J Am Acad Audiol 25:154-163 (2014)
Prediction of IOI-HA Scores Using Speech Reception Thresholds and Speech Discrimination Scores in Quiet DOI: 10.3766/jaaa.25.2.4 K. Jonas Brännström*§ Johannes Lantzt§ Lars Holme Nielsen^ Steen 0stergaard
Abstract Background: Outcome measures can be used to improve the quality of tlie rehabiiitation by identifying and understanding which variables influence the outcome. This information can be used to improve outcomes for clients. In clinical practice, pure-tone audiometry, speech reception thresholds (SRTs), and speech discrimination scores (SDSs) in quiet or in noise are common assessments made prior to hearing aid (HA) fittings. It is not known whether SRT and SDS in quiet relate to HA outcome measured with the International Outcome Inventory for Hearing Aids (IOI-HA). Purpose: The aim of the present study was to investigate the relationship between pure-tone average (PTA), SRT, and SDS in quiet and IOI-HA in both first-time and experienced HA users. Research Design: SRT and SDS were measured in a sample of HA users who also responded to the IOI-HA. Study Sample: Fifty-eight Danish-speaking adult HA users. Data Collection and Analysis: The psychometric properties were evaluated and compared to previous studies using the IOI-HA. The associations and differences between the outcome scores and a number of descriptive variables (age, gender, fitted monaurally/binaurally with HA, first-time/experienced HA users, years of HA use, time since last HA fitting, best ear PTA, best ear SRT, or best ear SDS) were examined. A multiple forward stepwise regression analysis was conducted using scores on the separate IOI-HA items, the global score, and scores on the introspection and interaction subscales as dependent variables to examine whether the descriptive variables could predict these outcome measures. Results: Scores on single IOI-HA items, the global score, and scores on the introspection (items 1,2,4, and 7) and interaction (items 3, 5, and 6) subscales closely resemble those previously reported. Multiple regression analysis showed that the best ear SDS predicts about 18-19% of the outcome on items 3 and 5 separately, and about 16% on the interaction subscale (sum of items 3, 5, and 6) Conclusions: The best ears SDS explains some of the variance displayed in the IOI-HA global score and the interaction subscale. The relation between SDS and IOI-HA suggests that a poor unaided SDS might in itself be a limiting factor for the HA rehabilitation efficacy and hence the IOI-HA outcome. The clinician could use this information to align the user's HA expectations to what is within possible reach. Key Words: Hearing aids, outcome, satisfaction Abbreviations: HA = hearing aid; IOI-HA = International Outcome Inventory for Hearing Aids; PCA = principal components analysis; PTA = pure-tone average; SDS = speech discrimination score; SRT = speech reception threshold
'Department of Logopedics, Phoniatrics and Audiology, Clinical Science in Lund, Lund University, Lund, Sweden; tGI\l Otometrics, Taastrup, Denmark; ^Research Laboratory, Department of Otorhinolaryngology, Head and Neck Surgery, University Hospital, Rigshospitalet, Copenhagen, Denmark; §HEARsound Laboratories, Network for Joint Hearing Research in the Oresund Region K. Jonas Brännström, Clinical Science in Lund, Department of Logopedics, Phoniatrics and Audiology, SE-221 00 Lund, Sweden; Phone; +46(0)46-177103; Fax; +46(0)46-171732; E-mail; [email protected]
154
Prediction of IOI-HA/Brännström et al
T
he International Outcome Inventory for Hearing Aids (IOI-HA) is a hearing-specific questionnaire developed by Cox et al (2000) with the purpose of assessing hearing aid (HA) rehabilitation efficacy. As an outcome measure, it can be used by the clinician to assess how well the rehabilitation has worked from the client's perspective (Uriarte et al, 2005). In a similar way, it can be used to improve the quality of the rehabilitation (Beck, 2000). One way to achieve the latter is by identifying and understanding which variables infiuence the outcome and use this knowledge to improve outcomes for clients (Hickson et al, 2010). In clinical practice, prior to HA fittings, pure-tone audiometry, speech reception thresholds (SRTs), and speech discrimination scores (SDSs) in quiet or in noise are often determined. It is not known whether SRT and SDS in quiet relate to and can be used to predict the HA outcome measured with the IOI-HA. If, for example, poor scores on these measures are related to poor scores on the IOI-HA, the clinician can use this information to improve interventions that also improve the clients' outcome. The psychometric properties of the IOI-HA have been examined in different languages (Cox and Alexander, 2002; Kramer et al, 2002, Stephens, 2002; Heuermann et al, 2005, Jespersen et al, 2005; Vestergaard, 2006; Öberg et al, 2007; Smith et al, 2009; Brännström and Wennerström, 2010). These studies clearly demonstrate that the seven items of the questionnaire can be divided into two subscales: introspection ("me and my hearing aid"; items 1, 2, 4, and 7) and interaction ("me and the rest of the world"; items 3, 5, and 6). However, as Jespersen et al (2005) and Vestergaard (2006) showed, the official Danish version presents a negative item-total correlation for one of the items (item 5, residual participation restriction). A negative correlation should not be present due to the direction of the response values (a higher score always represents a more positive response). For this particular item this indicates that subjects may have problems understanding the response values and hence reporting residual disability. This negative correlation has also been reported by Cox and Alexander (2002). Addressing this particular issue, Jespersen et al (2014) rephrased item 5.
The relationship between the IOI-HA and several demographic variables has been examined. Although some previous studies indicate that older subjects use their HAs less and report less benefit than younger subjects (Stephens, 2002; Öberg et al, 2007), others did not (Kramer et al, 2002; Heuermann et al, 2005; Cook and Hawkins, 2007; Brännström and Wennerström, 2010; Hickson et al, 2010). Increasing levels of hearing loss lead to increased HA usage, but the association between the IOI-HA and pure-tone hearing thresholds seems unclear
(Humes et al, 2002; Kramer et al, 2002; Larson et al, 2002; Stephens, 2002; Stark and Hickson, 2004; Heuermann et al, 2005; Meister et al, 2005; Metselaar et al, 2009; Smith et al, 2009; Brännström and Wennerström, 2010; Hickson et al, 2010). Also, the type of hearing loss explains a minor part of the variance (less than 5%) in scores on the IOI-HA introspection subscale (items 1, 2, 4, and 7); subjects with sensorineural hearing loss reported poorer scores than subjects with conductive or mixed hearing loss (Brännström and Wennerström, 2010). However, monaural or binaural HA fitting does not inñuence the IOI-HA global outcome (Heuermann et al, 2005; Cook and Hawkins, 2007; Brännström and Wennerström, 2010; Hickson et al, 2010), but binaurally fitted subjects seem to report higher scores on single items (residual activity limitation, item 3) (Liu et al, 2011), and they report longer HA use durations (HA use, item 1) (Stephens, 2002). Also, binaurally fitted subjects who have had longer hearing loss duration are more likely to demonstrate higher scores on the introspection subscale (Kramer et al, 2002; Stephens, 2002). It is unclear whether there is a gender effect on the IOI-HA (Cook and Hawkins, 2007; Liu et al, 2011) or not (Kramer et al, 2002; Heuermann et al, 2005; Williams et al, 2009; Brännström and Wennerström, 2010; Hickson et al, 2010). The IOI-HA global score increases slightly but significantly with increasing price of the HA, when subjects themselves finance their devices (Liu et al, 2011). Though it is a common clinical experience that HAs cannot fully compensate for hearing impairment, it is rarely known which factors, specifically, limit the HA outcome. It would therefore be relevant to be able to identify factors that infiuence the HA rehabilitation efficacy, for example, as measured with the IOI-HA. Taylor (2008) demonstrated an association between IOI-HA and the acceptable noise level (ANL) test (Nabelek et al, 1991; Nabelek, 2006), afindingthat Olsen and colleagues (2012) could not replicate in a larger sample. If the ANL shows that a subject only accepts little noise in the presence of speech, and this in turn affects the IOIHA outcome, then we might be able to increase the HA benefit for this particular subject by emplojdng noise reducing algorithms. Hickson and colleagues (2010) explained about 57% of the variance in the overall IOI-HA based on findings that higher average IOIHA scores are associated with (1) greater satisfaction when having a conversation with a single person, in small groups, in large groups, and outdoors and (2) greater satisfaction with HA attributes such as comfort with loud sounds, comfort of the physical fit, and clarity of tone and sound. Hence, the speech perception ability in quiet (as measured with the SRT and SDS) might affect the IOI-HA. If a relation between SRT and/or SDS and IOI-HA can be established, then we will know that a poor SRT and/or SDS might be a limiting factor
155
Journal of the American Academy of AudiologyA^olume 25, Number 2, 2014
for the HA rehabilitation efficacy and hence the IOI-HA outcome. With this information in hand we would better be able to explain why an HA in such a case does not compensate for the hearing loss to the same extent as when users have a good SRT/SDS. The aim of the present study was therefore (1) to provide preliminary psychometric properties of the revised Danish IOI-HA and (2) to investigate if the outcome of the SRT and SDS in quiet affects IOI-HA in first-time and experienced HA users.
Table 1. Demographic Data for All Included Users (n = 58) Variable Age (years)
Gender PTA (dB HL)
Right ear
Left ear
MATERIAL AND METHODS Best ear or fitted ear
Subjects Potential subjects were identified through our patient files at Rigshospitalet in Copenhagen. Initially, 85 HA users were recruited to the study. AU subjects were native Danish speakers. Inclusion criteria were sensorineural hearing loss, normal tympanograms (within ± 150 daPa) according to Margolis and Goycoolea (1993), and recordable contralateral acoustic refiexes at 1 kHz. Also, the subjects should have been fitted with HAs, and they were required to actually use them. No other inclusion criteria were used. Fifty-eight subjects met all these inclusion criteria. Of the 27 excluded subjects, 20 subjects did not meet all inclusion criteria, three subjects had too poor hearing to be able to adjust the speech signals to the most comfortable level in one part of the test, three additional subjects had too poor SRTs to be able to present SDSs at the maximum output level of the audiometer (i.e., 90 dB SPL), and one subject was not able to perform the measurement at all. The IOI-HA was administered at Rigshospitalet in Copenhagen in conjunction with the earlier mentioned project (Olsen et al, 2012). At the time that the IOI-HA was administered, fifty-one participants (88%) were binaurally fitted with HAs, and seven participants (12%) were monaurally fitted with HAs. Thirteen (22%) of the participants were fitted with HAs for the first time, while forty-five (78%) of the subjects had renewed their initial devices at least once over the period. The participants had HAs for a mean period of 10.7 yr (SD = 7.4, Range = 0.2-26), and the mean period of time since the last HA aid fitting was 2.4 yr (SD = 2.5; Range = 0.2-16). Demographic data and information on hearing thresholds (pure-tone average [PTA]), SRT, and SDS on the included subjects are presented in Table 1. Hearing thresholds are also reported as best ear hearing loss in the case of binaural fitting, but in the case of monaural fitting best ear means fitted ear. As in some previous studies (e.g.. Cox and Alexander, 2002; Jespersen et al, 2005), no effort was made to segregate the subjects by type of HA, type of signal processing, or any other fitting variable. All investigations were performed in accord-
15B
SRT (dB HL)
SDS (% correct)
Best ear or fitted ear
Best ear or fitted ear
Mean 69.3 SD 9.4 Range 39-87 Female 36.2% (n = 21) Male 63.8% (n = 37) Mean 48.4 SD 11.1 Range 26.3-72.5 Mean 47.6 SD 11.6 Range 20.0-72.5 Mean 46.1 SD 11.0 Range 20.0-68.8 Mean 30.3 SD Range Mean
14.3 0-60 88.6
SD Range
12.3 49-100 12.1% 87.9% 22.4% 77.6% 10.7 7.4 0.2-26
Hearing aid fitting Monaural Binaural First-time users Experienced users Years of use Mean SD Range
(n = 7) (n = 51) (n = 13) (n = 45)
Note: PTA was calculated as the average in dB HL for frequencies 0.5, 1, 2, and 4 kHz.
anee with the Helsinki declaration. The Scientific Ethical Review Board C for the Capital Region approved the project (H-C-2009-022.) Procedures and Equipment Initially, each subject completed a questionnaire concerning individual hearing status, HA use, and use of prescription drugs. Normal external ear canals and tjnnpanic membranes were then verified in all subjects by otoscopy followed by immittance measurements and pure-tone audiometry. Pure-tone audiometry was conducted in accordance with ISO 8253-1 (International Organization for Standardization, 1998) using the modified Hughson-Westlake technique (-10/+5 dB). Finally, SRT in quiet was performed followed by SDS in quiet. Pure-tone audiometry, SRT, and SDS were carried out using the GN Otometrics Astera Madsen Audiometer with Sennheiser HDA 200 circumaural earphones. SRT was measured using tracks 9-11 on the Dantale audio CD, which contains the validated speech material used in all Danish public hearing
Prediction of IOI-HA/Brännström et al
clinics (Elberling et al, 1989). The three tracks contain recordings of the ten monosyllabic numerals in Danish grouped in "triplets" (Olsen, 1996). Each presentation consists of one triplet (e.g., 2-1-3) that the listener is asked to repeat. The SRT is the lowest presentation level in dB HL at which the listener is able to repeat at least two out of three numerals. SDS was measured using tracks 1-8 on the Dantale audio CD. The tracks contain eight lists each of 25 one-syllable words containing a total of 80 phonemes. The presentation level when measuring the SDS is adjusted to the SRT + 30 dB. The test has to start with the first word of one of the eight lists, and the number of correct phonemes is counted (Keidser, 1993). At the end of the test, the percentage of correctly identified phonemes was calculated. The complete equipment setup (Astera Audiometer and the HDA 200 earphones) was calibrated in accordance with IEC 60318-2 (International Electrotechnical Commission, 1998), ISO 389-5 (International Organization for Standardization, 2006), and ISO 389-8 (International Organization for Standardization, 2004) using a Brüel and Kjaer 2610 measuring amplifier with a 4144 microphone in a 4152 acoustic coupler. A frequencymodulated 1000 Hz calibration tone (modulation rate = 20 Hz, modulation frequency = 250 Hz) from the Dantale CD with a 10 dB higher average root mean square than the speech and noise signals was used for the calibration of the speech signals where 0 dB HL = 15 dB SPL referring to the frequency-modulated 1000 Hz calibration tone on the CD (Olsen, 1996). All tests were performed in a double-walled soundproof booth, complying with the maximum permissible ambient sound pressure levels as specified in ISO 8253-1 (International Organization for Standardization, 1998). Acoustic refiexes were measured using the Madsen Electronics Zodiac 901 middle-ear analyzer. IOI-HA IOI-HA was designed to be universally applicable (Cox et al, 2000). The IOI-HA consists of seven items; hours of use (item 1), benefit (item 2), residual activity limitations (item 3), satisfaction (item 4), residual participation restriction (item 5), impact on others (item 6), and quality of life (item 7) (Cox et al, 2000; Cox and Alexander, 2002). Five responses are possible on each item ranging from worst (score 1) to best outcome (score 5). Results are reported as the global score, that is, the total sum of scores on all seven items (i.e., with a maximum of 35), where higher scores indicate better HA fitting outcome (Cox and Alexander, 2002). The psychometric properties of the IOI-HA have been evaluated for Danish-speaking populations ( Jespersen et al, 2005; Vestergaard, 2006). Previous studies suggest that the IOI-HA can be divided into two factors: introspec-
tion subscale ("me and my HA"; items 1, 2, 4, and 7) and interaction subscale ("me and the rest of the world"; items 3, 5, and 6) (Cox and Alexander, 2002; Kramer et al, 2002; Stephens, 2002; Vestergaard, 2006; Öberg et al, 2007; Brännström and Wennerström, 2010). A revised version of IOI-HA in which item 5 has been reformulated was used in the present study. The revised version was made by Jespersen et al (2014). Since Jespersen et al (2005) and Vestergaard (2006) demonstrated a negative item-total correlation for item 5 in the official version (Cox and Alexander, 2002), this indicates that subjects may have a problem understanding the response values and, hence, reporting residual disability. A negative correlation should not be present due to the direction of the response values (a higher score always represents a more positive response). This negative correlation has also been reported by Cox and Alexander (2002). Therefore, an IOI-HA with this rephrased version of item 5 was used to examine if it demonstrated more valid psychometric properties. The official version of item 5 reads, "Hvor meget har dine h0reproblemer med dit nuvaerende h0reapparat i de seneste to uger haft indfiydelse pa de ting, du kan g0re?" (Over the past two weeks, with your present hearing aid, how much have your hearing difficulties affected the things you can do?) with the response values "betydelig indfiydelse" (score 1, affected very much), "en hel del indfiydelse" (score 2, affected quite a lot), "moderat indfiydelse" (score 3, affected moderately), "lidt indfiydelse" (score 4, affected slightly), and "slet ingen indfiydelse" (score 5, affected not at all). The rephrased version reads, "Hvor meget har dine resterende h0reproblemer med dit nuvœrende h0reapparat begraenset dig i dit daglige liv i de seneste to uger?"(How much have your remaining hearing difficulties with your current hearing aid limited your daily life the last two weeks?) with the response values "begraenset mig betydeligt" (score 1, limited very much), "begraenset mig en hel del" (score 2, limited quite a lot), "begraenset mig moderat" (score 3, limited moderately), "begrsenset mig lidt" (score 4, limited slightly), and "slet ikke begraenset mig" (score 5, limited not at all). RESULTS Psychometric Properties Average scores and standard deviations (SDs) for each IOI-HA item, global score, and subscales are presented in Table 2. It can be seen that average scores for the individual items lie between 3.4 and 4.5. Associations between items were investigated using Pearson product-moment correlation coefficients (rho). Negative correlations should not be present in the IOI-HA due to the direction of its response values (where a higher
157
Journal of the American Academy of Audiology/Volume 25, Number 2, 2014
Table 2. Means, SDs, and Ranges for the Separate lOI-HA Items, Global Score (maximum score 35), and Introspection and Interaction Subjects (n = 58) 1. Use 2. Benefit 3. Residual activity restriction 4. Satisfaction 5. Residual participation restriction 6. Impact on others 7. Quality of life Global Introspection (items 1, 2, 4, and 7) Interaction (3, 5, and 6)
Mean
SD
Range
4.4 4.3 3.4 4.5 3.9 3.9 4.1 28.5 17.3 11.2
1.1 1.0 0.9 0.8 1.0 0.9 0.8 4.3 2.9 2.4
1-5 1-5 1-5 2-5 1-5 2-5 2-5 16-35 8-20 4-15
score always is a more positive response). No negative associations were seen. The IOI-ELA global score showed significant positive associations with the introspection subscale, the interaction subscale, and each of the separate items (degrees of freedom IDF] = 9, rho > 0.562, p :< 0.01). The IOI-HA introspection subscale showed significant positive associations with tbe IOI-HA interaction subscale and each of the separate items except item 5 (DF = 9, rho > 0.327, p < 0.05). The IOI-HA interaction subscale showed significant positive associations with each of the separate items except item 1 (DF = 9, rho > 0.285, p < 0.05). Most IOI-HA items correlated to each other significantly (DF, = 9, rho > 0.264, p < 0.05) although exceptions were seen: Item 1 (hours of use), 2 (benefit), 4 (satisfaction), and 7 (quality of life) did not show any association with item 5 (residual participation restriction), and item 1 was not correlated to item 3 (residual activity limitations). Also, the corrected item-total correlation coefficients, that is, the association between tbe score of a separate item and the total score of all other items were calculated. For all items except item 5, the item-total correlation coefficients ranged from rho 0.453 to rho 0.633 (p < 0.001). For item 5, this coefficient (rho) was 0.365 (p = 0.005). A principal components analysis (PCA) using varimax rotation and Kaiser normalization was made to confirm that the IOI-HA items were represented in the previously reported subscales (introspection and interaction subscales). The unrotated PCA verified that two factors met Kaiser's criterion (eigenvalues exceed 1.0) and lay above the Cattell's point of inñection. These two factors together accounted for 67.8% of the total variance. After rotation, factor 1 (items 1, 2, 4, and 7) accounted for 37.4% of the variance in scores, and factor 2 (items 3, 5, and 6) accounted for 30.4%. The factor loadings above 0.60 verified the previously reported division of the questionnaire into the two subscales. Cronbach's alpha was used to estimate the internal
15B
consistency reliability for the total scale (alpha = 0.790), the items loading on factor 1 (alpha = 0.809), and the items loading on factor 2 (alpha = 0.764). Differences and Associations It was investigated if there were any differences between male and female subjects, and between subjects who were fitted monaurally and binaurally with HA. Using independent-samples T-tests with Bonferroni correction, no differences were found in the reported scores on separate IOI-HA items, the global score, the introspection subscale, the interaction subscale, age, years of HA use, best ear PTA, best ear SRT, or best ear SDS. Also, it was investigated if there were any differences between first-time HA users and experienced HA users. Again using independent-samples T-tests with Bonferroni correction, no differences were found in the reported scores on separate IOI-HA items, the global score, the introspection subscale, or the interaction subscale. The experienced HA users had used HAs significantly longer than the first-time users (on average 12.8 and 3.5 yr, respectively) and had poorer best ear SRT (on average 33 and 19 dB HL, respectively) (p < 0.001). A correlation analysis was conducted (for variables age, years of HA use, time since last HA fitting, best ear PTA, best ear SRT, best ear SDS, and IOI-HA global score, introspection subscale, interaction subscale, and separate items) using Pearson product-moment correlation coefficients (rho). Table 3 shows the result of this analysis. Age showed only a significant negative association with best ear SDS (i.e., the best ear for binaurally fitted but the fitted ear for the monaurally fitted), suggesting that the SDS becomes poorer with increasing age. Unsurprisingly, the number of years of HA use showed a significant positive association with time since last HA fitting. Tbe number of years of HA use also showed significant positive associations with best ear PTA and best ear SRT, indicating that subjects with poorer PTAs and SRTs have used HAs longer. Time since last HA fitting showed significant positive associations with IOI-HA items 2 and 7, suggesting that subjects who have used their current HAs longer report more benefit (item 2) and higher quality of life (item 7). Best ear PTA showed a significant positive association with best ear SRT and a significant negative association with best ear SDS. This suggests that subjects with poorer PTAs also have poorer SRTs and SDSs. Best ear SDS showed significant positive associations with the interaction subscale and two of its separate items (items 3 and 5), suggesting that better SDSs are accompanied by fewer limitations and restrictions in their interaction with others. No other significant associations were found.
Prediction of IOI-HA/Brännström et al
o c5 Í2 co
0
CL
J=
E
o
co "ro o c T3
'•^ ro
CC
'.Í:
o -ji
CM o o CO
có .9- o 0 o w oó
I—
T—
CO
CO
ro S. CL
^ CO CO i n
g Tt 'i:
CO CM
CO LO
o
CM CD CM
O " Í CM in -^
r-o o
f— CM CM in
CM LO CD -^ r^
CO CO CO CO CO
in Oí CO CM
CM CO CO CM o C\J CO CO CO
in CTi CM N.
o CO
co LO K
CO CO
CO CO
CO CO
in LO CO o in
in
o
b CsO
•^
o o CO
CO
CO
CO
m c
g 0 o "ro 0 o co o. co o Z!
CO CO
o O 00 CO OJ T- o
cn
CO
CO
q q
c\j
CM CO
o
ce
o
CO
CO
1
CO c\j
t^
LU
-îi LO
^
CO CO
o
cji lo
O CO CO CM •^ O
q
I
m
CO in
CM en t~~ en
o -^ o
r
in oa CO o o c\j
I
CO en
o o
o
m CO
CD
co CM CT)
í CO CD O
co Ȓf o
1- O CD O CO ^ i
o
CO CO O) CO 1 - CO 1 - T- - ^
T- in r-- CO CO ^ • * h- Lo in c\j o -^ o -^
CM CO ••-
CO CO
O
o i 5
CO CM
O
• ^ LO LO O o CM
CM O
h-
g
u0
S O
-^ c\j CO h^ o CM
o
CO CO
CO
LO
^
CD
Ö
CD
0
q q
CD CO
C3) CJ)
cj
o o
> co D)
X
c X
(year E PTA ESRT ESDS
'co"
CÛ
co
o
m m O
0 Q.
c o
Benef
C/)
0
Use
5i
CO O CO CO CD CJl 'S- CO - i t^ 1 - 1 - O ' - T-
core Ction subscale :eractii subs cale
O)
.£ ro
riabl
X
\ use (yeai
Prediction of IOI-HA Scores Using Speech Reception Thresholds and Speech Discrimination Scores in Quiet DOI: 10.3766/jaaa.25.2.4 K. Jonas Brännström*§ Johannes Lantzt§ Lars Holme Nielsen^ Steen 0stergaard
Abstract Background: Outcome measures can be used to improve the quality of tlie rehabiiitation by identifying and understanding which variables influence the outcome. This information can be used to improve outcomes for clients. In clinical practice, pure-tone audiometry, speech reception thresholds (SRTs), and speech discrimination scores (SDSs) in quiet or in noise are common assessments made prior to hearing aid (HA) fittings. It is not known whether SRT and SDS in quiet relate to HA outcome measured with the International Outcome Inventory for Hearing Aids (IOI-HA). Purpose: The aim of the present study was to investigate the relationship between pure-tone average (PTA), SRT, and SDS in quiet and IOI-HA in both first-time and experienced HA users. Research Design: SRT and SDS were measured in a sample of HA users who also responded to the IOI-HA. Study Sample: Fifty-eight Danish-speaking adult HA users. Data Collection and Analysis: The psychometric properties were evaluated and compared to previous studies using the IOI-HA. The associations and differences between the outcome scores and a number of descriptive variables (age, gender, fitted monaurally/binaurally with HA, first-time/experienced HA users, years of HA use, time since last HA fitting, best ear PTA, best ear SRT, or best ear SDS) were examined. A multiple forward stepwise regression analysis was conducted using scores on the separate IOI-HA items, the global score, and scores on the introspection and interaction subscales as dependent variables to examine whether the descriptive variables could predict these outcome measures. Results: Scores on single IOI-HA items, the global score, and scores on the introspection (items 1,2,4, and 7) and interaction (items 3, 5, and 6) subscales closely resemble those previously reported. Multiple regression analysis showed that the best ear SDS predicts about 18-19% of the outcome on items 3 and 5 separately, and about 16% on the interaction subscale (sum of items 3, 5, and 6) Conclusions: The best ears SDS explains some of the variance displayed in the IOI-HA global score and the interaction subscale. The relation between SDS and IOI-HA suggests that a poor unaided SDS might in itself be a limiting factor for the HA rehabilitation efficacy and hence the IOI-HA outcome. The clinician could use this information to align the user's HA expectations to what is within possible reach. Key Words: Hearing aids, outcome, satisfaction Abbreviations: HA = hearing aid; IOI-HA = International Outcome Inventory for Hearing Aids; PCA = principal components analysis; PTA = pure-tone average; SDS = speech discrimination score; SRT = speech reception threshold
'Department of Logopedics, Phoniatrics and Audiology, Clinical Science in Lund, Lund University, Lund, Sweden; tGI\l Otometrics, Taastrup, Denmark; ^Research Laboratory, Department of Otorhinolaryngology, Head and Neck Surgery, University Hospital, Rigshospitalet, Copenhagen, Denmark; §HEARsound Laboratories, Network for Joint Hearing Research in the Oresund Region K. Jonas Brännström, Clinical Science in Lund, Department of Logopedics, Phoniatrics and Audiology, SE-221 00 Lund, Sweden; Phone; +46(0)46-177103; Fax; +46(0)46-171732; E-mail; [email protected]
154
Prediction of IOI-HA/Brännström et al
T
he International Outcome Inventory for Hearing Aids (IOI-HA) is a hearing-specific questionnaire developed by Cox et al (2000) with the purpose of assessing hearing aid (HA) rehabilitation efficacy. As an outcome measure, it can be used by the clinician to assess how well the rehabilitation has worked from the client's perspective (Uriarte et al, 2005). In a similar way, it can be used to improve the quality of the rehabilitation (Beck, 2000). One way to achieve the latter is by identifying and understanding which variables infiuence the outcome and use this knowledge to improve outcomes for clients (Hickson et al, 2010). In clinical practice, prior to HA fittings, pure-tone audiometry, speech reception thresholds (SRTs), and speech discrimination scores (SDSs) in quiet or in noise are often determined. It is not known whether SRT and SDS in quiet relate to and can be used to predict the HA outcome measured with the IOI-HA. If, for example, poor scores on these measures are related to poor scores on the IOI-HA, the clinician can use this information to improve interventions that also improve the clients' outcome. The psychometric properties of the IOI-HA have been examined in different languages (Cox and Alexander, 2002; Kramer et al, 2002, Stephens, 2002; Heuermann et al, 2005, Jespersen et al, 2005; Vestergaard, 2006; Öberg et al, 2007; Smith et al, 2009; Brännström and Wennerström, 2010). These studies clearly demonstrate that the seven items of the questionnaire can be divided into two subscales: introspection ("me and my hearing aid"; items 1, 2, 4, and 7) and interaction ("me and the rest of the world"; items 3, 5, and 6). However, as Jespersen et al (2005) and Vestergaard (2006) showed, the official Danish version presents a negative item-total correlation for one of the items (item 5, residual participation restriction). A negative correlation should not be present due to the direction of the response values (a higher score always represents a more positive response). For this particular item this indicates that subjects may have problems understanding the response values and hence reporting residual disability. This negative correlation has also been reported by Cox and Alexander (2002). Addressing this particular issue, Jespersen et al (2014) rephrased item 5.
The relationship between the IOI-HA and several demographic variables has been examined. Although some previous studies indicate that older subjects use their HAs less and report less benefit than younger subjects (Stephens, 2002; Öberg et al, 2007), others did not (Kramer et al, 2002; Heuermann et al, 2005; Cook and Hawkins, 2007; Brännström and Wennerström, 2010; Hickson et al, 2010). Increasing levels of hearing loss lead to increased HA usage, but the association between the IOI-HA and pure-tone hearing thresholds seems unclear
(Humes et al, 2002; Kramer et al, 2002; Larson et al, 2002; Stephens, 2002; Stark and Hickson, 2004; Heuermann et al, 2005; Meister et al, 2005; Metselaar et al, 2009; Smith et al, 2009; Brännström and Wennerström, 2010; Hickson et al, 2010). Also, the type of hearing loss explains a minor part of the variance (less than 5%) in scores on the IOI-HA introspection subscale (items 1, 2, 4, and 7); subjects with sensorineural hearing loss reported poorer scores than subjects with conductive or mixed hearing loss (Brännström and Wennerström, 2010). However, monaural or binaural HA fitting does not inñuence the IOI-HA global outcome (Heuermann et al, 2005; Cook and Hawkins, 2007; Brännström and Wennerström, 2010; Hickson et al, 2010), but binaurally fitted subjects seem to report higher scores on single items (residual activity limitation, item 3) (Liu et al, 2011), and they report longer HA use durations (HA use, item 1) (Stephens, 2002). Also, binaurally fitted subjects who have had longer hearing loss duration are more likely to demonstrate higher scores on the introspection subscale (Kramer et al, 2002; Stephens, 2002). It is unclear whether there is a gender effect on the IOI-HA (Cook and Hawkins, 2007; Liu et al, 2011) or not (Kramer et al, 2002; Heuermann et al, 2005; Williams et al, 2009; Brännström and Wennerström, 2010; Hickson et al, 2010). The IOI-HA global score increases slightly but significantly with increasing price of the HA, when subjects themselves finance their devices (Liu et al, 2011). Though it is a common clinical experience that HAs cannot fully compensate for hearing impairment, it is rarely known which factors, specifically, limit the HA outcome. It would therefore be relevant to be able to identify factors that infiuence the HA rehabilitation efficacy, for example, as measured with the IOI-HA. Taylor (2008) demonstrated an association between IOI-HA and the acceptable noise level (ANL) test (Nabelek et al, 1991; Nabelek, 2006), afindingthat Olsen and colleagues (2012) could not replicate in a larger sample. If the ANL shows that a subject only accepts little noise in the presence of speech, and this in turn affects the IOIHA outcome, then we might be able to increase the HA benefit for this particular subject by emplojdng noise reducing algorithms. Hickson and colleagues (2010) explained about 57% of the variance in the overall IOI-HA based on findings that higher average IOIHA scores are associated with (1) greater satisfaction when having a conversation with a single person, in small groups, in large groups, and outdoors and (2) greater satisfaction with HA attributes such as comfort with loud sounds, comfort of the physical fit, and clarity of tone and sound. Hence, the speech perception ability in quiet (as measured with the SRT and SDS) might affect the IOI-HA. If a relation between SRT and/or SDS and IOI-HA can be established, then we will know that a poor SRT and/or SDS might be a limiting factor
155
Journal of the American Academy of AudiologyA^olume 25, Number 2, 2014
for the HA rehabilitation efficacy and hence the IOI-HA outcome. With this information in hand we would better be able to explain why an HA in such a case does not compensate for the hearing loss to the same extent as when users have a good SRT/SDS. The aim of the present study was therefore (1) to provide preliminary psychometric properties of the revised Danish IOI-HA and (2) to investigate if the outcome of the SRT and SDS in quiet affects IOI-HA in first-time and experienced HA users.
Table 1. Demographic Data for All Included Users (n = 58) Variable Age (years)
Gender PTA (dB HL)
Right ear
Left ear
MATERIAL AND METHODS Best ear or fitted ear
Subjects Potential subjects were identified through our patient files at Rigshospitalet in Copenhagen. Initially, 85 HA users were recruited to the study. AU subjects were native Danish speakers. Inclusion criteria were sensorineural hearing loss, normal tympanograms (within ± 150 daPa) according to Margolis and Goycoolea (1993), and recordable contralateral acoustic refiexes at 1 kHz. Also, the subjects should have been fitted with HAs, and they were required to actually use them. No other inclusion criteria were used. Fifty-eight subjects met all these inclusion criteria. Of the 27 excluded subjects, 20 subjects did not meet all inclusion criteria, three subjects had too poor hearing to be able to adjust the speech signals to the most comfortable level in one part of the test, three additional subjects had too poor SRTs to be able to present SDSs at the maximum output level of the audiometer (i.e., 90 dB SPL), and one subject was not able to perform the measurement at all. The IOI-HA was administered at Rigshospitalet in Copenhagen in conjunction with the earlier mentioned project (Olsen et al, 2012). At the time that the IOI-HA was administered, fifty-one participants (88%) were binaurally fitted with HAs, and seven participants (12%) were monaurally fitted with HAs. Thirteen (22%) of the participants were fitted with HAs for the first time, while forty-five (78%) of the subjects had renewed their initial devices at least once over the period. The participants had HAs for a mean period of 10.7 yr (SD = 7.4, Range = 0.2-26), and the mean period of time since the last HA aid fitting was 2.4 yr (SD = 2.5; Range = 0.2-16). Demographic data and information on hearing thresholds (pure-tone average [PTA]), SRT, and SDS on the included subjects are presented in Table 1. Hearing thresholds are also reported as best ear hearing loss in the case of binaural fitting, but in the case of monaural fitting best ear means fitted ear. As in some previous studies (e.g.. Cox and Alexander, 2002; Jespersen et al, 2005), no effort was made to segregate the subjects by type of HA, type of signal processing, or any other fitting variable. All investigations were performed in accord-
15B
SRT (dB HL)
SDS (% correct)
Best ear or fitted ear
Best ear or fitted ear
Mean 69.3 SD 9.4 Range 39-87 Female 36.2% (n = 21) Male 63.8% (n = 37) Mean 48.4 SD 11.1 Range 26.3-72.5 Mean 47.6 SD 11.6 Range 20.0-72.5 Mean 46.1 SD 11.0 Range 20.0-68.8 Mean 30.3 SD Range Mean
14.3 0-60 88.6
SD Range
12.3 49-100 12.1% 87.9% 22.4% 77.6% 10.7 7.4 0.2-26
Hearing aid fitting Monaural Binaural First-time users Experienced users Years of use Mean SD Range
(n = 7) (n = 51) (n = 13) (n = 45)
Note: PTA was calculated as the average in dB HL for frequencies 0.5, 1, 2, and 4 kHz.
anee with the Helsinki declaration. The Scientific Ethical Review Board C for the Capital Region approved the project (H-C-2009-022.) Procedures and Equipment Initially, each subject completed a questionnaire concerning individual hearing status, HA use, and use of prescription drugs. Normal external ear canals and tjnnpanic membranes were then verified in all subjects by otoscopy followed by immittance measurements and pure-tone audiometry. Pure-tone audiometry was conducted in accordance with ISO 8253-1 (International Organization for Standardization, 1998) using the modified Hughson-Westlake technique (-10/+5 dB). Finally, SRT in quiet was performed followed by SDS in quiet. Pure-tone audiometry, SRT, and SDS were carried out using the GN Otometrics Astera Madsen Audiometer with Sennheiser HDA 200 circumaural earphones. SRT was measured using tracks 9-11 on the Dantale audio CD, which contains the validated speech material used in all Danish public hearing
Prediction of IOI-HA/Brännström et al
clinics (Elberling et al, 1989). The three tracks contain recordings of the ten monosyllabic numerals in Danish grouped in "triplets" (Olsen, 1996). Each presentation consists of one triplet (e.g., 2-1-3) that the listener is asked to repeat. The SRT is the lowest presentation level in dB HL at which the listener is able to repeat at least two out of three numerals. SDS was measured using tracks 1-8 on the Dantale audio CD. The tracks contain eight lists each of 25 one-syllable words containing a total of 80 phonemes. The presentation level when measuring the SDS is adjusted to the SRT + 30 dB. The test has to start with the first word of one of the eight lists, and the number of correct phonemes is counted (Keidser, 1993). At the end of the test, the percentage of correctly identified phonemes was calculated. The complete equipment setup (Astera Audiometer and the HDA 200 earphones) was calibrated in accordance with IEC 60318-2 (International Electrotechnical Commission, 1998), ISO 389-5 (International Organization for Standardization, 2006), and ISO 389-8 (International Organization for Standardization, 2004) using a Brüel and Kjaer 2610 measuring amplifier with a 4144 microphone in a 4152 acoustic coupler. A frequencymodulated 1000 Hz calibration tone (modulation rate = 20 Hz, modulation frequency = 250 Hz) from the Dantale CD with a 10 dB higher average root mean square than the speech and noise signals was used for the calibration of the speech signals where 0 dB HL = 15 dB SPL referring to the frequency-modulated 1000 Hz calibration tone on the CD (Olsen, 1996). All tests were performed in a double-walled soundproof booth, complying with the maximum permissible ambient sound pressure levels as specified in ISO 8253-1 (International Organization for Standardization, 1998). Acoustic refiexes were measured using the Madsen Electronics Zodiac 901 middle-ear analyzer. IOI-HA IOI-HA was designed to be universally applicable (Cox et al, 2000). The IOI-HA consists of seven items; hours of use (item 1), benefit (item 2), residual activity limitations (item 3), satisfaction (item 4), residual participation restriction (item 5), impact on others (item 6), and quality of life (item 7) (Cox et al, 2000; Cox and Alexander, 2002). Five responses are possible on each item ranging from worst (score 1) to best outcome (score 5). Results are reported as the global score, that is, the total sum of scores on all seven items (i.e., with a maximum of 35), where higher scores indicate better HA fitting outcome (Cox and Alexander, 2002). The psychometric properties of the IOI-HA have been evaluated for Danish-speaking populations ( Jespersen et al, 2005; Vestergaard, 2006). Previous studies suggest that the IOI-HA can be divided into two factors: introspec-
tion subscale ("me and my HA"; items 1, 2, 4, and 7) and interaction subscale ("me and the rest of the world"; items 3, 5, and 6) (Cox and Alexander, 2002; Kramer et al, 2002; Stephens, 2002; Vestergaard, 2006; Öberg et al, 2007; Brännström and Wennerström, 2010). A revised version of IOI-HA in which item 5 has been reformulated was used in the present study. The revised version was made by Jespersen et al (2014). Since Jespersen et al (2005) and Vestergaard (2006) demonstrated a negative item-total correlation for item 5 in the official version (Cox and Alexander, 2002), this indicates that subjects may have a problem understanding the response values and, hence, reporting residual disability. A negative correlation should not be present due to the direction of the response values (a higher score always represents a more positive response). This negative correlation has also been reported by Cox and Alexander (2002). Therefore, an IOI-HA with this rephrased version of item 5 was used to examine if it demonstrated more valid psychometric properties. The official version of item 5 reads, "Hvor meget har dine h0reproblemer med dit nuvaerende h0reapparat i de seneste to uger haft indfiydelse pa de ting, du kan g0re?" (Over the past two weeks, with your present hearing aid, how much have your hearing difficulties affected the things you can do?) with the response values "betydelig indfiydelse" (score 1, affected very much), "en hel del indfiydelse" (score 2, affected quite a lot), "moderat indfiydelse" (score 3, affected moderately), "lidt indfiydelse" (score 4, affected slightly), and "slet ingen indfiydelse" (score 5, affected not at all). The rephrased version reads, "Hvor meget har dine resterende h0reproblemer med dit nuvœrende h0reapparat begraenset dig i dit daglige liv i de seneste to uger?"(How much have your remaining hearing difficulties with your current hearing aid limited your daily life the last two weeks?) with the response values "begraenset mig betydeligt" (score 1, limited very much), "begraenset mig en hel del" (score 2, limited quite a lot), "begraenset mig moderat" (score 3, limited moderately), "begrsenset mig lidt" (score 4, limited slightly), and "slet ikke begraenset mig" (score 5, limited not at all). RESULTS Psychometric Properties Average scores and standard deviations (SDs) for each IOI-HA item, global score, and subscales are presented in Table 2. It can be seen that average scores for the individual items lie between 3.4 and 4.5. Associations between items were investigated using Pearson product-moment correlation coefficients (rho). Negative correlations should not be present in the IOI-HA due to the direction of its response values (where a higher
157
Journal of the American Academy of Audiology/Volume 25, Number 2, 2014
Table 2. Means, SDs, and Ranges for the Separate lOI-HA Items, Global Score (maximum score 35), and Introspection and Interaction Subjects (n = 58) 1. Use 2. Benefit 3. Residual activity restriction 4. Satisfaction 5. Residual participation restriction 6. Impact on others 7. Quality of life Global Introspection (items 1, 2, 4, and 7) Interaction (3, 5, and 6)
Mean
SD
Range
4.4 4.3 3.4 4.5 3.9 3.9 4.1 28.5 17.3 11.2
1.1 1.0 0.9 0.8 1.0 0.9 0.8 4.3 2.9 2.4
1-5 1-5 1-5 2-5 1-5 2-5 2-5 16-35 8-20 4-15
score always is a more positive response). No negative associations were seen. The IOI-ELA global score showed significant positive associations with the introspection subscale, the interaction subscale, and each of the separate items (degrees of freedom IDF] = 9, rho > 0.562, p :< 0.01). The IOI-HA introspection subscale showed significant positive associations with tbe IOI-HA interaction subscale and each of the separate items except item 5 (DF = 9, rho > 0.327, p < 0.05). The IOI-HA interaction subscale showed significant positive associations with each of the separate items except item 1 (DF = 9, rho > 0.285, p < 0.05). Most IOI-HA items correlated to each other significantly (DF, = 9, rho > 0.264, p < 0.05) although exceptions were seen: Item 1 (hours of use), 2 (benefit), 4 (satisfaction), and 7 (quality of life) did not show any association with item 5 (residual participation restriction), and item 1 was not correlated to item 3 (residual activity limitations). Also, the corrected item-total correlation coefficients, that is, the association between tbe score of a separate item and the total score of all other items were calculated. For all items except item 5, the item-total correlation coefficients ranged from rho 0.453 to rho 0.633 (p < 0.001). For item 5, this coefficient (rho) was 0.365 (p = 0.005). A principal components analysis (PCA) using varimax rotation and Kaiser normalization was made to confirm that the IOI-HA items were represented in the previously reported subscales (introspection and interaction subscales). The unrotated PCA verified that two factors met Kaiser's criterion (eigenvalues exceed 1.0) and lay above the Cattell's point of inñection. These two factors together accounted for 67.8% of the total variance. After rotation, factor 1 (items 1, 2, 4, and 7) accounted for 37.4% of the variance in scores, and factor 2 (items 3, 5, and 6) accounted for 30.4%. The factor loadings above 0.60 verified the previously reported division of the questionnaire into the two subscales. Cronbach's alpha was used to estimate the internal
15B
consistency reliability for the total scale (alpha = 0.790), the items loading on factor 1 (alpha = 0.809), and the items loading on factor 2 (alpha = 0.764). Differences and Associations It was investigated if there were any differences between male and female subjects, and between subjects who were fitted monaurally and binaurally with HA. Using independent-samples T-tests with Bonferroni correction, no differences were found in the reported scores on separate IOI-HA items, the global score, the introspection subscale, the interaction subscale, age, years of HA use, best ear PTA, best ear SRT, or best ear SDS. Also, it was investigated if there were any differences between first-time HA users and experienced HA users. Again using independent-samples T-tests with Bonferroni correction, no differences were found in the reported scores on separate IOI-HA items, the global score, the introspection subscale, or the interaction subscale. The experienced HA users had used HAs significantly longer than the first-time users (on average 12.8 and 3.5 yr, respectively) and had poorer best ear SRT (on average 33 and 19 dB HL, respectively) (p < 0.001). A correlation analysis was conducted (for variables age, years of HA use, time since last HA fitting, best ear PTA, best ear SRT, best ear SDS, and IOI-HA global score, introspection subscale, interaction subscale, and separate items) using Pearson product-moment correlation coefficients (rho). Table 3 shows the result of this analysis. Age showed only a significant negative association with best ear SDS (i.e., the best ear for binaurally fitted but the fitted ear for the monaurally fitted), suggesting that the SDS becomes poorer with increasing age. Unsurprisingly, the number of years of HA use showed a significant positive association with time since last HA fitting. Tbe number of years of HA use also showed significant positive associations with best ear PTA and best ear SRT, indicating that subjects with poorer PTAs and SRTs have used HAs longer. Time since last HA fitting showed significant positive associations with IOI-HA items 2 and 7, suggesting that subjects who have used their current HAs longer report more benefit (item 2) and higher quality of life (item 7). Best ear PTA showed a significant positive association with best ear SRT and a significant negative association with best ear SDS. This suggests that subjects with poorer PTAs also have poorer SRTs and SDSs. Best ear SDS showed significant positive associations with the interaction subscale and two of its separate items (items 3 and 5), suggesting that better SDSs are accompanied by fewer limitations and restrictions in their interaction with others. No other significant associations were found.
Prediction of IOI-HA/Brännström et al
o c5 Í2 co
0
CL
J=
E
o
co "ro o c T3
'•^ ro
CC
'.Í:
o -ji
CM o o CO
có .9- o 0 o w oó
I—
T—
CO
CO
ro S. CL
^ CO CO i n
g Tt 'i:
CO CM
CO LO
o
CM CD CM
O " Í CM in -^
r-o o
f— CM CM in
CM LO CD -^ r^
CO CO CO CO CO
in Oí CO CM
CM CO CO CM o C\J CO CO CO
in CTi CM N.
o CO
co LO K
CO CO
CO CO
CO CO
in LO CO o in
in
o
b CsO
•^
o o CO
CO
CO
CO
m c
g 0 o "ro 0 o co o. co o Z!
CO CO
o O 00 CO OJ T- o
cn
CO
CO
q q
c\j
CM CO
o
ce
o
CO
CO
1
CO c\j
t^
LU
-îi LO
^
CO CO
o
cji lo
O CO CO CM •^ O
q
I
m
CO in
CM en t~~ en
o -^ o
r
in oa CO o o c\j
I
CO en
o o
o
m CO
CD
co CM CT)
í CO CD O
co Ȓf o
1- O CD O CO ^ i
o
CO CO O) CO 1 - CO 1 - T- - ^
T- in r-- CO CO ^ • * h- Lo in c\j o -^ o -^
CM CO ••-
CO CO
O
o i 5
CO CM
O
• ^ LO LO O o CM
CM O
h-
g
u0
S O
-^ c\j CO h^ o CM
o
CO CO
CO
LO
^
CD
Ö
CD
0
q q
CD CO
C3) CJ)
cj
o o
> co D)
X
c X
(year E PTA ESRT ESDS
'co"
CÛ
co
o
m m O
0 Q.
c o
Benef
C/)
0
Use
5i
CO O CO CO CD CJl 'S- CO - i t^ 1 - 1 - O ' - T-
core Ction subscale :eractii subs cale
O)
.£ ro
riabl
X
\ use (yeai
