C HAPTER T WENTY-TWO

Transcription

1 C HAPTER T WENTY-TWO Hearing-Aid Outcome Measures in Older Adults Larry E. Humes For the past decade or so, my colleagues and I at Indiana University have been conducting research on hearing-aid outcome measures in older adults. This seemed like a fertile area for further research when we embarked on this journey a decade ago. Several observations supported this supposition. First, it was known that about two-thirds of the hearing aids sold in the U.S. were purchased by individuals 65 years of age or older (e.g., Skafte 2000), but that only about half of those who did so reported satisfaction with the result (e.g., Kochkin 1993a, 1993b, 1993c). Further, even though older adults were the primary purchasers of hearing aids, only about 20% of those older adults who could benefit from hearing aids sought them out (Kochkin 1992). Our primary research interest in this area was the identification of factors that were associated with individual differences in hearing-aid outcome among older adults. What is it that accounts for sizable differences in outcome often observed between two seemingly homogeneous older hearing aid wearers? Answering this question might lead to the development of ways to maximize outcome or, at the very least, to recognize a priori when success may be limited without additional intervention. As we began to address the issue of individual differences in hearing-aid outcome and the factors underlying them, it was apparent that we would first need to be able to measure hearing-aid outcome in older hearing aid wearers. Reviewing the literature available at the time this project was initiated revealed a myriad of choices available for potential outcome measures. There Address correspondence to:, Larry E. Humes, Ph.D., Professor, Department of Speech and Hearing Sciences, Indiana University, Bloomington, IN 47405, USA. humes@indiana.edu. were numerous instruments available, for example, that could be considered self-report measures of the relative benefit provided by hearing aids. Measures of aided hearing handicap reduction (Malinoff and Weinstein 1989; Newman, Jacobson, Hug, Weinstein and Malinoff 1991), using the Hearing Handicap Inventory for the Elderly (HHIE; Ventry and Weinstein 1982), represented one of the more common measures of self-reported benefit used at the time. Other measures of self-reported benefit using the Profile of Hearing Aid Benefit (PHAB; Cox, Gilmore and Alexander 1991; Cox and Rivera 1992) or the abbreviated version of this profile, the APHAB (Cox and Alexander 1995), were also available. In addition, not only was the Hearing Aid Performance Inventory (HAPI; Walden, Demorest and Hepler 1984) available, which, despite its name is actually a measure of relative benefit or helpfulness of hearing aids, but two shortened versions of this inventory had been developed for use with older adults (Schum 1992; Dillon 1994). Dillon and colleagues also developed an open-response self-report measure of benefit or relative improvement, the Client Oriented Scale of Improvement (COSI; Dillon, James and Ginis 1997). Although there were still more measures of relative benefit available at the time our initial work in this area was being planned, the HHIE, HAPI, PHAB, APHAB and COSI were among the most common and, in our opinion, among the most carefully developed measures. Self-reported benefit, however, whether measured in terms of handicap reduction or in terms of perceived helpfulness of the hearing aid in a variety of listening situations, seemed to represent just one potential piece of the hearing-aid outcome puzzle. Certainly, other potential pieces of the puzzle could include self-report measures of hearing aid satisfaction or usage. In the area of hearing aid satisfaction, there were many fewer widely 265

2 266 Hearing Care for Adults used outcome measures from which to choose, although many had been developed over the years (Wong, Hickson and McPherson 2003). The most widely used measure was (and is) unquestionably the MarkeTrak series of satisfaction surveys composed by Kochkin (1993a, 1993b, 1993c, 1997, 2000) which tens of thousands of hearing aid purchasers have completed. A more recently developed tool designed to measure satisfaction with hearing aids was the Satisfaction with Amplification in Daily Living (SADL) scale developed by Cox and Alexander (1999), with subsequent research on its utility by Hosford-Dunn and Halpern (2000, 2001) and Cox and Alexander (2001). The MarkeTrak measures of satisfaction made use of a five-point Likert scale of satisfaction, with response options ranging from very satisfied to very dissatisfied. The SADL, on the other hand, was an indirect measure of satisfaction which never made use of the word satisfaction in the ratings completed by the hearing aid wearer. Humes, Wilson, Humes, Barlow, Garner and Amos (2002), however, demonstrated that there were strong positive correlations (r = 0.75) between these two measures of satisfaction, despite using entirely different response formats. With regard to hearing-aid usage, there were no well-defined scales available a decade ago. Both self-report measures, ranging from daily logs to single-item recollections of daily usage over a specific period of time, and objective measures, including changes in battery weight and digital data loggers, had been described previously. Finally, some self-report scales of hearing aid outcome had been developed that tapped several potential domains in a single scale. Most notable in this regard was the Glasgow Hearing Aid Benefit Profile (GHABP; Gatehouse 1999) which measures, among other things, self-reported hearing-aid benefit, satisfaction, and usage. Another attractive feature of this self-report instrument is that it includes four predetermined listening situations in which the hearing aid must be evaluated by all respondents and four additional self-selected listening situations identified as being important to the specific hearing aid wearer. The self-selected items of the GHABP appeared to us to capture some of the same positive aspects of the open-response format used in the COSI. In addition to all of these potential measures of selfreported hearing aid benefit, satisfaction, and usage, objective measures of benefit based on measured speechrecognition scores also were viable candidates for a comprehensive assessment of hearing-aid outcome. In this case, one can examine the aided speech-recognition performance alone or the relative benefit by comparing the aided to the unaided speech-recognition performance. More often, the latter measure of relative benefit is employed because it is difficult to interpret the aided performance scores in isolation. On the other hand, it is the aided speech-recognition performance alone that reflects the impact of the hearing aid. This potential measure of hearing aid outcome itself had a myriad of choices available. Should the materials used to measure aided and unaided speech-recognition performance be nonsense syllables, words, sentences or connected discourse? Should the testing be completed in quiet or in a noise background? What level of speech should be used? If noise is to be used, what signal-to-noise ratio, what type of noise, and from what azimuth should it be presented? With so many potential outcome measures from which to choose, we were quickly beginning to have second thoughts about conducting research on individual differences in hearing-aid outcome. The documentation of hearing-aid outcome, after all, was just the initial step, but a very important one, on this journey. To make matters worse, although there were numerous outcome measures in various domains with which to work, with few exceptions (e.g., Cox and Alexander 1991; Gatehouse 1994; Dillon et al. 1997), studies comparing multiple measures of outcome in the same hearing-aid wearers were scarce. So, a fundamental question left unaddressed was whether all of these outcome measures were necessary. Most likely, there was considerable redundancy to be found among measures within a particular domain, such as self-reported benefit, but it was also possible that there could be considerable overlap in measures across domains. Basically, we were in a quandary as to what to measure to document individual differences in outcome. This was not our only quandary, however, for these preliminary steps necessary for the identification of factors underlying individual differences in hearing-aid outcome. Not only was there considerable uncertainty as to what we should measure, but there were no clear answers as to when we should obtain outcome measures. Following the pioneering work on hearing aid acclimatization by Gatehouse (1992) and Cox and Alexander (1992), a wide range of results had been obtained by various investigators with some suggesting significant longitudinal improvements in self-reported benefit or aided speech-recognition scores over the first 3 6 months following hearing aid delivery and others observing stable performance over time (see review by Turner, Humes, Bentler and Cox 1996).

3 Hearing-Aid Outcome Measures in Older Adults 267 Prior Outcomes Research at Indiana University What to Measure and When to Measure It With the foregoing in mind, we decided to temporarily set aside our ultimate objective of identifying the factors underlying individual differences in hearing aid outcome. Instead, we directed our efforts initially at both what should be measured to get a comprehensive assessment of hearing-aid outcome and when they should be measured (Humes 2001). There are various approaches one could take to answering these questions, but the approach pursued in our studies at Indiana University was to obtain a wide range of outcome measures from a large number of older hearing aid wearers fit with identical devices and to follow these individuals over a period of at least 12 months. In the first two studies of this type completed at Indiana University (Humes, Garner, Wilson and Barlow 2001; Humes, Humes and Wilson 2004), each employing a different hearing aid style [full concha in-the-ear (ITE) aids versus in-the-canal (ITC) aids] and circuitry (linear with output-limiting compression for the ITE aids and two-channel wide-dynamic-range-compression for the ITC aids), a total of twenty outcome measures were obtained from all study participants. These outcome measures are summarized in table 1. It was important to not only sample several potential domains of hearing-aid outcome, but also multiple measures within each domain. With so many outcome measures, and factor analyses identified as the tool of choice to examine redundancy among these measures, a large number of participants were required for this investigation. The initial study, which focused on what to measure, obtained measures from 173 older Table 1. List of the 20 outcome measures included in the initial studies of hearing-aid outcome at Indiana University. HDABI = Hearing Disability and Benefit Inventory (Gatehouse 1999); CST = Connected Speech Test (Cox, Alexander, Gilmore and Puskalich 1988); NST = Nonsense Syllable Test (Levitt and Resnick 1978). Other acronyms are defined in the text. Outcome Categor y Variable Description Hearing-aid Usage HDABI_use Mean frequency of hearing-aid usage, 14 situations GHABP_use Mean frequency of hearing-aid usage, 4 situations Daily use Mean hours per day hearing-aid usage from daily diary Subjective Benefit HAPI_spn Mean score for speech-in-noise HAPI subscale HAPI_spq Mean score for speech-in-quiet HAPI subscale HAPI_rd_cues Mean score for HAPI subscale, speech--reduced cues HAPI_nonspch Mean score for HAPI subscale, non-speech sounds HABI_ben Mean helpfulness of hearing aids, 14 listening situations GHABP_ben Mean helpfulness of hearing aids, 4 listening situations GHABP_perf Mean aided performance rating, 4 situations Satisfaction MarkeTrakIV Mean satisfaction rating on 28 items, 5-point scale GHABP_sat Mean satisfaction rating, 4 listening situations Speech Recognition NST_unaid Percentage correct on NST, 65 db SPL, +8 db SNR NST_aided Percentage correct on NST, 65 db SPL, +8 db SNR CST50_unaid Percentage correct on CST, 50 db SPL, in quiet CST65_unaid Percentage correct on CST, 65 db SPL, +8 db SNR CST80_unaid Percentage correct on CST, 80 db SPL, 0 db SNR CST50_aided Percentage correct on CST, 50 db SPL, in quiet CST65_aided Percentage correct on CST, 65 db SPL, +8 db SNR CST80_aided Percentage correct on CST, 80 db SPL, 0 db SNR

4 268 Hearing Care for Adults adults at hearing-aid fit, as well as at two-weeks and one-month post fit (Humes et al. 2001). These data suggested that the 20 outcome measures in table 1 could be reduced to three (Humes 2003), or at least that there were three dimensions or domains of outcome. In the end, whether one chooses to obtain only a single measure from each domain would depend on the reliability of the measures available in each domain. The three dimensions or domains of outcome identified were self-reported hearing-aid benefit and satisfaction combined, hereafter referred to as benefaction (courtesy of Mead Killion), speech-recognition performance (aided and unaided, quiet and in noise), and hearing-aid usage. The second IU study (Humes et al. 2004) involved 53 older ITC wearers followed for six months and the same findings were observed with regard to what to measure. The study sample was smaller because we approached this study with the knowledge gained from the initial larger study and were mainly interested in whether the same results would be obtained with different circuitry and a different style of hearing aid. The same three dimensions of outcome were identified as in the initial larger study: benefaction, speech recognition, and usage (Humes 2003). As to when to obtain these measures, we failed to find much evidence in support of acclimatization of hearing-aid benefit, whether measured via self report or objectively with speech-recognition scores. This was the case when 134 of the ITE participants returned for additional measurements at 6 and 12 months post-fit (Humes, Wilson, Barlow and Garner 2002; Humes, Wilson, Barlow, Garner and Amos 2002), as well as for a smaller sample (N = 34) of ITC wearers who completed the same outcome measures at one-month and six-months post-fit (Humes et al. 2004). In addition, we were able to document the lack of significant acclimatization effects in selfreport and objective measures of benefit for smaller groups of older adults wearing the ITE devices for either two years (N = 49; Humes, Wilson, Barlow and Garner 2002; Humes, Wilson, Barlow, Garner and Amos 2002) or three years (N = 9; Humes and Wilson 2003). Individual Differences in Hearing-Aid Outcome Having established three primary domains of hearing aid outcome from the 20 outcome measures employed in both studies, as well as when measures from each domain should be obtained, we were now ready to return to our primary goal: the identification of factors underlying individual differences in hearing aid outcome. In the initial study of outcome measures completed at Indiana University, 173 older adults not only completed the outcome measures at one-month post-fit, but they also had completed a large battery of tests prior to receiving amplification. These pre-fit measures included a wide array of audiological measures from simple pure-tone thresholds to auditory brainstem responses and measures of auditory processing. Non-auditory measures also were included, such as measures of manual dexterity, cognitive function, and locus of control. Details regarding the wide array of pre-fit measures included can be found in Humes (2002, 2003). The results of the analyses of individual differences in hearingaid outcome were presented in detail in Humes (2003) and will only be summarized here. Briefly, only 12% of the variance in the benefaction domain could be accounted for by pre-fit variables. Of the three significant pre-fit predictors identified, two were related to sound quality or loudness discomfort. For hearing-aid usage, 35% of the variance could be accounted for, but the best predictor of current hearing aid usage was prior hearing aid usage. Although informative, this predictor was not particularly helpful in determining factors associated with individual differences in usage for new hearing aid users. Finally, 55% of the variance in speech-recognition performance could be accounted for by pre-fit variables with the three key predictors being audibility (hearing loss and measured hearing-aid gain), age, and verbal IQ. In summary, of the three domains of hearing-aid outcome, pre-fit predictor variables could account for substantial amounts of variance only for the speech-recognition domain. For speech recognition, roughly twothirds of the variance that could be accounted for was accounted for by audibility. Still, age and verbal IQ accounted for sizable additional portions of the variance in speech recognition. This suggested that the audibility was the primary contributor, but not sole contributor, to individual differences in speech-recognition performance by older hearing aid wearers. One interpretation of this finding is that, once a group of older adults with impaired hearing has been formed based on similar audiograms, individual differences in performance among this group will be determined primarily by individual differences in age and verbal IQ. There were several likely reasons that audibility emerged as the primary predictor of speech-recognition performance in this study. One reason could have been the decision to include both aided and unaided

5 Hearing-Aid Outcome Measures in Older Adults 269 mea-sures of speech-recognition performance in the factor analysis of outcome measures. Clearly, one would expect unaided performance to be largely determined by hearing loss. Other possible reasons for the strong influence of audibility on the speech-recognition domain include the type of hearing aid employed (single-channel linear with output-limiting compression), the clinical gain-prescription formula employed (NAL-RP; Byrne, Parkinson and Newall 1990), limits in our ability to achieve the desired target gain in the higher frequencies with the technology employed, and the user s desire to wear the aid, at least for the first month, at a less than optimal gain setting (Humes, Barlow, Garner and Wilson 2000). As noted in Humes (2002), this combination of factors often resulted in the inaudibility of speech (and noise) stimuli above 3000 Hz, even when amplified. Ongoing research on hearing-aid outcome at Indiana University is attempting to address these audibility issues through use of multi-channel digital ITE hearing aids fit using the NAL-NL1 prescriptive procedure (Dillon 1999). The electroacoustic flexibility of the hearing aid allowed for better matches to the desired gain at all frequencies and the option for feedback management with these devices increased the likelihood of achieving the desired high-frequency gain in the wearer s ears. Recall, however, that much lower percentages of variance were accounted for in the benefaction and usage domains of hearing-aid outcome than in the speechrecognition domain. In these two self-report domains, our primary goal in the ongoing study was to expand the range of potential predictor variables. As the initial study of hearing aid outcome at Indiana University was underway and nearing completion, information about the correlation of personality with self-report measures of hearing-aid outcome was published (Cox, Alexander and Gray 1999). In addition, subsequent analysis of the data from the NIDCD/VA clinical trial of several types of single-channel hearing aids (Larson et al. 2000) also suggested that measures of sound quality and aided loudness might be predictors of hearing-aid satisfaction (Humes 2003). Finally, more recent work of Nabelek and colleagues on acceptable noise level suggested that this measure might be predictive of hearing aid usage (Nabelek, Freyaldenhoven, Tampas, Burchfield and Muenchen, 2006). In the ongoing study, we have added potential predictive measures not included previously, including measures of personality, emotional state and well being, aided loudness and sound quality, and acceptable noise level. We also eliminated those predictors that were not informative in the earlier studies, including most of the advanced audiological measures (otoacoustic emissions, auditory brainstem responses, and word-recognition roll-over), measures of manual dexterity, and measures of auditory processing. Finally, in the ongoing study of hearing-aid outcome at Indiana University, we have expanded the outcome measures in the area of satisfaction by adding the SADL to the other measures of satisfaction included Table 2. List of outcome measures included in the ongoing study of hearing-aid outcome at Indiana University. Outcome Categor y Variable Description Hearing-aid Usage GHABP_use Mean frequency of hearing-aid usage, 4-8 situations Daily use-a Mean hours per day of hearing-aid usage from diary Daily use-b Mean hours per day of hearing-aid ownership from diary Subjective Benefit HAPI_global Mean score for all HAPI items GHABP_ben Mean helpfulness of hearing aids, 4-8 listening situations Satisfaction MarkeTrakIVa Mean satisfaction rating for hearing-aid features MarkeTrakIVb Mean satisfaction rating for listening situations GHABP_sat Mean satisfaction rating, 4-8 listening situations SADL_global Mean SADL global score Speech Recognition CST65_aideda Percentage correct, CST, 65 db SPL, +8 db SNR, at fit CST65_aidedb Percentage correct, CST, 65 db SPL, +8 db SNR, 6 wks

6 270 Hearing Care for Adults previously (the MarkeTrak instrument and the satisfaction portion of the GHABP). We also reduced the number of speech-recognition measures obtained, since all of these measures were strongly correlated and associated with a single, largely independent, domain of outcome; one for which predictive factors could more easily be identified. In addition, we decided to focus on aided speech-recognition performance alone in these analyses, rather than including both unaided and aided measures as in previous analyses. In the remaining paragraphs of this paper, we report the findings to date from this ongoing investigation of individual differences in hearing aid outcome. Ongoing Outcomes Research at Indiana University In the ongoing study at Indiana University, complete data are currently available from 79 older adults fit bilaterally with ITE multi-band, multi-channel digital hearing aids. The 11 outcome measures completed at six-weeks post-fit are listed in table 2. An additional purpose for this study was the examination of differences in hearingaid outcome for hearing aids with directional versus omnidirectional microphones. Every other participant enrolled was assigned sequentially to the directional group and the other half of the participants were assigned to the omnidirectional group. Of all the outcome measures listed in table 2, thus far only one significant difference in hearing-aid outcome has been observed in the mean data for each of these microphone-technology groups. This difference was observed for the aided speech-recognition score obtained at six- weeks post-fit, with those having directional microphones scoring significantly higher than those having omnidirectional microphones. Since the overwhelming majority of outcome measures shown in table 2, however, exhibit no significant differences associated with microphone technology, these outcome measures have been pooled for all 79 subjects in subsequent analyses of hearing-aid outcome described below. Principal-components factor analysis (Gorsuch 1983) was then applied to the 11 hearing-aid outcome measures in table 2 obtained from the 79 older adults. Three factors were identified (using the criterion of eigenvalues greater than 1.0; Gorsuch 1983; Humes 2003). Observed communalities for the 11 outcome measures were all greater than 0.69 (9 of 11 >_ 0.80) and the total variance accounted for by these three factors was 83.8% (Gorsuch 1983). Both of these observations indicate that the factor solution provided a good fit to the data. The three factors identified were benefaction, usage, and aided speech recognition; a solution identical to that from the two prior studies of outcome at Indiana University that made use of similar outcome measures, but different technologies (Humes, 2003). Further, as in the prior studies, it was more appropriate to model the three domains of outcome as correlated domains (i.e., oblique rotation of components), with a correlation of r = 0.52 between the domains of benefaction and usage. Correlations between the other pairs of domains were negligible (r = 0.11 between the domains of speechrecognition performance and benefaction, and r = 0.01 between the domains of speech recognition performance and usage). The linkages among the three outcome domains are also similar to those observed in our earlier studies (Humes 2003). Individual Differences in Hearing-Aid Outcome Having established, once again, these three domains of hearing-aid outcome, we next examined the ability of various pre-fit measures to account for individual differences in each domain. Step-wise multiple linear regression was used to identify significant predictor variables for each outcome domain. Table 3 summarizes the results of these analyses for the 79 older adults who have completed the study to date. As shown in table 3, four predictor variables could account for 46% of the variance in the benefaction domain. This is a vast improvement over the 12% of the variance in this outcome domain that could be accounted for previously (Humes 2003). Note that two of the four predictor variables were measures of aided loudness (Profile of Aided Loudness, PAL; Mueller and Palmer 1998) and aided sound quality (Gabrielsson, Schenkman and Hagerman 1988; Narendran and Humes 2003). For the PAL, the loudness ratings for three of the four soft loudness items of this scale (indicated as PAL_soft in table 3) emerged as a predictor of benefaction. For measures of aided soundquality, the loudness related items of the sound-quality ratings (softness, nearness, and loudness) were separated from the other five dimensions of sound quality (clarity, fullness, brightness, spaciousness, and total impression) with two composite ratings resulting; one for loudness and the other for global sound quality. It is the global sound-quality measure for a moderate-intensity (70 db SPL) speech stimulus that emerged as a predictor of benefaction in the regression analyses summarized in table 3. The other two predictors identified were

7 Hearing-Aid Outcome Measures in Older Adults 271 Table 3. Summary of results of regression analyses for ongoing study of hearing-aid outcome at Indiana University (N = 79). All regression equations were significant at p <.001. Outcome Domain % Variance Predictor Beta Weight Benefaction 18.4 PAL_soft ECHO JSQ_global HA problems Total = 46.1 Usage 20.4 HA problems PAL_soft PriorDailyUse ECHO 0.21 Total = 41.7 Aided Speech Recognition 42.4 CST unaided WAISverbal Age Total = 48.8 a measure of hearing aid expectations (ECHO; Cox and Alexander 2000) and a checklist of hearing aid problems experienced by the wearer during a hearing-aid followup session completed at two-weeks post-fit. In general, the regression results are consistent with the following interpretation: lower aided loudness for soft sounds; fewer observed or reported hearing aid problems; higher aided sound quality ratings; and higher expectations yield greater self-reported benefaction. With regard to hearing-aid usage, table 3 indicates that four variables accounted for 42% of the variance in the usage domain. Three of these variables, moreover, were among the four predictor variables identified for the benefaction domain and probably reflect the partial overlap of these two domains (i.e., correlation between benefaction and usage of r = 0.52). The other predictor, consistent with similar analyses of the data from the prior studies at Indiana University (Humes 2003), was the amount of prior daily hearing aid use. For the usage domain, the interpretation of the regression analysis is that lower aided loudness for soft sounds, fewer hearing aid problems, higher expectations, and greater amounts of prior hearing aid use lead to greater current hearing aid usage. As noted, three of the four predictive measures for benefaction and usage are the same. As a result, these measures probably deserve some more discussion here. One of these measures, however, the Expected Consequences of Hearing aid Ownership (ECHO; Cox and Alexander 2000), has been described in detail previously and does not require additional discussion. In these analyses, we only used the global ECHO score, rather than subscale scores. Another measure, the PAL, had been described previously (Mueller and Palmer 1998), but we modified its use slightly in this study. In particular, various subscales were formed here based on an orthogonal principal-components analysis of wearer s responses to the loudness portion of the PAL items, rather than using the item groupings proposed by the scale developers. One subgroup of items, however, represented three of the four soft loudness items (items 3, 4 and 8) grouped in the original PAL. The factor scores associated with the loudness ratings of these three items of the PAL represent the predictor variable labeled PAL_soft in table 3. Note, however, that these are only the loudness ratings for these items of the PAL and not the ratings of the wearer s satisfaction with the observed loudness; ratings that also are included in the PAL and were obtained from the wearer. We did not feel that it was appropriate to use one measure of satisfaction with amplification to predict another measure of satisfaction (or, more appropriately, benefaction). Finally, with regard to the hearing-aid problems checklist, this measure appears in table 4 and was developed specifically for use in this study. Basically, the hearing-aid wearer is asked by the audiologist at this two-week follow-up session if he or she has experienced any difficulties with regard to the topics listed in the left column of table 4 (insertion/removal of hearing aids, comfort of

8 272 Hearing Care for Adults aids, etc.). If no problems are noted, a score of 0 is entered. If problems are noted, they are recorded for the ear involved and each receives a score of 1 (i.e., a score of 2 is recorded, if problems are reported in both ears for a given item). Wearers also are queried with regard to any additional problems not covered in the left column, and these are tabulated at the bottom of the checklist. A total score is then computed by simply summing the scores on the checklist. In general, those who indicated a greater number of problems at the two-week follow-up had lower benefaction and usage at six-weeks post-fit when the outcome measures were completed. Returning to the outcome domain of aided speechrecognition performance, as shown in table 3, 49% of the total variance could be accounted for by three predictor variables: unaided speech-recognition performance (for the same test conditions as aided performance); verbal cognitive function; and age. The majority of the variance (42% of the 49% explained variance) was explained by unaided speech-recognition performance, with verbal cognitive function and age each accounting for an additional 3 4% of the total variance in aided speech-recognition performance. Regression analysis of unaided speech recognition performance for these same 79 participants indicated that 45% of the variance in unaided performance could be explained by average high-frequency (1000, 2000 and 4000 Hz) hearing loss, with an additional 8% associated with verbal cognitive function. In general, the milder the average high-frequency hearing loss and the higher the verbal cognitive function, the higher the unaided speech-recognition score. In turn, younger participants with higher unaided speech-recognition scores and higher verbal cognitive function yielded higher aided speech-recognition scores. By the time this ongoing study is complete, it is most likely that an even greater percentage of variance will be accounted for in the aided speech-recognition domain. To explain, recall that group differences in aided speech-recognition performance at six-weeks post-fit were observed for the subgroup receiving directional microphones. Electroacoustic measures of microphone directivity were not available for each participant at the time of these interim analyses, but will be available by study completion. Given the observed group differences in aided speech-recognition performance due to microphone technology, it is likely that this factor will account for additional variance in the final analyses of individual differences in this domain as well. Finally, table 3 only identifies the predictors in each outcome domain that could account for significant portions of the variance in each domain. It would be informative to know the factors that were included in the regression analyses, but did not account for any of the individual variations in outcome. The set of possible predictor variables included in these interim analyses were age, average high-frequency hearing loss, loudness discomfort level at 2000 Hz, unaided speech-recognition scores, unaided acceptable noise level (which was strongly correlated with aided acceptable noise level), global sound quality for speech, global sound quality for music, ECHO, six factor scores from the Communication Profile for the Hearing Impaired (CPHI; Demorest and Erdman 1987), three factor scores from the PAL, one emotional well-being factor score, four personality factor scores from the Myers-Briggs Type Indicator, five cognitive factor scores from the third edition of the Wechsler Adult Intelligence Scale (Wechsler 1997), and the hearing-aid problems checklist (table 4). Sometimes, the variables that do not enter the regression equation are just as informative as those that do. Conclusions The interim conclusions that can be drawn from this ongoing work on hearing-aid outcome measures in older adults are as follows. First, this is the third independent confirmation, each with a different hearing-aid technology, of the existence of three domains of hearing-aid outcome in older adults: (1) benefaction; (2) usage; and (3) speech recognition. Second, while there are strong linkages between the two self-report domains of benefaction and usage, speech recognition is a largely independent aspect of outcome. Third, progress continues to be made in the identification of variables that may explain or predict individual differences in outcome in each of these three domains. These interim analyses from the ongoing study of hearing-aid outcome at Indiana University suggest that 40 50% of the variance can be accounted for in each outcome domain by various predictor variables (table 3). Assuming that many of these outcome measures have test-retest correlations of approximately 0.8 (Humes, Halling and Coughlin 1996), an upper bound for the total variance that could be explained by other variables is about 64% (with the balance considered to be unexplained or error variance). In that context, accounting for 40 50% of the total variance amounts to accounting for 62 78% of the systematic or explainable variance.

9 Hearing-Aid Outcome Measures in Older Adults 273 Table 4. The hearing-aid problems checklist used in ongoing study. Hearing Aid Problems Checklist Check-mark that item was reviewed. Check-mark if problems/no problems. INSERTION AND REMOVAL OF AIDS COMFORT OF AIDS TURNING AIDS ON/OFF ADJUSTING VOLUME USING TELEPHONE BUTTON/T-COIL CHANGING BATTERIES CLEANING THE HEARING AIDS LOUDNESS OF LOUD SOUNDS Did subject report problems with items other than those listed above? NO YES IF YES, HOW MANY: FEW (1-3) SEVERAL (4 or more)

10 274 Hearing Care for Adults Finally, any factor analysis of outcome measures, applied to identify separate domains of outcome, is necessarily limited by the outcome measures included in the analysis. Although a large set of outcome measures has been included in all of the outcome studies reviewed here, other relevant domains could be sampled in future analyses as additional tools are developed to tap these domains. For example, the speech, spatial and qualities of hearing scale, SSQ (Gatehouse and Noble 2004), which taps a wider range of auditory perception than other scales used to date, could prove useful in future outcomes research and could lead to the identification of additional important domains of hearing-aid outcome. Acknowledgements The author would like to thank the following clinical faculty collaborators on the current and prior studies of hearing aid outcome at Indiana University: Dana (Wilson) Kinney, Elizabeth Thompson, Nathan Amos, Amy Arthur, Nancy Barlow, Gretchen Burk, Carolyn Garner, and Lisa Goerner. In addition, Lauren Strauser assisted with the development of the numerous response forms and data-entry forms used in the on-going investigation of hearing-aid outcome. This research on individual differences in hearing-aid outcome has benefited from several discussions of individual-differences research over the years with colleagues at Indiana University, especially Chuck Watson and Gary Kidd. Finally, this work was supported, in part, by a research grant from the National Institute on Aging (R01 AG08293). References Byrne, D., Parkinson, A., and Newall, P Hearing aid gain and frequency response requirements for the severely/profoundly hearing impaired. Ear and Hearing 11: Cox, R.M., and Alexander, G.C Hearing aid benefit in everyday environments. Ear and Hearing 12: Cox, R., and Alexander, G Maturation of hearing aid benefit: Objective and subjective measurements. Ear and Hearing 13: Cox, R.M., and Alexander, G.C The abbreviated profile of hearing aid benefit. Ear and Hearing 16: Cox, R.M., and Alexander, G.C Measuring satisfaction with amplification in daily life: The SADL scale. Ear and Hearing 20: Cox, R.M., and Alexander, G.C Expectations about hearing aids and their relationship to fitting outcome. Journal of the American Academy of Audiology 11: Cox, R.M., and Alexander, G.C Validation of the SADL questionnaire, Ear and Hearing 22: Cox, R.M., and Rivera, I.M Predictability and reliability of hearing aid benefit measured using the PHAB. Journal of the American Academy of Audiology 3: Cox, R.M., Alexander, G.C., and Gray, G.A Personality and the subjective assessment of hearing aids. Journal of the American Academy of Audiology 10 (1):1 13. Cox, R.M., Gilmore, C.G., and Alexander, G.C Comparison of two questionnaires for patient-assessed hearing aid benefit. Journal of the American Academy of Audiology 2: Cox, R., Alexander, G., Gilmore, C., and Puskalich, K.M Use of the Connected Speech Test (CST) with hearing-impaired listeners. Ear and Hearing 9: Demorest, M., and Erdman, S Development of the communication profile for the hearing impaired. Journal of Speech and Hearing Disorders 52: Dillon, H NAL-NL1: A new prescriptive fitting procedure for non-linear hearing aids. The Hearing Journal 52(4): Dillon, H Shortened Hearing Aid Performance Inventory for the Elderly (SHAPIE): A statistical approach. Australian Journal of Audiology 16: Dillon, H., James, A., and Ginis, J The Client Oriented Scale of Improvement (COSI) and its relationship to several other measures of benefit and satisfaction provided by hearing aids. Journal of the American Academy of Audiology 8: Gabrielsson, A., Schenkman, B.N., and Hagerman, B The effects of different frequency responses on sound quality judgments and speech intelligibility. Journal of Speech and Hearing Research 31: Gatehouse, S The time course and magnitude of perceptual acclimatization to frequency responses: Evidence from monaural fitting of hearing aids. Journal of the Acoustical Society of America 92: Gatehouse, S Components and determinants of hearing aid benefit. Ear and Hearing 15: Gatehouse, S Glasgow Hearing Aid Benefit Profile: Derivation and validation of a client-centered outcome measure for hearing aid services. Journal of the American Academy of Audiology 10:

11 Hearing-Aid Outcome Measures in Older Adults 275 Gatehouse, S. and Noble, W The speech, spatial, and qualities of hearing scale (SSQ). International Journal of Audiology 43: Gorsuch, R.L Factor analysis, 2 nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates. Hosford-Dunn, H., and Halpern, J Clinical application of the Satisfaction with Amplification in Daily Life scale in private practice. I: Statistical, content and factorial validity. Journal of the American Academy of Audiology 11: Hosford-Dunn, H.L., and Halpern, J Validation of the SADL. II: Influence of patient and technological variables. Journal of the American Academy of Audiology 12: Humes, L.E Issues in evaluating the effectiveness of hearing aids in the elderly: What to measure and when. Seminars in Hearing 22: Humes, L.E Factors underlying the speechrecgnition performance of elderly hearing-aid wearers. Journal of the Acoustical Society of America 112: Humes, L.E Modeling and predicting hearing aid outcome. Trends in Amplification 7: Humes, L.E., and Wilson, D.L An examination of changes in hearing-aid performance and benefit in the elderly over a 3-year period of hearing-aid use. Journal of Speech, Language, and Hearing Research 46: Humes, L.E., Halling, D., and Coughlin, M Reliability and stability of various hearing aid outcome measures in a group of elderly hearing aid wearers. Journal of Speech and Hearing Research 39: Humes, L.E., Humes, L.E., and Wilson, D.L A comparison of single-channel linear amplification and two-channel wide-dynamic-range-compression amplification by means of an independent-group design. American Journal of Audiology 13: Humes, L.E., Barlow, N.N., Garner, C.B., and Wilson, D.L Prescribed clinician-fit versus as-worn coupler gain in a group of elderly hearing aid wearers. Journal of Speech, Language, and Hearing Research 43: Humes, L.E., Wilson, D.L., Barlow, N.N., and Garner, C.B Changes in hearing-aid benefit following one or two years of hearing-aid use by the elderly. Journal of Speech, Language, and Hearing Research 45: Humes, L.E., Garner, C.B., Wilson, D.L., and Barlow, N.N Hearing-aid outcome measures following one month of hearing aid use by the elderly. Journal of Speech, Language, and Hearing Research 44: Humes, L.E., Wilson, D.L., Barlow, N.N., Garner, C.B., and Amos, N.E Longitudinal changes in hearing-aid satisfaction and usage in the elderly over a period of one or two years after hearing-aid delivery. Ear and Hearing 23: Humes, L.E., Wilson, D.L., Humes, L.E., Barlow, N.N., Garner, C.B., and Amos, N.E A comparison of two measures of hearing-aid satisfaction in a group of elderly hearing aid wearers. Ear and Hearing 23: Kochkin, S MarketTrak III: Higher hearing aid sales don t signal better market penetration. Hearing Journal 45(7): Kochkin, S. 1993a. MarkeTrak III: Why 20 million in U.S. don t use hearing aids for their hearing loss. Part I. Hearing Journal 46(1): Kochkin, S. 1993b. MarkeTrak III: Why 20 million in U.S. don t use hearing aids for their hearing loss. Part II. Hearing Journal 46(2): Kochkin, S. 1993c. MarkeTrak III: Why 20 million in U.S. don t use hearing aids for their hearing loss. Part III. Hearing Journal 46(4): Kochkin, S Subjective measures of satisfaction and benefit: establishing norms. Seminars in Hearing 18 (1): Kochkin, S MarkeTrak V: Consumer satisfaction revisited. Hearing Journal 53(1): Larson, V.D., Williams, D.W., Henderson, W.G., Luethke, L.E., Beck, L.D., et al Efficacy of 3 commonly used hearing aid circuits: A crossover trial. Journal of the American Medical Association 284: Levitt, H., and Resnick, S.B Speech reception by the hearing impaired: Methods of testing and development of materials. Scandinavian Audiology Suppl. 6: Malinoff, R., and Weinstein, B Measurement of hearing aid benefit in the elderly. Ear and Hearing 10: Mueller, H.G., and Palmer, C.V The profile of aided loudness: A new PAL for 98. Hearing Journal 51(1): Nabelek, A.K., Freyaldenhoven, M.C., Tampas, J.W., Burchfield, S.B., and Muenchen, R.A Acceptable noise level as a predictor of hearing aid use. Journal of the American Academy of Audiology 17: Narendran, M.M., and Humes, L.E Reliability and validity of judgments of sound quality in elderly hearing aid wearers. Ear and Hearing 24:4 11.

12 276 Hearing Care for Adults Newman, C., Jacobson, G., Hug, G., Weinstein, B., and Malinoff, R Practical method for quantifying hearing aid benefit in older adults. Journal of the American Academy of Audiology 2: Schum, D Responses of elderly hearing aid users on the Hearing Aid Performance Inventory. Journal of the American Academy of Audiology 3: Skafte, M.D The 1999 hearing instrument market the dispensers perspective. Hearing Review 7(6):8 40. Turner, C.W., Humes, L.E., Bentler, R.A., and Cox, R.M A review of past research on changes in hearing aid benefit over time. Ear and Hearing 17:14S 25S. Ventry, I., and Weinstein, B The hearing handicap inventory for the elderly: A new tool. Ear and Hearing 3: Walden, B.E., Demorest, M.E. and Hepler, E.L Self-report approach to assessing benefit derived from amplification. Journal of Speech and Hearing Research 27: Wechsler, D Wechsler Adult Intelligence Scale-III. San Antonio, TX: The Psychological Corporation. Wong, L.L.N., Hickson, L., and McPherson, B Hearing aid satisfaction: What does research from the past 20 years say? Trends in Amplification 7: