Clinical Application of the SADL Scale in Private Practice II

J Am Acad Audiol 12 : 15-36 (2001)
Clinical Application of the SADL
Scale in Private Practice 11 :
Predictive Validity of Fitting Variables
Holly Hosford-Dunn*
Jerry Halpern'
Abstract
Predictive validity of 44 independent variables and their interactions with Satisfaction with
Amplification in Daily Life (SADL) scores was assessed . SADL scores were influenced by
patient age, years of hearing aid experience, hours of use per day, perceived hearing difficulty, pure-tone average, hearing aid style, processor type, and manufacturer's invoice cost .
The relative importance of these variables to SADL measures was complex and very small,
but the variables and their squares and interactions improved r2 predictions of SADL Global
and subscale scores in a separate stepwise multiple linear regression procedure by 12 to 33
percent compared to SADL norms alone. More research with additional variables is needed
to develop a clinically useful model for predicting wearer satisfaction . Clinically, SADL scores
yield subscale-specific patterns of satisfaction and dissatisfaction that help in intervention
planning and serve as graphic "snapshots" of satisfaction status . A series of patient profiles
are presented illustrating the potential usefulness of the SADL in predicting hearing aid satisfaction. With its good construct and psychometric properties, the SADL could serve as a
gold standard for satisfaction outcomes and a basis for development of a predictive model
of hearing aid fitting success.
Key Words : Hearing aids, outcome, satisfaction, Satisfaction with Amplification in Daily Life,
validity
Abbreviations: ANOVA = analysis of variance, BTE = behind the ear, CIC = completely in
the canal, DSP = digital signal processing, ITC = in the canal, ITE = in the ear, PP-SADL =
private practice SADL group, PTA = pure-tone average, SADL = Satisfaction with Amplification
in Daily Life
art I of this study verified that the Satisfaction with Amplification in Daily Life
(SADL) scale quantifies hearing aid
users' responses in content domains that contribute to SADL scores in predictable ways (Hosford-Dunn and Halpern, 2000) . The SADL's
sound construct validity and psychometric properties enable its use as an outcome measure of
hearing aid satisfaction in a variety of settings,
with a wide range of patients, over periods of
months to years postfitting (Cox and Alexander,
1999 ; Hosford-Dunn and Halpern, 2000).
The SADL profiles satisfaction domains in
four subscales containing a total of 15 items. The
P
*Tai Inc ., Tucson, Arizona ; tDepartment of Biostatistics,
Stanford University, Stanford, California
Reprint requests : Holly Hosford-Dunn, TAI, Inc ., PO
Box 32168, Tucson, AZ 85751
most important, and heaviest weighted, subscale is Positive Effect, which is complemented
by the Service/Cost subscale . Together, these
two subscales contain nine items that address
benefit/value content areas. The remaining subscales are Negative Features, which quantifies
problems of patient-technology mismatch, and
Positive Image, which combines cosmetic and
stigmatizing concerns . These two subscales have
lesser weighting in the SADL because they represent content areas that are important to some,
but not all, patients . However, both subscales
contain one item each that is highly important
to satisfaction . Telephone use and appearance
are elements that influence satisfaction for
almost all users.
A number of variables affect satisfaction in
one or more domains (Hosford-Dunn and Baxter,
1985 ; Smedley, 1990 ; Dillon et al, 1997 ; Humes,
1999) in ways that remain unclear. Humes (1999)
15
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
concluded that satisfaction "may depend on a
complex combination of severity of hearing loss,
perceived handicap . . ., aided sound quality, reliability . . .of the instruments, and the personality
of the wearer" (p . 38). With that shopping list in
mind, the SADL may be a candidate for ferreting
out complex relationships ofpsychological, demographic, and technical variables with satisfaction domains. No data are published relating
SADL elements to these classes of independent
variables. For instance, statistically higher Negative Features subscale scores in a private practice population (Hosford-Dunn and Halpern,
2000) may be due to the conservative estimate provided by Cox and Alexander (1999) or may represent effects of environmental, patient, or
technology variables (e.g ., practice procedures,
degree of hearing loss, multichannel processing) .
The importance of identifying relations
between variables and SADL scores lies in the
idea of developing tools and techniques to
improve preassessment needs and predict treatment outcomes . By knowing which variables
affect satisfaction in what ways, it might be
possible to predict hearing aid user satisfaction
with some degree of certainty prior to fitting
while using the SADL to verify satisfaction at
periodic intervals subsequent to treatment.
The sequential goals of Part II of this study
were to
0
Identify effects of a select group of independent variables, and their interactions, on
SADL scores ;
Evaluate SADL scores for different hearing
aid types, if significant hearing aid influences
were discovered in the first goal, and compare
these scores to interim norms proposed by
Cox and Alexander (1999) ; and
Investigate the possibility of using the relationships identified in goal 1 to start developing
fitting profiles that could serve eventually as
generalized predictors of positive hearing aid
outcome in a variety of dispensing settings .
METHOD
S
ubjects, materials, and procedures were
described in detail in Part I (Hosford-Dunn
and Halpern, 2000). To summarize, all patients
fitted with hearing aids in 1996 and 1997 at a single private practice were asked to complete the
15-item SADL scale at 1-year postfitting. Those
who complied (referred to as "the PP-SADL
group") constitute the subjects in Parts I and II
of this study. Subjects in the PP-SADL group
16
represented a wide range of ages, degrees of
hearing impairment and disability, hearing aid
use, and hearing aid experience . Patients also
completed four multiple-choice items on the
SADL regarding years of hearing aid experience, daily hearing aid use, and perceived degree
of unaided hearing difficulty. These demographics
were treated as subjective independent variables in the analyses .
Procedures were incorporated into the
patients' regular office visits to encourage participation . Hearing aids were selected and fitted
in a series of appointments in which individual
participants received services from one of three
audiologists employed in the private practice
(including the first author). The evaluation and
fitting appointment were 90 minutes each .
Follow-up appointments were 30 minutes each .
The first follow-up was between 48-hours to
1-week postfitting, depending on hearing aid
style (e .g., completely-in-the-canal [CIC1 fittings
were scheduled for two appointments in the first
week, the first at 48-hours postfitting and the second at 1-week postfitting) . Weekly follow-up
appointments continued until the patient
expressed satisfaction with amplification in daily
life . Subsequent follow-up appointments were
scheduled quarterly or biannually. Patients were
scheduled for annual appointments to test hearing and review hearing aid fittings .
Instrument Selection
Hearing aid recommendations used a clientcentered approach of "consultative selling"
(Sweetow, 1999) in which audiologists used a
tiered technology chart illustrating three levels
of technology (conventional, analog programmable, and digital), which were described in
terms of circuitry, style, application, and price
range. Tier assignments were made according to
the following operational definitions :
Tier I. Single-channel linear or compression instruments that were not adjustable using
Noah-based manufacturers' software modules .
Tier IL Analog instruments that included
multichannel or multimemory processing and/or
were adjusted using Noah-based fitting technology.
Tier III. Instruments with digital signal
processing (DSP).
Four styles were defined: behind the ear
(BTE), in the ear (ITE), in the canal (ITC), and
CIC .
The tier chart was updated frequently during 1996 and 1997 to reflect technology changes
SADL Validation II/Hosford-Dunn and Halpern
and additions . A typical tier chart is shown in
Appendix A . Appendix B lists the instruments
used in the study, grouped by technology tier .
Audiologists used the technology tier chart for
sequential purposes : (1) to explain which hearing aid styles were available in each technology level (e .g ., DSP CIC hearing aids were not
available until June 1997) ; (2) to advise patients
on which devices and styles matched their
hearing and listening needs best ; (3) to discuss the pros and cons of selections at each
technology level, including performance and
price considerations associated with monau-
were extracted from patients'responses to demographic questions on the SADL and from patient
records. They were categorized as intrinsic or
extrinsic, as shown in Table 1. Intrinsic variables
described characteristics of the patients, and
extrinsic variables were specific to the instruments and fitting histories . Two conventions
were adopted in the Results and Discussion sections for purposes of clarity: (1) intrinsic variables
are referred to as intrinsic or patient related;
extrinsic variables are referred to as extrinsic or
hearing aid related; and (2) because the statistical analyses included so many comparisons
that some might be significant by chance, we
elected to refer to a correlation as "significant"
only if the associated p value was less than .005 .
Hearing aid selection proceeded by establishing whether the fit should be binaural or
monaural and then by indicating particular
instruments at different levels of the tier sheet
and rank ordering them according to "best fit"
styles . The technology tier chart served as a
selection guide, with the patient making an educated purchase based on the audiologist's recommendations . Patients did not always choose
the top-ranked recommendation, usually due
to cosmetic or financial reasons, nor did they
always allow a binaural fitting when it was recommended . The tier sheet was inserted in each
patient's chart in case original recommendations needed to be revisited during subsequent
fitting and follow-up appointments .
RESULTS
ral and binaural fittings ; and (4) to ensure that
patients' expectations were in line with selected
technologies, styles, costs, and monaural/
binaural fitting recommendations .
Data Analysis
Data were stored and analyzed on the Stanford University Sun Sparc Ultra II computer,
Solaris 2 .6 Operating System, using programs
from the Statistical Analysis System . Crosscorrelation, analysis of variance (ANOVA), stepwise multivariant analyses, and logistic
regression procedures analyzed data .
The PP-SADL group consisted of patients in
the cohort who returned usable SADL scales . Criteria for acceptable SADL questionnaires were
described in Part I (Hosford-Dunn and Halpern,
2000). Of 282 returned scales, 257 were used for
Global scores, 275 for the Positive Effect subscale, 274 for Service/Cost, 271 for Negative
Features, and 270 for Positive Effect .
Dependent variables were Global, Positive
Effect, Service/Cost, Negative Features, and
Positive Image SADL scores, as well as scores
on the 15 SADL items. The independent variables selected for examination in this study
ge, gender, and audiometric characterisA tics of the cohort and PP-SADL group are
described in Part I of this study (Hosford-Dunn
and Halpern, 2000). The PP-SADL group (68 .7%
of the cohort) was similar to the population in
most respects but differed according to some
hearing aid-related variables . The PP-SADL
participants used higher priced (t = -2 .92),
higher tier technology (t = -2 .87), binaural fittings (t = -3 .05) more than the total cohort
(p < .005 for each). They also returned for more
hearing aid follow-up checks (t =-6.01, p < .005).
Table 1
Type
Intrinsic
Summary of Dependent Variables
Description
Patient-related variables
Gender
Age
Total hearing aid experience (yr)
Daily use (hr)
Perceived hearing difficulty (unaided)
Pure-tone average (four frequency, left and
right ears averaged)
Monaural/binaural status
Extrinsic Hearing aid-related variables
Hearing aid style (BTE, ITE, ITC, CIC)
Technology tier (conventional,
programmable, DSP)
Number of processing channels (1, 2, or 3)
Invoice cost
Retail price
Number of repairs (in first year)
Total visits (number of clinical appointments)
Total time (sum of clinical appointments in
minutes
17
Journal of the American Academy of Audiology/ Volume 12, Number 1, January 2001
Table 2 Number of Patients Fit in
Each Instrument Category (N = 374)
Conventional Programmable
BTE
ITE
ITC
CIC
Total
12
35
58
21
126(34%)
56
30
47
20
153(41%)
DSP
Total (%)
29
17
31
18
95(25%)
97 (26)
82 (22)
136 (36)
59 (16)
374 (100)
Hearing Aid Selections
The distribution of fittings for different combinations of hearing aid technology and styles
are shown in Table 2 for the cohort . Proportions
were similar whether the data were viewed as
total instruments or total fittings (Fig . lA and
1B), indicating that binaural or monaural fittings
did not occur disproportionately in any technology or style category. One-third of the instruments fitted in this study were conventional
(Tier 1), 41 percent were programmable (Tier II),
and one-quarter were DSP (Tier III) . Fittings
were fairly evenly distributed among styles . The
most common style was ITC (38%). Almost half
of the instruments were BTE or ITE. Sixteen percent of the instruments were CIC.
Table 3 and Figure 2 summarize PP-SADL
group response rates according to hearing aid
style and technology categories . Logistic regression analysis showed a statistical difference
(p = .016) in response rates among the three technology levels, indicating that the response rate
of DSP users was significantly higher than that
of patients who wore conventional aids . However,
there was no strong evidence of true differences
between the conventional and programmable
or between the programmable and DSP response
rates.
Relation of Intrinsic
and Extrinsic Variables
Correlation analyses were performed for all
dependent variables in Table 1. The following
variables were excluded from subsequent analyses because they did not correlate significantly
with any dependent variables in the study
(p > .005) : gender, monaural/binaural status,'
number of independent processing channels
(i .e ., one to three), retail price, number of repairs,
total visits, and total time .
Table 4 examines the relationships of the
remaining variables to each other. Intrinsic vari18
ables in Table 4 correlated significantly (p < .005
or . 001) with one another except age, which correlated only with four-frequency pure-tone average (PTA) (500, 1000, 2000, 3000 Hz).2,3 These
intrinsic correlations indicated that patients
with greater hearing loss reported more hearing difficulty, more years of hearing aid use,
and greater daily wearing time .
Extrinsic variables were uncorrelated, with
a few predictable exceptions : (1) higher technology tiers had higher invoice costs and
(2) smaller, more expensive instruments had
more first-year repairs. Style and technology
were unrelated, implying that technology selection did not drive style selection or vice versa.
The number of hearing aid changes during the
trial period did not differ for different hearing
aid styles or technology tiers.
The only variable that linked patients to
instrumentation was style, which correlated
significantly and negatively with all intrinsic
variables . The correlations with style in Table
4 indicate that smaller instruments were fitted
on younger patients with less hearing loss and
hearing aid experience, who reported less hearing difficulty and used their instruments fewer
hours per day. The relationship between age
and repairs is due to style: older patients tended
to wear larger hearing aids, which had fewer
first-year repairs .
Variable Influences on SADL Scores
Univariate Analyses
Significant correlation coefficients between
dependent and independent variables are shown
in Tables 5 and 6. One-way ANOVA procedures
'Kochkin (2000) has reported a 3 percent improvement in overall satisfaction for binaural wearers, based on
national survey results . Our variable is not directly comparable and undoubtedly underestimates the effect of a
binaural effect because it is based on the number of aids
fitted, so that binaural users who purchased one instrument were coded as monaural fittings .
2 PTA was selected as the variable of choice for hearing loss because hearing loss effects were not frequency
dependent when audiometric thresholds at 500 to 4000 Hz
were correlated with other study variables .
'Functional gain at 500 to 4000 Hz also correlated with
these patient variables, independent of test frequency, but
did not influence SADL scores systematically and was not
used as an extrinsic variable in this study. Similarly, Beamer
et al (in preparation) found no predictive value of National
Acoustic Laboratories' target gain match on perceived
hearing aid benefit .
SADL Validation II/Hosford-Dunn and Halpern
A
B
# Aids
# Aids
# Fittings
# Fittings
Tier
I
Tier
II
Tier
III
Figure l
BTE
ITE
ITC
~_ CIC
Distribution of hearing aids (n = 608) and hearing aid fittings (n = 374) by technology tier (A) and style (B).
19
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
Table 3 Distribution of Hearing Aid Fittings
for Patients Who Returned SAM Scales
BTE
ITE
ITC
CIC
Total
Conventional
Programmable
4
25
42
16
87(69%)
37
24
36
17
114(75%)
DSP
Total (51o)
26
67(69)
15
64(78)
26
104(76)
14
47(80)
81 (85%) 282(75%)
Percent response rate for each aid category shown in
parentheses.
were performed when appropriate and indicated. ANOVA results are described in the text.
Global Scores
Global satisfaction correlated with age . Age
had small negative effects on Global and Positive Effect scores (see Table 5) and on telephone
use (see Table 6) . These effects were not artifacts
of the skewed age distribution of subjects because
t-tests of Global and subscale scores for older
(>59 years old, n = 253) and younger (<60 years
old, n = 22) subjects were not significantly different (p > .29) . One-way ANOVA showed a significant effect of style on Global satisfaction
(F3,253 = 3 .167, p = .025) in which smaller instruments were associated with higher Global scores
(Figure 3) .
Subscale and Constituent Item Scores
Variables in this study affected SADL scores
in a dichotomous manner, as shown in Table 5
by the arrangement of intrinsic variables in the
top and extrinsic variables in the bottom portions
of the chart. Intrinsic variables correlated with
Table 4
Age
Total hearing aid experience
Hours/day
Perceived hearing difficulty
Pure-tone average
Style
Technology tiers
Invoice
Repairs
Aid changes
20
Style
Figure 2 Percentage of patients who completed the
SADL, according to hearing aid technology tier and hearing aid style (n = 282 respondents) .
the Positive Effect subscale and extrinsic variables correlated with the Personal Image subscale . Only the Negative Features subscale
correlations reflected influences of both intrinsic and extrinsic variables. The effects were not
great: none of the correlations was large, and
Service/Cost did not correlate with any variables in the study.
A further dichotomy was that intrinsic variables correlated with either Positive Effect or
Negative Features subscales, but not both . Those
patients who reported more daily use and greater
perceived hearing difficulty also reported higher
Positive Effect, manifest by significant correlations of these variables with most items in that
subscale (see Table 6) . Patients who reported
more years of hearing aid use and had greater
audiometric hearing loss reported lower Negative Features scores due to increased dissatisfaction with feedback and background noise.
Spearman's Rho Correlation Coefficients Between Study Variables
ge
*p < .005 ; **p < .001 .
Tier
Total
Hearing
Aid
Experience
Hours/
Day
0 .25**
0 .20*
0 .42**
-0 .25**
-
0 .41 **
0 .26**
-0 .28**
-
-
-
Perceived
PureHearing
Tone
Difficulty Average
Style
Technology
Tiers
Invoice
0.26**
0 .75**
-
0 .18**
0 .22**
-0 .16*
-0 .15*
-
0 .52**
-0 .41**
-0 .46**
SADL Validation ll/Hosford-Dunn and Halpern
Table 5
Spearman's Rho Correlation Coefficients Between SADL Subscales and Related Variables
Variables
Intrinsic
Age
Total hearing aid experience
Global
Positive Effect
-0 .18*
-0 .22*
Service/Cost
Negative Feature
-0.23*
033*
0 .25*
Hours/day
Perceived hearing difficulty
Pure-tone average
Personal Image
-0 .29**
Extrinsic
Style
0 .31**
Technology tiers
0 .25**
0 .19*
Invoice
*p < 005; **p < .001 .
Style emerged as the key extrinsic variable
for predicting SADL scores, manifest in the Negative Features and Personal Image subscales.
Five of the six items in these two subscales
showed small but significant correlations with
style (see Table 6), and item 4 barely missed significance (p = .006) by the criterion of this study.
The meaning of these relationships is illustrated
in Figure 4, showing less dissatisfaction with
Negative Features and more satisfaction with
Personal Image for smaller instrument designs.
The relationship of size with satisfaction in
these subscales was confirmed by one-way
ANOVAs (Negative Features : F,, 267 = 9 .49,
p = .0001 ; Positive Image: F,, 266 = 729, p = .0001).
Figure 5 shows the effects of style on items
in the Negative Features and Positive Image subscales, where larger instruments (BTE and ITE)
produced lower satisfactions than smaller instruments (ITC and CIC) on all items. One-way
Table 6 Spearman's Rho Correlation Coefficients Between
Study Variables and SADL Items, Organized by Subsale
ge
Positive Effect
Item 1
Item 3
Item 5
Item 6
Item 9
Item 10
Total
Hearing
Aid
Experience
Hours/
Day
0 .25**
0 .22*
0 .32**
0 .33**
0 .3**
-0 18*
-0 .24**
-
-0 .17*
0 .21 **
Perceived
Hearing
Loss
-0 .19*
PureTone
Average
0.31**
0.21**
0 .31 **
0 .30**
0 .20**
0 .27**
it
nvoice
Aid
epairs Changes
Service/Cost
Item 12
Item 14
Item 15
-0 .32**
Negative Features
Item 2
Item 7
Item 11
-0 .21 *
0.30**
-
-0 .20**
0 .24**
-
-
-
0.25**
0.18*
-0 .27**
-0 .28**
0.20*
-
Personal Image
Item 4
Item 8
Item 13
0 .20*
0.19*
-0 .20*
-
*p< .005 :**p< .001 .
21
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
5.4
L 5. 2
O
v
N
significant . Invoice cost, number of repairs, and
number of hearing aid changes had small but significant effects on subscale scores . Higher-cost
instruments were associated with higher Positive Image subscale scores (see Table 5) and
more satisfaction with telephone use (see Table
6) . The only other significant correlations were
in areas that are predictable from clinical experience . Patients who had more hearing aid
repairs were less impressed with the dependability of their instruments (item 15). Patients
who exchanged their instruments for other styles
or technology tier types during their trial periods expressed less satisfaction with the appearance of their instruments (item 8) .
+1SE
Mean
1SE
5. 0
O
0 4.8
4.6
45
4;r
4~V
CK
~~~
~~'
Figure 3 Mean SADL Global scores and standard
errors for four hearing aid styles (n = 257 SADL scores).
ANOVAs for each item, classified by style, were
significant (p < .05 or less), but the main effects
were due to one item in each subscale . Both
telephone use (item 11) and appearance (item 8)
were rated significantly higher by CIC users
than users of ITC, ITE, or BTE instruments
(F3,251 = 5 .97, p = .0006 ; F3,264 = 5 .11, p = .0019) .
Technology tiers did not correlate with SADL
scores or item scores, nor were one-way ANOVAs
between technology tiers and SADL subscales
BTE (n=64.66)
0 ITE.(n=62-63)
ITC (n=99-100)
OCi
®
~
^"~
Q02
J
V0
. G`
mt'
0
~a1
~0
Wr
;~
OC
~~
Q
Figure 4 Mean SADL subscale scores for four hearing
aid styles .
22
Multivariate and Factor Analyses
Aprincipal research question was whether
combinations of the independent variables in our
study affected SADL scores in measurable ways .
As a first step in this direction, we looked at the
isolated effects of style and technology tiers to
see if the lack of effect of technology tiers in
univariant analyses was due to a confounding
of size and circuitry. Two-way ANOVA for technology tiers X style without interaction paralleled the findings reported in the previous section
for style alone, indicating that size influenced satisfaction more than processing on SADL measures. Significant relationships due to style, but
not technology tiers, were found for Global
(F5,251 = 3 .51, p = .01) and Negative Features
(F5,265 = 10 .11, p = .0001) scores . Despite the
independence of style, the two-way ANOVA did
show a secondary effect of technology tiers on
Personal Image scores, which were significantly
better for DSP and programmable instruments
(Tiers III and II) compared with conventional
(Tier I) instruments (F test for style F3,264 = 8.36,
p < .0001; F test for technology tiers: F5,264 = 2.95,
p = .05; omnibus F test for the model: F5,264 = 5.71,
p = .0001) .
A separate stepwise multiple linear regression procedure was used in the following manner to determine whether selected variables
contributed to satisfaction in more complex ways .
Ideally, the procedure should use all variables to
produce a predictive equation that explains how
the variables, their squares, and their interactions may be used in a predictive mode . However,
preliminary work showed that using all of the
variables in the model produced results that
served to confuse, rather than clarify, the clinical picture provided by the SADL. Therefore, we
limited the model to specific variables that came
SADL Validation II/Hosford-Dunn and Halpern
Item 2
Item 7
Item 11
ITC (n=78-101)
E BTE (n=51-67)
ITE (n=52-63) / CIC (n=32-46)
Figure 5
Item 4
Item 8
Item 13
ITC (n=93-101)
BTE (n=56-64)
(n=58-61)
/
CIC (n=39-44)
C1 ITE
Mean scores for four hearing aid styles on Negative Feature items (A) and Personal Image items (B) .
to our attention as the result of the lower-order,
single-variant analyses (see Tables 5 and 6) . We
also excluded aid changes and repairs because
these variables represented rare events . The
intent was to get an idea of which universal variables remained important in the presence of the
other variables (i .e ., gave information in addition
to the others). The model for the stepwise backward selection procedure contained the eight
variables from Table 5, their linear and quadratic combinations including squares, for a total
of 44 variables.' The model was as follows:
Global = age, PTA, perceived hearing difficulty, daily use, total hearing aid experience, style, technology tiers, invoice, and
the squares and pairwise products (interactions) of these eight variables.
Table 7 gives the results of the procedure for
Global and SADL subscores . In each case, no
more than seven variables remained when the
criterion for remaining in the model was p < .05 .
All eight of the original variables selected for the
model contributed to the results-as a linear
term, its square, or in combination with other
linear terms . RI values were too small to develop
a predictive model, but some observations about
the data are worth noting. As Table 7 shows, the
'Our goal in pursuing this analytic approach was to
try to understand contributions by allowing variables and
their squares and interactions . We did not require a hierarchical model with linear terms but allowed quadratic
variables to appear without their linear constituents .
relative importance of the variables to SADL
measures was complex and very small. However,
the significance of the p values is strong evidence
that there are relationships and that it is worthwhile to investigate the nature of the relationships further.
Age and its square were the most important
variables for three SADL measures . Age exercised an exponential and negative effect on
Global and Positive Effect scores and a linear
negative effect on Service/Cost . It also had a
positive effect on these scores when combined
with invoice and perceived hearing difficulty .
Perceived hearing difficulty and its square
had a negative effect on Service/Cost and Personal Image scores . However, perceived hearing
difficulty in combination with other variables
(age, daily use, PTA, and style) exercised positive effects on all SADL measures except Negative Features .
Daily use was an important but complicated variable, which affected all SADL scores
except Service/Cost . Daily use x style had a
positive effect on Global scores . Daily use alone
had an exponential, negative effect on Negative Features but affected that subscale positively
in interactions with style or technology tier .
Daily use x total hearing aid experience and
x PTA increased Personal Image scores, but
daily use x PTA reduced Positive Effect scores .
Total hearing aid experience had an exponential and positive effect on Negative Features
but a linear and negative effect on Personal
Image. When combined with PTA or technology tiers, the effect on Negative Features was
23
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
Table 7
Results of Stepwise Backward Selective Procedure for Global and Subscale Scores,
Using Eight Variables,Their Squares and Interactions (44 Variables Total)
Variable
Parameter Estimate
Probability > F
Daily use x style
Perceived hearing difficulty x PTA
Age x invoice
PTA x invoice
-0 .00014
0 .04224
0 .00324
0 .000
-0 .000
0001
0018
0036
.0073
0221
Positive Effect (n = 243) r2 = .33
Age2
Daily use x PTA
Age x perceived hearing difficulty
Style x invoice
Perceived hearing difficulty'
PTA x invoice
-0 .00059
0 .00749
0 .01929
0 .00010
-0 .21586
0 .00010
0001
0001
0021
.0109
.0118
0109
-0 .06211
0 .01884
0019
0056
Global (n = 230) r2 = .117
Age 2
Service/Cost (n = 244) r2 = .17
Age
Age x perceived hearing difficulty
Perceived hearing difficulty'
-0 .20041
.0246
Negative Features (n = 240) r2 = .187
Daily use x style
Daily use'
Total hearing aid experience x PTA
Daily use x technology tiers
Total hearing aid experience'
Total hearing aid experience x technology tiers
0 .09503
-0 .13838
-0 .00631
0 .30100
0 .12598
-0 .28429
.0004
0009
0019
0057
.0144
0152
Personal image (n = 239) r2 = .140
Daily use x PTA
Total hearing aid experience x daily use
Perceived hearing difficulty x style
Perceived hearing difficulty x PTA
Total hearing aid experience
Age x invoice
Perceived hearing difficulty
-0 .01671
0 .24800
0 .07631
0 .01520
-0 .72887
-0 .72887
-0 .73282
0002
0005
0006
.0008
0025
.0025
.0040
negative . When combined with daily use, the
effect on Personal Image was positive .
PTA had mixed effects on all SADL scores
except Service/Cost in interactions with perceived hearing difficulty, invoice, daily use, or
total hearing aid experience .
Style interacted with daily use, invoice, and
perceived hearing difficulty to positively affect
all SADL scores except Service/Cost. Positive
effects indicated increased satisfaction with
smaller instruments .
Technology tiers affected only the Negative
Features subscale. In combination with daily use,
the effect was positive . In combination with
total hearing aid use, the effect was negative .
Invoice x age or x PTA had positive and
negative effects on Global, Positive Effect, and
Personal Image scores. Invoice x style increased
Positive Effect scores .
24
We further limited the scope of our inquiry
to the effect of intrinsic variable combinations
on Global scoring. Using the model
Global = age, PTA, perceived hearing difficulty, daily use, total hearing aid experience, and their square and interactions, age
was the only variable selected .
After age was entered, no further variable
in the model was significant (p < .05), but the
daily use x perceived hearing loss interaction
was borderline significant (p = .059). In realworld terms, even this very limited model
emphasized the small effect of age on overall satisfaction : for every decade of increased
age in our cohort, Global SADL scores
decreased by -0 .12 points on the 7-point
scale .
SADL Validation II/Hosford-Dunn and Halpern
DISCUSSION
Effects of Independent
Variables on SADL Scores
As conceptualized by Cox and Alexander
(1999), an important part of the clinical success
of the SADL lies in the correspondence of its subscales to domains that reflect wearer satisfaction, underscoring the multidimensional nature
of the satisfaction construct. When the SADL is
used clinically, the different patterns of scores
that emerge quickly make it apparent that several factors contribute to satisfactory hearing aid
fitting outcomes, with each factor affected by
variables that interact in complex and unique
patterns depending on the individual patient.
According to the theory of its construction,
the SADL score should meet or exceed norms
whenever a patient's needs are surmised correctly, appropriate counseling is provided (and
heeded by the patient), and a patient is fitted and
maintained with optimal instrumentation . In
such a theoretical world, score deviations below
SADL norms reflect a failure in the fitting
process due to one or more variables that were
improperly weighted during the counseling,
selection, fitting, or follow-up . Due to the clinical nature of our study, the audiologists strove
to ensure "best match" fittings by assessing and
treating patient- and hearing aid-related vari-
ables over 1-year intervals . To the extent that
they succeeded with each fitting, none of the
study variables should have reduced satisfaction
measures below norms . Likewise, if no study
variables are associated with inherently more
satisfaction, none should increase satisfaction
measures above norms . Therefore, the effects of
our study variables on SADL scores in the present data reflect variances that were unanticipated or poorly addressed in fitting and
treatment (e .g ., cosmetic concern) or are due to
inherent satisfaction differences associated with
the variables (e .g ., age, style) .
Overall, the results of this study reinforce the
observations that objective measures of hearing
disability (e .g., PTA) and hearing aid fitting (e .g .,
functional gain) and subjective measures of benefit (e .g ., Positive Effect) do not account for some
important aspects of wearer satisfaction with
hearing aids in daily life. Our attempts to identify other predictors of wearer satisfaction met
with only slight success. Although some variables are almost certainly related to long-term
satisfaction with hearing aid use, the data make
it clear that additional variables and interac-
tions must be analyzed before we can hope to predict satisfaction a priori . The results of this study
identify some areas of variance in SADL data as
a first step toward describing and codifying complex interactions of variables that affect long-term
hearing aid user satisfaction . At the same time,
we recognize that there are many complex variables such as personality, psychosocial adjustment to hearing loss, and overall health that
are weighed in clinical decision making that
were not included in our study.
Using univariate analyses, we dropped out
variables that did not affect satisfaction according to any SADL measures but kept other variables that appeared relevant to specific domains
of user satisfaction . In particular, five patientrelated variables correlated with benefit measures and three technology-related variables
affected cosmetic concerns . The only domain in
which both types of variables convened was
Negative Features, where style and patient variables made differences in satisfaction .
Although design differences prevent direct
comparisons, the same-direction correlation
trends reported in other research (cf., Schum,
1992 ; Dillon et al, 1999 ; Kochkin, 2000 ; Beamer
et al, 2000) support the validity of our observations (e .g ., negative correlation of PTA with Negative Features, positive correlation of daily use
with Positive Effect and with degree of hearing
loss). Thus, the first-order analyses encouraged
pursuit of a predictive model of satisfaction based
on the eight consequential variables shown in
Table 5. However, it became clear from higher-level
analyses that variables that affected satisfaction
domains did so in the presence of other influential variables and higher-order interactions, so
that satisfaction outcomes were affected in
extremely complex ways . Although we were able
to use the variables to improve outcome estimates
over those provided by SADL mean scores, the
improvements were not sufficient to develop a
model for clinical predictions .
The present data encourage further research
to complete a predictive model. They also allow
some speculation as to the paths that such
research might take . A few variables affected
SADL scores in patterns that suggest predictive
fitting profiles . Their recurrent significance in
relation to different SADL measures, as demonstrated by a variety of statistical analyses,
strongly suggested that the relationships are
important. The following discussion looks at the
effects of those variables and their interactions
on the SADL and the profiles that are suggested
by these effects .
25
Journal of the American Academy of Audiology/ Volume 12, Number 1, January 2001
Age
As patients aged, their hearing losses
increased but their impressions of hearing difficulty did not. They tended to use larger instruments, but their daily wearing times did not
increase . Their satisfaction with acoustic aspects
of hearing aid benefit decreased to the point
that Global satisfaction and Positive Benefit
scores declined as an exponential effect of age.
Indirectly, the inverse relation of age and acoustic
benefit is consistent with well-documented
reports in the literature of the deleterious effects
of age on speech understanding, due to cognitive
slowing or auditory processing deficits (Jerger
et al, 1990 ; Pichora-Fuller et al, 1995 ; GordonSalant and Fitzgibbons, 1999). More directly,
negative relationships have been reported for age
and hearing aid benefit (Beamer et al, in preparation) and for age and hearing aid user satisfaction (Smedley, 1990).
Age also reduced satisfaction with
Service/Cost . Interestingly, these negative effects
of age alone were offset somewhat by increases
in Global, Positive Effect, and/or Service Cost
scores for those older patients who perceived
more hearing difficulty and/or purchased higher
cost instruments. The latter almost certainly
reflects higher technology tier fittings, given
the high correlation of invoice cost with technology tiers. The former suggests an appreciation of services that may be tied to a heightened
awareness of disability in some older patients .
Clinically, one could speculate from these
results that hearing aid satisfaction in older
patients might be improved with a sequential,
twofold aural rehabilitation program emphasizing needs-matched technology applications
prior to purchase and systematically raising
expectations for hearing performance in postfitting aural rehabilitation sessions . This
approach is in line with data from Schum (1999)
indicating that prefitting expectations and postfitting perceived benefit operate as separate
domains.
Age-related dissatisfaction focused on the
four auditory benefit items on the SADL : speech
understanding (item 1), reduced repetitions
(item 5), naturalness of sound (item 10), and telephone help (item 11). However, satisfaction with
value aspects of hearing aid benefit did not
decline with age, so that older patients felt that
their instruments were "worth it" or "in their best
interest" even though they felt that the instruments were only partially successful in restoring auditory function. This may suggest that
26
older patients have lowered expectations for
amplification, perhaps because they have gradually grown used to their disability and perceive proportionally less handicap associated
with their hearing loss (cf., Gatehouse, 1999).
An example is illustrated in Figure 6. This
85-year-old man rated his hearing difficulty as
"moderate" in the presence of 60 dB PTA and
70 dB HL thresholds at 2000 Hz bilaterally. He
reported lower than average benefit on all auditory benefit items, including telephone use
(hence, the low Negative Features score) . Yet, he
expressed satisfaction with value, cost, benefit,
and appearance with his bilateral, full-shell,
multichannel, programmable instruments.
Daily Use
Patients who wore their instruments the
most every day had more hearing loss at all frequencies, had more years of hearing aid experience, reported more perceived hearing difficulty,
and wore larger instruments. In univariant
analyses, those patients' SADL scores reflected
more benefit in the Positive Benefit subscale
and more problems with feedback (item 7). Daily
use was not significantly related to Global satisfaction but had the highest correlation with Positive Effect (r = .33; see Table 5) . Other studies
using different outcome measures and analyses
also have found small but significant correlations between routine use and benefit (Schum,
1992 ; Dillon et al, 1997 ; Beamer et al, in preparation). Daily use did not influence other subscale
scores independently. Patients who did not use
their instruments (<1 hour/day reported on the
SADL) accounted for only 5.7 percent of the
Figure 6 SADL score sheet for elderly patient with low
auditory benefit and good psychological benefit, as
reported on Positive Effect subscale .
SADL Validation II/Hosford-Dunn and Halpern
group, notably less than the 16 percent reported
in MarkeTrak V (Kochkin, 1999).
Higher-order analyses allowing variable
interactions showed a negative, exponential
effect of daily use on Negative Features . Otherwise, daily use had complicated interactions
with other variables, which affected most SADL
domains. Patients who wore smaller hearing
aids or higher technology tier instruments a lot
each day reported higher Global and Negative
Features scores, perhaps due to fewer problems
with feedback . Patients with greater hearing
loss and/or longer histories of hearing aid use
who wore their instruments a lot each day were
less concerned with stigma and appearance, but
those with greater hearing loss also were less satisfied with Positive Effect . Clinical speculation
from these data suggests two profiles for predicting and manipulation satisfaction, based on
daily use:
0 Patients with significant and long-term hear-
ing loss who have grown accustomed to and
dependent on amplification over the years,
even in the presence of less than satisfactory
negative features . These patients run the risk
of settling into inertia by accepting instruments because they have worn them before
and because they have a great need for amplification . In order to improve satisfaction with
future fittings, counseling could be aimed at
educating these patients to ways in which
benefit, feedback, and background noise might
be improved by technological applications .
0 Patients with milder losses and smaller hear-
ing aids who encounter few or no wearing
obstacles (e.g ., feedback on the phone) . This is
an ideal profile . The daily use results serve to
remind clinicians of the importance of examining even "easy" fittings in terms of this
variable to undercover overt or covert obstacles that may discourage patients from incorporating their hearing instruments into
full-time daily use and therefore may reduce
wearer satisfaction .
In the literature, daily use is a frequently
reported measure of hearing aid fitting success
that is considered independent of but somehow
related to satisfaction (Dillon et al, 1997 ; Humes,
1999). Our findings concur with this assessment and suggest that daily use itself is a multidimensional construct, influenced by other
variables, that exercises a variety of effects on
satisfaction . Daily use is an important patient
outcome measure that affects satisfaction but
does not capture the full spectrum of satisfaction
dimensions .
The patient profiled in Figure 7 demonstrates an example of the limitations of daily use
as a satisfaction outcome variable . As Humes
(1999) points out, "It is not at all clear . . . that
higher hearing aid satisfaction will lead to higher
hearing aid use ." The patient in Figure 7 is very
satisfied but uses her instruments only a few
hours a day. She has a bilateral, mild highfrequency sensorineural hearing loss and purchased DSP instruments for one purpose : to
improve understanding of evening television .
This patient was made aware of infrared systems
for this purpose but wanted the communication
flexibility provided by amplification . One year
later, she reported daily use of 1 to 4 hours and
expressed higher than average satisfaction on
all SADL scales .
Total Hearing Aid Experience
Patients with longer histories of hearing
aid use had greater audiometric hearing losses,
had greater perceived hearing difficulty, and
wore their instruments more . Univariate analyses with SADL measures showed that patients
with longer total hearing aid experience reported
greater dependence on amplification (item 9) .
They also reported significantly more feedback
problems (item 7) and lower satisfaction in the
Negative Features subscale . Higher-order analyses confirmed that this variable alone had a
negative, exponential effect on Negative Features
scores . Patients with long histories of hearing aid
use who had greater hearing loss and/or wore
higher-technology instruments were more likely
to report Negative Features problems . Conversely, the long-time wearers who had greater
Figure 7 SADL score sheet for satisfied patient with
low daily use of amplification .
27
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
hearing loss and/or used smaller instruments
were less likely to voice Personal Image concerns .
Clinical speculation suggests a profile of
long-term hearing aid users who have few stigma
or cosmetic concerns but who are inherently
more difficult to satisfy because their degrees of
hearing loss or gain preferences push the technological envelope . Counseling for satisfaction
in this group needs to stress the importance of
technological applications to control negative
features but simultaneously stress the limitations of even the most sophisticated technologies
in high-gain fittings .
Figure 8 is an example of a long-term user
with bilateral severe hearing loss who is fitted
with multichannel, directional, dual-receiver
power aids with manual volume controls . He
has no measurable speech discrimination under
earphones and does not use the telephone .
According to the SADL, the fitting is successful
except for feedback problems that prevent the
patient from turning both aids to maximum volume. The patient and his family were made
aware of the fitting challenge and counseled
that periodic replacement of his ear molds would
help, but not alleviate, the problem.
Hearing Loss and Perceived
Hearing Difficulty
The relation between degree of hearing loss
and degree of perceived hearing difficulty manifest in our data is well documented in the literature (Cox and Alexander, 1999 ; Kochkin,
2000). Perceived hearing difficulty was positively correlated with audiometric hearing loss,
independent of frequency, and both variables
correlated with all intrinsic variables except
age.
First-order analyses showed that these two
variables affected SADL scores in different but
predictable ways : the greater the perceived hearing difficulty, the greater the satisfaction with
benefit, as reported in other studies (cf., Schum,
1999). The greater the hearing loss was, the
less satisfaction there was with Negative Features due to increased problems with feedback
and background noise.
Higher-order analyses showed a more complicated picture in which both variables interacted with a number of other variables and with
each other to influence satisfaction in most
SADL domains. Notably, patients with greater
hearing loss who reported greater perceived difficulty had higher Global SADL scores . Perceived hearing difficulty, alone or as an
interaction, affected all SADL domains except
Negative Features . PTA did not emerge as an
independent variable, but its interaction with
other variables affected all SADL domains except
Service/Cost .
To some extent, the complexity of PTA effects
is probably related to the hearing losses of the
sample population, as discussed in Dillon et al
(1999) . The lack of correlation in univariant
analyses between PTA and Positive Effect parallels their comment that "the use of average
hearing loss as a predictor of likely benefit
appears to be totally without support" (p . 40). The
interactive influences of PTA on satisfaction
domains further support their suggestion that
stronger correlations between hearing loss and
hearing benefit should emerge in sample groups
that include more patients with hearing in the
normal and profound regions.
Beyond these basic observations, it seems
as though PTA and perceived hearing difficulty are similar to daily use in that they function as multidimensional outcome measures
that overlap with satisfaction but fail to clarify its components .
Style
Figure S SADL score sheet for long-time hearing aid
user with severe hearing loss .
28
Smaller hearing aids were worn by younger
patients with less hearing loss, who reported less
perceived hearing difficulty, less daily use, and
shorter hearing aid histories. On the SADL,
smaller hearing aids were associated with
higher Global, Negative Features, and Personal
Image scores . Patients with small (ITC and
CIC) instruments reported more satisfaction
on these scales than did patients wearing larger
instruments (ITE and BTE) . In both subscales,
successively smaller instruments produced
SADL Validation II/Hosford-Dunn and Halpern
incremental increases in satisfaction, particularly for CIC wearers in terms of satisfaction
with telephone use (item 11) and appearance
(item 8) . Higher-order analyses confirmed that
style was important but emphasized the positive effects of style interactions with daily use
and perceived hearing difficulty to promote
higher Global, Negative Features, and Personal
Image scores . Style and technology tiers also
interacted at least weakly, with users of smaller
Tiers II and III instruments scoring higher on
the Personal Image subscale .
Clinical speculation for improving satisfaction based on this variable is clearly focused on
ways in which smaller, more sophisticated instruments can be fitted to more patients with greater
hearing loss without inducing negative features
(c£, Van Vliet, 2000) . The predictive profile is constantly changing as processing chips and ear
mold technology are developed to accommodate
ear canal geometries and enhance signal-to-noise
ratios . Counseling for long-term satisfaction
requires that the provider maintain in-depth
knowledge of emerging technologies and applications in order to optimize processor/packaging
recommendations . When fitting patients with
larger instruments, the predictive profile underscores the importance of spending extra time
with patients prior to fitting to conduct needsbased counseling and educate them on expectations . Postfitting training on telephone use and
assistive listening devices is also likely to improve
satisfaction scores for those patients using larger
instruments .
Technology Tier and Invoice Cost
Increasing circuit sophistication usually
raised the cost of instruments (based on manufacturers' invoicing) but did not correlate with
other independent variables or SADL measures .
More sophisticated circuitry interacted with
daily wearing time and years of hearing aid
use, respectively, to produce increases and
decreases in Negative Features satisfaction .
Personal Image scores were higher for wearers
of small programmable or DSP instruments.
Reviewing the data, it is clear that the independent contribution of technological sophistication to satisfaction was underestimated by
the statistical assumption of an ordered relationship between technology tiers and SADL
scores . In our method, instruments were rank
ordered according to their programmability and
processing circuitry. By that categorization, DSP
instruments (Tier IIl) had the highest satisfac-
tion ratings on 60 percent of SADL items (including cost) and on all SADL scores except Positive
Effect . However, Tier I instruments scored
slightly higher than Tier 11 instruments on one
subscale (Negative Features) and on several
SADL items, effectively reducing correlations in
the statistics . The trends were not great because
t-tests for differences in means did not reach significance (p < .05) in any comparison . Nevertheless, it seems likely that some other means
of describing instrumentation might reveal more
about its importance to satisfaction domains .
Exactly how this might be done is not clear
because our attempts to categorize the data in
different ways (e .g., number of processing chan-
nels, compression vs noncompression) did not
result in higher correlations with SADL measures . One possibility is to compare DSP versus
analog instruments or programmable versus
linear instruments . For instance, Kochkin (1996)
reported 13 percent higher overall satisfaction
ratings for programmable analog instruments
compared with "typical product[s] in the marketplace ."
Clinical speculation, supported less by statistical significance than by trends, suggests
that DSP instrumentation with automatic, intelligent management of feedback and environmental nonspeech noise will reduce obstacles to
full-time use and contribute to other satisfaction
domains as well .
Comparison of Private Practice
Scores to SADL Norms
The fact that the SADL isolates and quantifies satisfaction and dissatisfaction associated
with hearing aid style in the larger context of
overall satisfaction is an important clinical
result . In Part I, we found that Negative Features scores in the PP-SADL group were significantly higher than the conservative interim
norms reported by Cox and Alexander (1999) . In
Part II, we found that subscale patterns for
PP-SADL patients differed according to hearing
aid style so that Negative Features and Personal Image scores were higher for smaller aids,
especially CICs . Comparisons of subscale norms
with subscale mean scores for CIC wearers in the
PP-SADL group (Table 8) show that CIC wearers in the present study were less dissatisfied
with Negative Features and more satisfied with
Personal Image than subjects in the normative
group. The differences shown in Table 8 are
based on a small sample and need replication
before it is determined that SADL norms require
29
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
Table 8 Comparison of Interim SADL
Norms to SADL Statistics for CIC Wearers
Low Positive Benefit Subscale
The SADL is a clinically useful outcome
measure because its four subscales analytically
and visually profile factors that reflect hearing
aid fitting successes and failures, as shown in the
following examples . Failure in a specific subscale shows up as a readily identified scoring pattern that can be used to document problems in
a fitting, plan efficient intervention strategies,
and quickly and flexibly redirect intervention .
The patterns facilitate counseling and recommendations by including the patients in the
process. Patients can see how they compare to
other hearing aid wearers, and clinicians can
use the patterns to isolate and translate dissatisfactions into workable goals for improvement
that are understandable to the patients . Examples of subscale-specific profiles and their interpretations are discussed in the following sections .
The patient in Figure 9A is a 90-year-old
man with moderate hearing loss (PTAs of 40
and 45 dB HL) who is a first-time hearing aid
user. He is fitted with binaural canal instruments of Tier I technology. His SADL pattern
shows very low Positive Effect but satisfaction
on other subscores. This pattern signifies low
benefit, especially because telephone use is
scored very low in the otherwise satisfactory
Negative Features subscale . The pattern was
rarely encountered in our data . Only three
patients (all Tier I instruments ; two ITC, one
CIC) scored below the 20th percentile on Positive Effect but at or above norms on the remaining subscales. One reason for this is the high
correlation of Service/Cost and Positive Effect
subscales so that these subscores usually moved
in the same direction (as in Fig. 9B). The patient
in Figure 9A was an exception because he
remained pleased with the service and cost,
even though he felt that the aids provided little,
if any, acoustic or psychological benefit. Despite
low benefit, Global satisfaction remains in the
acceptable range, in large part because of the
absence of technical problems with the fitting.
One of the important values of the SADL is that
satisfaction does not rely solely on a single index
of satisfaction but reveals covert areas of dissatisfaction that may be less obvious than overt
problems such as feedback. The recurring pattern shown in Figure 9A and elsewhere of an
acceptable Global score in the presence of subscore dissatisfaction underscores the comment
that "absence of problems with a hearing aid fitting should thus not be regarded as an indica-
A
B
Negative Features
Personal Image
Norms CIC Scores Norms CIC Scores
(n = 256) (n = 45) (n = 103) (n = 45)
Mean
SD
20th percentile
80th percentile
3 .6
1 .4
2.3
5 .0
4 .6
1 .2
3 .6
5 .7
5 .6
1 .1
5 .0
6 .7
6 .3
0 .7
6 .0
7 .0
adjustments for hearing aid type or for practice
setting.
Clinical Applications of
the SADL by Profiling
~
Nvm
Figure 9 SADL score sheets for two patients with low Positive Effect profiles . Service/Cost
scores are high for the
first patient (A) and low for the second patient (B).
30
SADL Validation II/Hosford-Dunn and Halpern
tion of a successful hearing aid fitting" (Dillon
et al, 1997).
Inadequate gain or instrument malfunctions are the first things to rule out when a low
Positive Effect pattern appears. Assuming adequate gain and function, the low Positive Effect
pattern in Figure 9A is consistent with three scenarios : (1) the patient's expectations for the
selected amplification are too high, (2) the patient
is not using the instruments regularly, and/or
(3) the patient is not using the instruments efficiently (e .g ., setting volume control very low or
failing to adjust for different environments) .
Each scenario points to patient-related variables that deserve counseling efforts. Also, all scenarios open the possibility of recommending
higher-technology instruments either to better
match the patient's expectations or facilitate
audibility and adjustment via automatic volume control.
The pattern in Figure 9A is especially conducive to effective counseling that is likely to
improve satisfaction because the patient remains
pleased with Service/Cost issues . The pattern in
Figure 9B was more common in the data set .
Assuming adequate gain, the scenarios suggested by this pattern are the same as Figure 9A,
involving mainly intrinsic variables, and the
approach revolves around counseling issues .
However, the likelihood of improving patient
satisfaction is lower because the credibility of the
provider and/or instruments is doubtful in these
patients' minds .
Low ServicelCost Subscale
Correlation and factor analyses showed
that the Service/Cost subscale was closely
related to satisfaction with benefit. Service/Cost
also correlated significantly with other subscales . It seems likely that almost all patients
rate this subscale by calibrating it to their satisfaction with benefit and, to a lesser extent,
with negative features and cosmetic/stigma
concerns . There were no subjects who scored
below the 20th percentile for Service/Cost if
their satisfaction in the other subscales was at
or above subscale means. The SADL in Figure
10 almost met criteria, but Positive Effect was
just below the mean . This atypical pattern
gives explicit information that the hearing aid's
poor repair record reflects poorly on the clinician . Unless decisive steps are taken to fix the
repair problem, it seems likely that the patient
will pursue new amplification with a different
vendor.
Figure 10 SADL score sheet for patient with low
Service/Cost profile .
Low Negative Features Subscale
Figure 11 illustrates the SADL results from
an 87-year-old man with bilaterally symmetric
hearing loss and an average PTA of 61 dB HL .
His SADL scores meet or exceed norms except
for Negative Features, which shows almost complete dissatisfaction with all items in the subscale . His comments are informative : "I have
eliminated going to any meeting where there
may be several people talking at once, or where
there may be background noise ." The low Negative Features pattern signals a patienttechnology mismatch, which may be addressed
to the extent that technological applications
exist to correct the problems (e .g ., stronger t-coil,
directional microphones, tighter earmold, handheld microphone, or personal fm system) .
Low Negative Features was the most common of the single low subscale score patterns and
suggested some interesting findings related to
Figure 11 SADL score sheet for patient with low Negative Features profile .
31
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
hearing aid variables. Of the 10 outliers who
comprised this group, 8 wore BTE instruments,
9 were fitted with Tier II technology instruments, and none wore CIC aids . Dissatisfaction
with telephone use in this group is likely to be
due to microphone location in the case of BTEs
and degree of hearing loss for BTE, ITE, and ITC
fittings . Dissatisfaction with feedback and background noise is likely to be higher for Tier II
instruments because those instruments in our
study characteristically had wider dynamic
ranges and frequency responses than most Tier
I instruments, but less active control of feedback
and gain than Tier III instruments (Appendix B) .
Thus, when patients were given choices in several tiers during hearing aid selection, those
who opted for higher technology at lesser cost by
choosing programmable rather than DSP instruments may have obtained the benefit they
wanted but at the price of more negative features. This "good news/bad news" aspect of Tier
II selections may have confounded any direct
relations between technology tiers and SADL satisfactions .
Clinically, the approach to the Figure 11
pattern centers on technology recommendations
that are appropriate for the individual patient's
hearing loss, hearing aid experience, and daily
use . The SADL can be used to improve the
patient's fitting, first by using the Positive Benefit score to reassure the patient that it is consistent with other wearers'reports and then by
using the low Negative Features subscale to let
the patient know that this much dissatisfaction
is atypical . Discussion of Negative Features
components allows education and recommendations on ways to reduce the negative aspects
of the fitting by switching to a different hearing
aid style or to higher-technology instruments for
better control of feedback and reduced loudness
intolerance of environmental noise. In the case
of the patient in Figure 11, he had upgraded from
programmable to DSP instruments and felt that
the benefit was sufficiently improved that a further upgrade to directional DSP instruments
was not warranted. Instead, he was fitted with
new ear molds to reduce feedback, was given a
handheld extension microphone for use in noise,
and was instructed on the use of a speakerphone. He declined other assistive listening
device options.
problems . Nine patients scored Personal Image
below the 20th percentile in the presence of
scores in the other subscales that were at or
above norms. None of these patients wore CIC
instruments; otherwise, no particular pattern of
technology and/or style was noted.
The patient in Figure 12 was a 66-year-old
man with an asymmetric hearing loss with right
and left PTAs of 61 and 35 dB HL, respectively.
He was a first-time user. The fitting was monaural and right, with a multichannel BTE with
automatic volume control. In contrast with
patients who rated Personal Image low based on
cosmetics (cf., Figs . 6C and 6D in Hosford-Dunn
and Halpern, 2000), the patient in Figure 12
expressed strong dissatisfaction with the two
stigma items in the subscale but was not
unhappy with appearance .
The low Personal Image pattern telegraphs
a need for counseling of patients regarding their
expectations for those who want smaller instruments and information about hearing aid use in
general for those who perceive stigma . The fact
that these patients report good benefit is an
important starting point. Appearance issues
may be reduced if patients learn from the SADL
that they meet or exceed the satisfaction of other
hearing aid wearers who use the same styles of
instruments. This pattern can also signal a need
to re-evaluate the patient's fitting, in light of new
technological advances that may allow more
acceptable packaging of the processing features
required for good benefit .
Low Personal Image Subscale
Reduced Scores on Multiple Subscales
Figure 12 shows a low Personal Image pattern with good benefit and no negative feature
As the single-subscale dissatisfaction patterns demonstrate, satisfaction in most areas is
32
Figure 12 SADL score sheet for patient with low Personal Image profile .
SADL Validation II/Hosford-Dunn and Halpern
sustainable only so long as instrumentation is
appropriate to the patient's condition . Lower
than average scores on Positive Effect and Negative Features subscales should trigger discussion with the patient about expectations,
listening situations, assistive devices, and appropriateness of current amplification for the degree
and type of hearing loss . For example, fitting a
CIC instrument to satisfy a cosmetic concern in
the presence of severe to profound hearing loss
runs the risk of provoking dissatisfaction in the
benefit domain. Lower than expected Positive
Effect scores with a cosmetically concerned
patient may serve as a red flag to revisit the
attributes of the fitting and counsel the patient
accordingly. At the same time, the Negative Features subscale score may serve as the red flag
by signalling problems with feedback and telephone use.
Figure 13 is a case in point. This patient was
an 81-year-old man with a history of full-time
ITC hearing aid use with recurrent feedback
problems for many years. He had a stable, bilateral hearing loss with left and right PTAs of 50
and 60 dB HL . Over the years, he repeatedly
resisted recommendations for a BTE or ITE fitting but was anxious to purchase DSP instruments in a canal style, despite obvious feedback
and our negative recommendation . One-year
postfitting, his SADL results are as shown in
Figure 13 . He reported lower than normal satisfaction in all SADL categories . Low benefit
was the direct result of gain compromises dictated by persistent feedback in the fitting.
Service/Cost followed Positive Effect, as was
typical in our data . Negative Features reflected
feedback problems in general and with the telephone. Personal Image showed the stigma
expressed over the years by this patient, even
MX a aap
7aJX Z(M lo !M PercmYe
Figure 13 SADL score sheet for patient with dissatisfaction expressed on all scales .
though appearance of the instrument was
satisfactory.
In this case, the SADL results dictated a
counseling approach that focused on technology and expectations because the patient's history made it likely that stigma would persist
regardless of the fitting. The SADL results were
presented to the patient, with emphasis on the
low Positive Effect and Negative Features subscores, as support for our prior recommendation
for appropriate BTE technology. The graphic
representation of dissatisfaction in many categories was a strong and convincing argument for
this gentleman, who finally agreed to wear programmable BTE instruments that have proven
satisfactory in terms of reducing feedback and
providing adequate gain .
CONCLUSIONS
P
arts I and II of this study underscore
Abrams and Hnath-Chisolm's (2000) observation that satisfaction is a multidimensional
construct requiring independent assessment .
The SADL probes domains of importance and
satisfaction and scores those domains in a
psychometrically sound manner (Cox and
Alexander, 1999). In the absence of other multidimensional outcome measures of satisfaction,
and given the good construct and psychometric
properties of the SADL, a case can be made for
using the SADL as a gold standard for establishing satisfaction outcomes . In that case, SADL
scores enable a search for independent variables that predict wearer satisfaction a priori .
In this study, as in others, satisfaction was
not well predicted by specific hearing aid technologies, measures of hearing loss or hearing
difficulty, hearing aid fitting, or benefit. However,
the significance of five intrinsic and three extrinsic variables in predicting SADL scores was
strong evidence that these variables are related
to SADL scores and that it is worth studying further the complex relationships of these variables
on satisfaction . Patient-related variables that
affected SADL scores were patient age, years of
hearing aid experience, hours of use per day,
perceived hearing difficulty, and combined-ear
PTA (four frequency, 500 to 3000 Hz) . Hearing
aid-related variables that affected SADL scores
were hearing aid style, performance characteristics (as defined by three levels of circuit
sophistication), and manufacturer's invoice
cost . The patient- and hearing aid-related variables, their squares and interactions, improved
SADL subscore predictions by as much as 33
33
Journal of the American Academy of Audiology/Volume 12, Number 1, January 2001
percent over that provided by SADL norms
alone, based on the r2 values in Table 7 . However, Global and other subscore predictions
were improved by as little as 12 percent compared with norms alone, leading us to conclude
that our endeavor encourages more research
with additional variables before a clinically
useful model can be developed .
Some clinically useful information and some
potentially useful trends emerged from the
study:
" For a privat e practice cohort , sati s faction is
greater with smaller instruments, especially
in the Negative Features and Personal Image
subscales and especially for telephone use in
CIC wearers . It is fair to say that overall satisfaction with hearing aids would have been
improved significantly if telephone use were
satisfactorily addressed for non-CIC wearers
in this cohort. If replicated, the differences
may be sufficient to warrant separate norms
for Negative Features and Personal Image
subscales for CIC fittings .
0 Technological
sophistication of instruments
did not affect satisfaction in a direct manner, at least for the three categories defined
in our study. Due to the nature of our study,
any interpretation from this finding that hightechnology instruments are no more satisfactory than low-technology instruments is
superficial and incorrect . Patients were fitted
according to their needs and expressed wants,
with an end goal of tailoring recommendations
to optimize individual satisfaction . To the
extent that this goal was achieved, differenttiered instruments fitted by need and want
should yield equivalent satisfaction . Nevertheless, higher-tiered technologies in smaller
instruments were associated with greater
satisfaction on the Personal Image subscale .
DSP (Tier III) technology ranked highest in
satisfaction on a majority of items in this and
other studies (Sandridge and Newman, 1998).
These findings suggest that Tier III-type
instruments are likely to prove more satisfactory in daily life, at least for private practice patients who are fitted according to
needs-based counseling.
" Satisfaction was lower for o ld er pat i ents,
mainly due to reduced auditory benefit in
general and with more difficulty using the telephone.
0 Hours of daily use,
degree of hearing loss,
and perceived hearing difficulty were measures that related to several SADL scores in
34
0
complex, overlapping ways, suggesting that
these measures themselves are multidimensional outcome measures that are essential to
satisfaction but do not capture its full essence.
SADL scores yield subscale-specific patterns
of satisfaction and dissatisfaction that are
useful graphic "snapshots" for the clinician
and patient. In particular, the areas of dissatisfaction are helpful in planning intervention . As Resnick (1998) comments,
"identification of the variables contributing to
an unsuccessful fitting [may be] as useful as
identification of those giving rise to a successful one" (p . 133) . Different patterns of
dissatisfaction suggest the efficacy of emphasizing a counseling approach to change patient
attitudes and behaviors or of pursuing a more
action-oriented approach to change the hearing instruments, their components, or accessories, depending on the pattern. Patients
easily understand the SADL patterns, which
lend structure and credibility to the provider's
counseling and recommendations .
Acknowledgments. Mary Gansheimer, MS, Nermana
Hrustic, Judy Huch, MS, Alicia Hutzel, MS, Michael Irby,
MS, Julie Leonard, MS, and Sherri MacMillan, MS, participated in data collection . Robyn Cox, PhD, Director of
the Hearing Aid Research Laboratory, Memphis State
University, provided SADL forms and the SADL scoring
program. Drs. Harvey Abram, Robyn Cox, and Donald
Schum reviewed the manuscript and offered helpful
criticism.
REFERENCES
Abrams H, Hnath-Chisolm T. (2000) . Outcomes . In :
Hosford-Dunn HL, Roeser R, Valente M, eds. Audiology
Practice Management. New York : Thieme, 69-95 .
Beamer SB, Grant KW, Walden BE . (2000) . Hearing aid
benefit in patients with high-frequency hearing loss . J
Amer Acad Audiol 11 :429-437 .
Cox RM, Alexander GC . (1999) . Measuring satisfaction
with amplification in daily life: the SADL scale . Ear Hear
20 :306-319 .
Dillon H, James A, Ginis J. (1997). Client Oriented Scale
of Improvement (COSI) and its relationship to several
other measures of benefit and satisfaction provided by
hearing aids . JAm Acad Audiol 8:27-43 .
Dillon H, Birtles G, Lovegrove R. (1999) . Measuring the
outcomes of a national rehabilitation program: normative data for the Client Oriented Scale of Improvement
(COSI) and the HearingAid Users Questionnaire (HAUQ) .
JAm Acad Audiol 10 :67-79 .
Gatehouse S. (1999). Glasgow HearingAid Benefit Profile :
derivation and validation of a client-centered outcome
measure for hearing aid services . J Am Acad Audiol
10 :80-103.
SADL Validation II/Hosford-Dunn and Halpern
Gordon-Salant S, Fitzgibbons PJ . (1999) . Profile of auditory temporal processing in older listeners . JSpeech Lang
Hear Res 42 :300-311 .
Hosford-Dunn HL, Baxter JH . (1985) . Prediction and validation of hearing aid wearer benefit: preliminary findings .
Hear Instr 36 :34-41 .
Hosford-Dunn H, Halpern J. (2000) . Clinical application
of the Satisfaction with Amplification in Daily Life Scale
in private practice I: statistical, content, and factorial
validity. JAm Acad Audiol 11 :523-539 .
Huch JL, Hosford-Dunn H. (2000). Inventory of self-report
outcome measures of hearing aid use. In : Sandlin R,
McCandless G, eds. Hearing and Amplification . 2nd Ed .
San Diego: Singular, 489-555.
Humes LE . (1999) . Dimensions of hearing aid outcomes .
J Am Acad Audiol 10 :26-39 .
Jerger J, Mahurin R, Pirozzolo F. (1990) . The reparability of central auditory and cognitive deficits : implications
for the elderly. J Am Acad Audiol 1:116-119 .
Kochkin S. (1996) . Customer satisfaction and subjective
benefit with high-performance hearing instruments. Hear
Rev 3(12):16-26 .
Kochkin S. (1999) . MarkeTrak V: "Baby Boomers" spur
growth in potential market, but penetration rate declines .
Hear J 52(1):33-48 .
Kochkin S. (2000) . Quantifying the obvious: the impact
of hearing instruments on quality of life . Hear Rev
7(1) :6-34.
Pichora-Fuller KM, Schneider BA, Daneman M. (1995) .
How young and old adults listen and remember speech
in noise. JAcoust Soc Am 97 :593-608 .
Resnick S. (1998) . Breakdown in the fitting process. In :
Tobin H, ed . Practical Hearing Aid Selection and Fitting.
Washington, DC : Department of Veterans Affairs.
Sandridge SA, Newman CW . (1998) . Subjective
Satisfaction Ratings for Digital Signal Processing Hearing
Aids . Paper presented at the American Speech-LanguageHearing Association Annual Convention . San Antonio,
TX.
Schum D. (1992) . Responses of elderly hearing aid users
on the Hearing Aid Performance Inventory. J Am Acad
Audiol 3 :308-314 .
Schum D. (1999) . Perceived hearing aid benefit in relation to perceived need . J Am Acad Audiol 10 :40-45 .
Smedley TC . (1990) . Self-assessed satisfaction levels in
elderly hearing aid, eyeglass, and denture wearers. Ear
Hear 11(Suppl 5) :41S-47S .
Sweetow RW. (1999) . Counseling : it's the key to successful hearing aid fittings . Hear J 52(3) :10-17 .
Van Vliet D . (2000) . Hide and seek. Hear J 53(1) :88 .
APPENDIX A
Typical Tier Sheet* Used in 1996/1997 for Hearing Aid Recommendations
that greatly improve
The hearing aid industry has introduced many new and innovative products
when
choosing new hearhearing aid performance . These new products add to the options you have
their features :
summarized
ing aids . To help you decide which product best suits your needs, we have
Tier 1. Conventional Instruments
Designed for:
environments
" Those whose main concern is hearing spouse and/or television in quiet
" Those who want to have a manual volume control
instrument
" Previous hearing aid wearers interested in upgrading a standard
Tier II. High Performance and Programmable Analog Instruments
multimemory, remote conFeaturing one or more of the following: computer based, multichannel,
Designed for :
trolled, fully automatic, or manual volume control, multimicrophone.
Active individuals
People with poor speech intelligibility
Previous hearing aid users looking to improve hearing in noise
Individuals who do not want to adjust their hearing aids at ear level
Those who are interested in advanced technologies
Tier III. DSP Instruments
Designed for:
" Enhanced speech and reduced noise
" High fidelity and natural sound reproduction
" Feedback cancellation
" Quiet, distortion-free sound
updated frequently to reflect changes
-Typical technology tier chart used during the period of this study. Charts were
July
1997)
.
in available instruments (i .e ., DSP CICs were not available until
35
Journal of the American Academy of Audiology/Volume 12, Number 1, January
2001
APPENDIX B
List of Manufacturers' Instruments Used in This Study,
Categorized by Tier
Manufacturer
Tier I
Phonak
Rexton
Starkey
UHS
Widex
Tier II
Danavox
Oticon
Phonak
Resound
Siemens
Starkey
Widex
Tier III
Oticon
Widex
Model (Descriptors)
Inca CIC (Class A or D ; linear or Super Compression)
Miniprimo BTE
CE 8 and 9 ITE (Class A or D)
Intra IV and V ITC (Class A or D)
Secret Ear ITC (Class A or D)
Sequel CIC (Class D, single channel, nonprogrammable)
SM-AGC BTE
Ultima III ITC and ITE
ES2-T BTE
Aura BTE
Multifocus ITE and BTE
Microfocus ITC
Primofocus ITC and ITE
Piconet AZ BTE
Sonoforte AZ BTE
BT2-E BTE
ED3 and ED3-E ITE and BTE
IE4 ITC (with and without remote)
Intelivenience CIC
Music CIC
Interra BTE
Sequel CIC
Sequel 675 AV or AGC BTE
L12T BTE
Digifocus ITE and BTE
Senso CIC, ITC, ITE, and BTE