31

10/24/2016 31

When thinking about “how close is close enough,” three things to consider:

 The preciseness of the target

calculations

 The variability of individual loudness

preferences

 The overall shape of the amplified

speech spectrum

The work from the NAL revealed that there is a large range of preferred gain for inviduals with the same hearing loss

5% Too Soft 49% Just Right 46% Too Loud

When thinking about “how close is close enough,” three things to consider:

 The preciseness of the target

calculations

 The variability of individual loudness

preferences

 The overall shape of the amplified

speech spectrum

slide-32
SLIDE 32

10/24/2016 32

Not all target misses are equal!

Questions from

  • ur workshops

“Sometimes when I present the soft speech signal, the output is below the patient’s threshold, but yet the patient says he hears it. Why is that?”

It’s good to look at the entire speech spectrum, not just the “average” line

We could turn up gain by 10 dB and the “average line” would still be below threshold, but 30-40% of the speech signal would be audible Dark line represents “average” of soft speech

slide-33
SLIDE 33

10/24/2016 33

Questions from

  • ur workshops

“With some patients, the NAL target is actually below the patient’s threshold. This can’t be

  • right. Don’t we want audibility?”

Questions from

  • ur workshops

“I use DSL for fitting to target. It seems like I get different targets for the same audiogram—is this possible?”

The effects of LDLs on DSL target gain and target match

Match to target when average LDLs used Match to target when measured LDLs used

(hearing aid programming not changed)

slide-34
SLIDE 34

10/24/2016 34

Questions from

  • ur workshops

“My boss is sort of a “nut” about probe-mic measures. Whenever speechmapping doesn’t look quite right, she wants us to also do insertion gain to see if speechmapping is really correct. Does this seem reasonable?”

Comparison of REIG vs. speechmapping for manufacturer’s “First Fit” response: Same hearing aid, same patient; two different verification procedures.

Insertion Gain REAR Target for 65 dB input Target for 65 dB input 23 minus 8 = 15 dB 81 minus 68 = 13 dB

Questions from

  • ur workshops

“I seem to always have problems hitting target for 4000 Hz. My colleagues say they don’t have the same problem, and they are fitting the same hearing aids. Am I doing something wrong?

slide-35
SLIDE 35

10/24/2016 35

Probe tube 2-3 mm too shallow

Match to target with probe at correct depth

Probe tube pulled back another 3 mm

Questions from

  • ur workshops

“I was told by my rep that I should use “live voice” for fitting hearing

  • aids. What do you think of that?”
slide-36
SLIDE 36

10/24/2016 36

Thomas Edison summed it up when he said: Questions from

  • ur workshops

“I totally don’t understand the amplified speech spectrum I see when I test hearing aids using speechmapping. I have compression set pretty high, so why isn’t the spectrum “squashed” like it is supposed to be?” Common effects of compression

slide-37
SLIDE 37

10/24/2016 37

Common effects of compression

The effects of release time: 60 msec vs. 1800 msec for a 3:1 ratio

Short release times Long release times

Some “interesting” questions from Rush University students . . .

slide-38
SLIDE 38

10/24/2016 38

My supervisor:

Conducts probe-mic testing, but doesn’t put any targets on the screen.

My supervisor:

Says not to pay attention to targets—just get an aided SII

  • f .70 or so and you’re okay.

Output (dB) Frequency (Hz)

Input=75 dB SPL: Real Speech of Verifit System

slide-39
SLIDE 39

10/24/2016 39

My supervisor:

Uses the old NAL-NL1 rather than NL2 because she says that she gets better matches to target.

My supervisor:

Says you should mostly ignore the targets, and just make the aided speech above the patient’s thresholds.

My supervisor:

Starts off with the manufacturer’s default fitting, but then adjusts gain and output to match NAL-NL2 targets.

slide-40
SLIDE 40

10/24/2016 40

My supervisor:

Says that the special features

  • f the hearing aid only will

work if you fit to the manufacturer’s default fitting, so no need to do probe-mic.

My supervisor:

Says that if you get a close match for average, no need to do soft or loud as they will be okay.

Output (dB) Frequency (Hz)

Input=65 dB SPL: Real Speech of Verifit System

slide-41
SLIDE 41

10/24/2016 41

Frequency (Hz) Output (dB)

Input=55 dB SPL: Real Speech of Verifit System

Output (dB) Frequency (Hz)

Input=75 dB SPL: Real Speech of Verifit System

  • H. Gustav Mueller

Professor, Vanderbilt University, Nashville, TN. Consultant, Sivantos Group Contributing Editor, AudiologyOnline

Probe-Mic Hearing Aid Verification: Can you afford NOT to do it?