THE RELIABILITY AND VALIDITY OF THE COMFORT LEVEL METHOD OF SETTING HEARING AID GAIN
Brian E. Walden, Gerald I. Schuchman, and Roy K. Sedge Walter Reed Army Medical Center, Washington, D.C.
The comfort level method (Carhart, 1946) probably is the most widely used procedure for setting the acoustic gain of hearing aids. A series of experiments were conducted to determine the test-retest reliability of the comfort level method and the relationship between the comfort settings established in a clinical test suite and the comfort settings utilized in more realistic daily listening situations. Adults with bilateral sensorineural hearing impairments were subjects. The results suggest that the comfort level method has good test-retest reliability for most clinical purposes. Further, clinically established comfort settings may accurately represent typical daily-use settings if the input level used to establish the comfort settings in the clinical environment is 70 dB SPL.
A variety of evaluation procedures has been developed to assist the clinician in selecting an appropriate hearing aid for a given patient. A prerequisite to each procedure is that the volume control dial of the instrument be adjusted prior to testing. T h i s adjustment can affect the outcome of tile evaluation since changes in the volume control setting can influence the level of distortion, signal-to-noise ratio, and frequency response, as well as the acoustic gain being delivered by the hearing aid (Kasten, Lotterman, and Revoile, 1967; L o t t e r m a n and Kasten, 1967; Bode and Kasten, 1971). Several methods of setting tile volume control dial of a hearing aid have been suggested. However, the most widely used procedure is probably the comfort level method (CLM) suggested by Carhart (1946). According to this method, the volume control dial is adjusted so that a standard input signal is amplified to the patient's most comfortable listening level. Any procedure for establishing volume control settings must be reliable and must yield settings which are comparable to those used by the patient in more typical listening situations. Few investigations of the reliability of the C L M appear in the literature. Davis and Ross (1945) and McConnell, Silber, and McDonald (1960) questioned the reliability of the CLM. Carhart (1946), however, concluded that the reliability of tile CLM is adequate for clinical purposes. T h e purpose of the series of experiments reported here was to study 455
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
456
JOURNAL OF SPEECH AND HEARING DISORDERS
XLII 455-461
1977
further the test-retest reliability of the CLM and to determine the relationship between the gain provided at clinically established comfort settings and that used in typical listening situations. PROCEDURES
T h e subjects were 40 adults with bilateral sensorineural hearing impairments. They ranged in age from 20 to 70 years (mean: 46.9 years), had unaided speech reception thresholds between 22 and 68 dB H T L (mean: 39.1 dB H T L ) , and had word-recognition ability (PB-50) of between 60 and 94% (mean: 80.7%). Half of the subjects participated in an evaluation of the testretest reliability of the comfort level method and the other half were involved in an evaluation of the validity of the method, that is, the relationship between clinically determined comfort level settings and daily-use settings. All testing was conducted in a large sound-treated test suite. Reliability Experiment
T e n of the 20 subjects were experienced hearing aid users, and 10 were not. T h r e e high-gain hearing aids, which differed considerably in frequency response, were used, and each of the 20 subjects was tested with all three aids. T h e three hearing aids were not necessarily the most appropriate to the subject's hearing impairment. It was assumed that the test-retest reliability demonstrated by a subject with these instruments would be no worse than the reliability he would show with instruments carefully selected for his particular impairment. A recording of connected speech was used as the input to the aids during all comfort level adjustments. It was presented at 60-dB SPL and wa's measured at the location of the subject's head without the subject being present. T h e examiner adjusted the volume control of the aid to the subject's comfort setting. This was determined by bracketing the volume control dialsetting at the level at which the speech was judged to be most comfortably loud by the subject. Beginning with the volume control set at its lowest setting, the dial was adyanced slowly until the subject judged the speech to be most comfortable. T h e n the volume control was further advanced until the subject judged the speech to be too loud. Next, the volume was decreased until the output was judged to be too soft. Finally, the output was increased again until the final comfort setting was achieved. Speech reception thresholds were obtained and measures of acoustic gain were made at each of two comfort settings with each hearing aid. T h e two comfort level adjustments with a particular aid were separated by approximately one hour, and the order of testing of the three aids was counterbalanced across subjects. T h e T i l l m a n and Carhart (1966) threshold procedure was used to obtain aided speech reception thresholds, using recorded randomizations of the 22 most homogeneous CID spondaic words (Bowling and El-
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
WALDEN, SCHUCHMAN, SEDGE: Comfort Level Method 457 pern, 1960). T h e hearing aid measurement system used to make the acoustic gain measurements was that described by Kasten et al. (1967). Acoustic gain was taken as the average of 500, 1000, and 2000 Hz, using a 60-dB SPL input. Both the connected speech and the spondees were presented in the sound field to a subject seated in the test suite at a distance of six feet and an angle of 45 degrees from the loudspeaker. T h e nontest ear was covered by a highattenuation acoustic muff. For the experienced hearing aid users, the test ear was the one in which they typically wore their own aid. For the inexperienced subjects, the test ear was the one for which they were being considered for amplification.
Validity Experiment T h e 20 subjects were new hearing aid users. T h e instruments were the subjects' own aids, issued to them at the time of the initial hearing aid evaluation. T h e hearing aids represented a wide range of styles and electroacoustic characteristics. T h e relationship between clinically established comfort level settings and typical daily-use settings was determined by making five acoustic gain measurements over an ll-day period. The first measurement evaluated a comfort level adjustment made in the test suite during the initial hearing aid evaluation. This setting was determined by the same procedures used in the reliability experiment, except that the unaided ear was unoccluded. The remaining four measurements were of volume control settings established by the subjects as they wore their aids. All subjects attended an inpatient aural rehabilitation program, eight hours per day for two weeks, following their hearing aid evaluation. T h e environment was similar to that of a classroom or business office. After one day of hearing aid use, each subject was interrupted without warning at a time during the day which was selected at random. His aid was removed and, without changing the volume control setting, the acoustic gain was measured. Similar measurements were made on the fourth, seventh, and tenth day of hearing aid use. RESU LTS
Reliability Experiment The data from the reliability experiment consisted of an aided speech reception threshold and an acoustic gain measure for two independent comfort settings, with each of three aids, by experienced and inexperienced hearing aid users. An analysis of variance did not reveal significant differences between the experienced and inexperienced hearing aid users, or among the three aids (a > 0.10). Therefore, the data were pooled across groups and aids. T h e pooled data are summarized in Table 1 and suggest relatively good test-retest reliability, both for the aided speech reception thresholds and for
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
458
JOURNAL OF SPEECH AND HEARING DISORDERS
XLII
455-461
1977
TABLE I. Mean speech reception thresholds and m e a n acoustic gain measurements for the test and retest conditions. Also shown are standard deviations a n d correlations between the test a n d retest conditions for each measure.
Test SRT Mean SD Gain Mean SD
Retest
Correlation Coefficient
12.9 6.9
12.0 7.5
0.78
30.7 8.3
30.4 7.6
0.67
the acoustic gain measures. Inspection of the data of individual subjects revealed that the first and second aided speech reception thresholds were within 3 dB for half of the 60 comparisons (that is, 20 subjects • 3 aids). Ninety percent of the test-retest comparisons were within 8 dB. T h e first and second gain measures were within 4 dB for half of the comparisons and within 8.5 dB for 90%.
Validity Experiment T h e data from the validity experiment are presented in Table 2. Shown is TABLE 2. Mean acoustic gain and standard deviations for the test-suite and the daily-use conditions. i
Mean SD
Test Suite
1
24.9 11.1
15.6 11.3
Daily Use 2 3 15.8 12.4
16.7 11.7
4 16.6 11.5
the mean acoustic gain for the initial comfort setting in the test suite and the four daily-use settings. An analysis of variance and subsequent probing procedure revealed that the mean acoustic gain required to achieve a comfort setting in the test suite was significantly different (a < 0.01) from each of the subsequent daily-use settings. Further, there were no significant differences among the acoustic gain measurements for the four daily-use settings (c~ >
0.20). Inspection of the data for individual subjects revealed that there were differences among the four daily-use measurements, ranging from 3 to 16 dB (mean: 7.8 dB) across the 20 subjects. However, there was no consistent trend for either increasing or decreasing gain as these new hearing aid users acquired experience with their aids.
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
WALDEN, SCHUCHMAN, SEDGE: Comfort Level Method
459
T h e relatively large discrepancy between the test suite comfort setting and the daily-use settings may be because hearing aid users tend to adjust the gain of their instruments in response to varying levels of acoustic input. As the overall acoustic level in the environment varies over a period of time, the listener will adjust the gain of his instrument to maintain a comfortable loudness level. T h e difference between the clinical and daily-use comfort settings may reflect a comparable difference in the input levels used to establish those settings. To explore this possibility, two follow-up experiments were accomplished with 16 new hearing aid users who were drawn from the same clinic population as the original sample. Again, the hearing aids were the subjects' own instruments. In the first follow-up experiment, connected discourse was presented to six listeners, and the hearing aids were adjusted to comfort settings. However, the procedure was repeated for 50, 60, 70, and 80-dB SPL input levels, presented in counterbalanced order. T h e results are presented in Figure 1. T h e relative gain is shown at a comfort setting for each of these m
Z
Brian E. Walden, Gerald I. Schuchman, and Roy K. Sedge Walter Reed Army Medical Center, Washington, D.C.
The comfort level method (Carhart, 1946) probably is the most widely used procedure for setting the acoustic gain of hearing aids. A series of experiments were conducted to determine the test-retest reliability of the comfort level method and the relationship between the comfort settings established in a clinical test suite and the comfort settings utilized in more realistic daily listening situations. Adults with bilateral sensorineural hearing impairments were subjects. The results suggest that the comfort level method has good test-retest reliability for most clinical purposes. Further, clinically established comfort settings may accurately represent typical daily-use settings if the input level used to establish the comfort settings in the clinical environment is 70 dB SPL.
A variety of evaluation procedures has been developed to assist the clinician in selecting an appropriate hearing aid for a given patient. A prerequisite to each procedure is that the volume control dial of the instrument be adjusted prior to testing. T h i s adjustment can affect the outcome of tile evaluation since changes in the volume control setting can influence the level of distortion, signal-to-noise ratio, and frequency response, as well as the acoustic gain being delivered by the hearing aid (Kasten, Lotterman, and Revoile, 1967; L o t t e r m a n and Kasten, 1967; Bode and Kasten, 1971). Several methods of setting tile volume control dial of a hearing aid have been suggested. However, the most widely used procedure is probably the comfort level method (CLM) suggested by Carhart (1946). According to this method, the volume control dial is adjusted so that a standard input signal is amplified to the patient's most comfortable listening level. Any procedure for establishing volume control settings must be reliable and must yield settings which are comparable to those used by the patient in more typical listening situations. Few investigations of the reliability of the C L M appear in the literature. Davis and Ross (1945) and McConnell, Silber, and McDonald (1960) questioned the reliability of the CLM. Carhart (1946), however, concluded that the reliability of tile CLM is adequate for clinical purposes. T h e purpose of the series of experiments reported here was to study 455
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
456
JOURNAL OF SPEECH AND HEARING DISORDERS
XLII 455-461
1977
further the test-retest reliability of the CLM and to determine the relationship between the gain provided at clinically established comfort settings and that used in typical listening situations. PROCEDURES
T h e subjects were 40 adults with bilateral sensorineural hearing impairments. They ranged in age from 20 to 70 years (mean: 46.9 years), had unaided speech reception thresholds between 22 and 68 dB H T L (mean: 39.1 dB H T L ) , and had word-recognition ability (PB-50) of between 60 and 94% (mean: 80.7%). Half of the subjects participated in an evaluation of the testretest reliability of the comfort level method and the other half were involved in an evaluation of the validity of the method, that is, the relationship between clinically determined comfort level settings and daily-use settings. All testing was conducted in a large sound-treated test suite. Reliability Experiment
T e n of the 20 subjects were experienced hearing aid users, and 10 were not. T h r e e high-gain hearing aids, which differed considerably in frequency response, were used, and each of the 20 subjects was tested with all three aids. T h e three hearing aids were not necessarily the most appropriate to the subject's hearing impairment. It was assumed that the test-retest reliability demonstrated by a subject with these instruments would be no worse than the reliability he would show with instruments carefully selected for his particular impairment. A recording of connected speech was used as the input to the aids during all comfort level adjustments. It was presented at 60-dB SPL and wa's measured at the location of the subject's head without the subject being present. T h e examiner adjusted the volume control of the aid to the subject's comfort setting. This was determined by bracketing the volume control dialsetting at the level at which the speech was judged to be most comfortably loud by the subject. Beginning with the volume control set at its lowest setting, the dial was adyanced slowly until the subject judged the speech to be most comfortable. T h e n the volume control was further advanced until the subject judged the speech to be too loud. Next, the volume was decreased until the output was judged to be too soft. Finally, the output was increased again until the final comfort setting was achieved. Speech reception thresholds were obtained and measures of acoustic gain were made at each of two comfort settings with each hearing aid. T h e two comfort level adjustments with a particular aid were separated by approximately one hour, and the order of testing of the three aids was counterbalanced across subjects. T h e T i l l m a n and Carhart (1966) threshold procedure was used to obtain aided speech reception thresholds, using recorded randomizations of the 22 most homogeneous CID spondaic words (Bowling and El-
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
WALDEN, SCHUCHMAN, SEDGE: Comfort Level Method 457 pern, 1960). T h e hearing aid measurement system used to make the acoustic gain measurements was that described by Kasten et al. (1967). Acoustic gain was taken as the average of 500, 1000, and 2000 Hz, using a 60-dB SPL input. Both the connected speech and the spondees were presented in the sound field to a subject seated in the test suite at a distance of six feet and an angle of 45 degrees from the loudspeaker. T h e nontest ear was covered by a highattenuation acoustic muff. For the experienced hearing aid users, the test ear was the one in which they typically wore their own aid. For the inexperienced subjects, the test ear was the one for which they were being considered for amplification.
Validity Experiment T h e 20 subjects were new hearing aid users. T h e instruments were the subjects' own aids, issued to them at the time of the initial hearing aid evaluation. T h e hearing aids represented a wide range of styles and electroacoustic characteristics. T h e relationship between clinically established comfort level settings and typical daily-use settings was determined by making five acoustic gain measurements over an ll-day period. The first measurement evaluated a comfort level adjustment made in the test suite during the initial hearing aid evaluation. This setting was determined by the same procedures used in the reliability experiment, except that the unaided ear was unoccluded. The remaining four measurements were of volume control settings established by the subjects as they wore their aids. All subjects attended an inpatient aural rehabilitation program, eight hours per day for two weeks, following their hearing aid evaluation. T h e environment was similar to that of a classroom or business office. After one day of hearing aid use, each subject was interrupted without warning at a time during the day which was selected at random. His aid was removed and, without changing the volume control setting, the acoustic gain was measured. Similar measurements were made on the fourth, seventh, and tenth day of hearing aid use. RESU LTS
Reliability Experiment The data from the reliability experiment consisted of an aided speech reception threshold and an acoustic gain measure for two independent comfort settings, with each of three aids, by experienced and inexperienced hearing aid users. An analysis of variance did not reveal significant differences between the experienced and inexperienced hearing aid users, or among the three aids (a > 0.10). Therefore, the data were pooled across groups and aids. T h e pooled data are summarized in Table 1 and suggest relatively good test-retest reliability, both for the aided speech reception thresholds and for
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
458
JOURNAL OF SPEECH AND HEARING DISORDERS
XLII
455-461
1977
TABLE I. Mean speech reception thresholds and m e a n acoustic gain measurements for the test and retest conditions. Also shown are standard deviations a n d correlations between the test a n d retest conditions for each measure.
Test SRT Mean SD Gain Mean SD
Retest
Correlation Coefficient
12.9 6.9
12.0 7.5
0.78
30.7 8.3
30.4 7.6
0.67
the acoustic gain measures. Inspection of the data of individual subjects revealed that the first and second aided speech reception thresholds were within 3 dB for half of the 60 comparisons (that is, 20 subjects • 3 aids). Ninety percent of the test-retest comparisons were within 8 dB. T h e first and second gain measures were within 4 dB for half of the comparisons and within 8.5 dB for 90%.
Validity Experiment T h e data from the validity experiment are presented in Table 2. Shown is TABLE 2. Mean acoustic gain and standard deviations for the test-suite and the daily-use conditions. i
Mean SD
Test Suite
1
24.9 11.1
15.6 11.3
Daily Use 2 3 15.8 12.4
16.7 11.7
4 16.6 11.5
the mean acoustic gain for the initial comfort setting in the test suite and the four daily-use settings. An analysis of variance and subsequent probing procedure revealed that the mean acoustic gain required to achieve a comfort setting in the test suite was significantly different (a < 0.01) from each of the subsequent daily-use settings. Further, there were no significant differences among the acoustic gain measurements for the four daily-use settings (c~ >
0.20). Inspection of the data for individual subjects revealed that there were differences among the four daily-use measurements, ranging from 3 to 16 dB (mean: 7.8 dB) across the 20 subjects. However, there was no consistent trend for either increasing or decreasing gain as these new hearing aid users acquired experience with their aids.
Downloaded From: https://jshd.pubs.asha.org/ by a University of Texas, Austin User on 03/27/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx
WALDEN, SCHUCHMAN, SEDGE: Comfort Level Method
459
T h e relatively large discrepancy between the test suite comfort setting and the daily-use settings may be because hearing aid users tend to adjust the gain of their instruments in response to varying levels of acoustic input. As the overall acoustic level in the environment varies over a period of time, the listener will adjust the gain of his instrument to maintain a comfortable loudness level. T h e difference between the clinical and daily-use comfort settings may reflect a comparable difference in the input levels used to establish those settings. To explore this possibility, two follow-up experiments were accomplished with 16 new hearing aid users who were drawn from the same clinic population as the original sample. Again, the hearing aids were the subjects' own instruments. In the first follow-up experiment, connected discourse was presented to six listeners, and the hearing aids were adjusted to comfort settings. However, the procedure was repeated for 50, 60, 70, and 80-dB SPL input levels, presented in counterbalanced order. T h e results are presented in Figure 1. T h e relative gain is shown at a comfort setting for each of these m
Z
