NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Feltner C, Wallace IF, Kistler CE, et al. Screening for Hearing Loss in Older Adults: An Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2021 Mar. (Evidence Synthesis, No. 200.)
Screening for Hearing Loss in Older Adults: An Evidence Review for the U.S. Preventive Services Task Force [Internet].
Show detailsLiterature Search
This review identified 4,681 unique records and assessed 281 full-text articles for eligibility (Figure 2). The review excluded 236 studies for various reasons, detailed in Appendix C, and included 41 unique studies (described in 45 publications). Of the included studies, one RCT reported some eligible outcomes for KQ 1, 34 studies (described in 35 publications) evaluated the accuracy of one or more screening tests for hearing loss (KQ 2), and no studies met eligibility criteria for KQ 3 (harms of screening). Six RCTs (described in 8 publications) addressed the benefits (KQ 4) of amplification compared with no amplification for treatment of screen-detected hearing loss, and no studies assessed harms of amplification (KQ 5). Details of quality assessments of included studies and studies excluded because of poor quality are in Appendix D Tables 1-16.
Figure 2
Summary of Evidence Search and Selection.
Results by Key Question
KQ 1a. Does Screening for Hearing Loss in Asymptomatic Adults Age 50 Years or Older Improve Health Outcomes?
Summary
One randomized trial (included in the prior USPSTF review) found that screening with the HHIE-S, the AudioScope, or both was not associated with any statistically significant difference in hearing-related QOL compared with no screening at 1 year. Although the trial did not find a difference between groups for health outcomes, it reported that screening was associated with greater hearing aid use (its primary outcome) at 1 year compared with no screening.71, 72 Effects of screening on hearing aid use appeared to be limited to patients with perceived hearing loss at baseline. Because 74 percent of patients enrolled in the trial reported perceived hearing loss at baseline and all patients were eligible to receive free hearing aids, results are likely to be most applicable to high-prevalence settings in which the cost of hearing aids is not a barrier.
Detailed Evidence
The review identified one randomized trial of screening for hearing loss (n=2,305), the Screening for Auditory Impairment—Which Hearing Assessment Test (SAI-WHAT) trial (Table 1).71, 72 The trial compared three different screening strategies with a nonscreened control group: the AudioScope (Welch Allyn, Skaneateles Falls, New York), based on inability to hear a 40-dB tone at 2,000 Hz in either ear; the HHIE-S (based on a score >10, range 0 to 40); or the AudioScope plus the HHIE-S. Included participants were predominantly male (94%), age 50 years or older (mean 61 years), and recruited from a VA Medical Center. All participants were eligible to receive free, VA-issued hearing aids. Most participants (74%) reported perceived hearing loss at enrollment (based on a “yes” or “maybe” response to the question “Do you think you have hearing loss?”). The SAI-WHAT trial was rated as fair quality, primarily because of concerns about potential deviation from intended interventions and high overall attrition for hearing-related function (23%). The study aims to compare screening with usual care; however, baseline assessment (prior to randomization) includes an assessment of self-perceived hearing loss. Participants who screened positive for hearing loss in any of the screening arms were told that they might have hearing loss and were given written instructions to call the audiology clinic for an evaluation (no referral was required to schedule an appointment). The nonscreened group was provided with a number for the audiology clinic if they wanted further assessment. Although there was no differential attrition, the high overall attrition for hearing-related function is a potential source of bias because it is possible that participants with worse function were less likely to respond to the survey.
Table 1
Characteristics and Outcomes of Included RCT for KQ 1.
Among screened groups, the proportion who screened positive was lowest in the AudioScope arm (19%) and higher in the HHIE-S arm (59%) and combined arm (64%). Hearing aid use at 1 year, the primary outcome, was significantly higher among the AudioScope arm and combined arm than the nonscreened arm (6.3% and 7.4% vs. 3.3%, respectively; p<0.01) but not among the HHIE-S arm compared with the nonscreened arm (4.1% vs. 3.3%; p>0.40). In a post hoc subgroup analysis, hearing aid use was greater among participants with perceived hearing loss at baseline (5.7% to 9.6% in screened arms vs. 4.4% in the control arm), but among those without perceived hearing loss, hearing aid use was minimal regardless of screening status (0% to 1.6%).
There was no difference in the proportion of patients who experienced a minimum clinically important difference (>6 points of improvement on a 0 to 100 scale) on the Inner Effectiveness of Aural Rehabilitation scale (a measure of hearing-related function) at 1 year (36% to 40% in the screened arms vs. 36% in the nonscreened group; p=0.39).
KQ 1b. Does the Effectiveness of Screening for Hearing Loss Differ for Subpopulations Defined by Age, Sex, Race/Ethnicity, Risk of Past Noise Exposure, or Comorbid Condition?
The SAI-WHAT trial of screening (described above) conducted post hoc analyses of hearing-related function for subpopulations defined by age.71, 72 There were no differences between screened and nonscreened groups in the proportion who experienced improvement on the Inner Effectiveness of Aural Rehabilitation scale when groups were stratified by age (50-64 years vs. ≥65 years) and according to whether they had perceived hearing loss at baseline, except in a subgroup that had both perceived hearing loss at baseline and was age 65 years or older (54% in the AudioScope arm, 34% in the HHIE-S arm, 40% in the combined arm, and 34% in the control arm; p=0.035).
KQ 2. What is the accuracy of primary care–relevant screening tests for hearing loss in adults age 50 years or older?
Summary
Thirty-four studies (reported in 35 articles) evaluated the diagnostic accuracy of clinical tests, a single question, a questionnaire, a handheld audiometric device, or a mobile-based audiometric application for identifying mild to moderate hearing loss in older adults. For detecting mild hearing loss (>20 to 25 dB), single-question screening had a pooled sensitivity of 66 percent (95% CI, 58% to 73%) and pooled specificity of 76 percent (95% CI, 68% to 83%) (10 studies, 12,637 participants);73–82 for detecting moderate hearing loss (>35 to 40 dB averaged over 2 to 4 frequencies), pooled sensitivity was 80 percent (95% CI, 68% to 88%) and the pooled specificity was 74 percent (95% CI, 59% to 85%) (6 studies, 8,774 participants).74–76, 82–84 Too few studies reported sufficient data to pool accuracy of the HHIE-S for detecting mild hearing loss (>25 dB at 2 to 4 frequencies); across 4 studies (7,194 participants) sensitivity of HHIE-S ranged from 34 to 58 percent, and specificity ranged from 76 to 95 percent.75, 78, 85, 86 For detecting moderate hearing loss (>40 dB at 2 to 4 frequencies), the pooled sensitivity of HHIE-S (5 studies; 2,820 participants) was 68 percent (95% CI, 52% to 81%) and pooled specificity was 78 percent (95% CI, 67% to 86%).3, 44, 75, 83, 85 For detecting mild hearing loss (>25 to 30 dB), pooled sensitivity of the whispered voice test was 94 percent (95% CI, 31% to 100%) and pooled specificity was 87 percent (82% to 90%) (5 studies; 669 participants).42, 73, 87–89 Fewer studies reported on the accuracy of whispered voice to detect moderate hearing loss (>40 dB) sensitivity ranged from 30 to 60 percent and specificity ranged from 80 to 98 percent (3 studies; 296 participants).73, 88, 90 Two studies (215 participants) assessed the accuracy of the AudioScope to detect at least mild hearing loss (>25 to >30 dB); sensitivities ranged from 64 to 93 percent, and specificities ranged from 70 to 91 percent.85, 91 For detecting moderate hearing loss (>40 dB), four studies (411 participants) found relatively high sensitivity (94% to 100%) and variable specificity (range: 24% to 80%) for the AudioScope.42, 44, 85, 92
Detailed Evidence
Six good74, 79, 82, 85, 93, 94 and 28 fair-quality studies (reported in 29 articles)3, 42, 44, 73, 75–78, 80, 81, 83, 84, 86–92, 95–104 assessed the accuracy of 18 different screening tools for hearing loss in older adults (Table 3). Nineteen of the included studies were in the review conducted for the USPSTF in 2011 (as noted in Table 3).5
Nine studies evaluated a clinical test (e.g., whispered voice, finger rub).42, 73, 87–90, 98, 101, 102 Thirteen studies evaluated a single question (e.g., “Do you have difficulty hearing?”);73–84, 86 11 studies (reported in 12 articles) evaluated a hearing questionnaire (e.g., HHIE-S);3, 44, 75, 78, 83, 85, 86, 94, 95, 97, 100, 104 and 10 studies evaluated a handheld or mobile-based audiometric device.42, 44, 85, 90–93, 96, 99, 103 Many studies assessed multiple screening tools.
All the studies used PTA as the reference standard, although the thresholds and the criteria used to diagnose hearing loss varied both across and within studies; specific criteria are shown in Appendix E Tables 1-4 along with the test accuracy outcomes. For instance, some studies examined accuracy in relation to mild (>25 dB) or moderate (>40 dB) levels of severity, and studies varied by the particular cut point at which a determination of mild or moderate was made. Another variation both within and across studies was whether the better ear, worse ear, or both individual ears were used to obtain hearing thresholds. Finally, studies varied on whether thresholds were speech frequency averages (i.e., 0.5, 1, 2 kHz), four-frequency averages (i.e., 0.5, 1, 2, 4 kHz), or high frequency averages (i.e., 4, 6, 8 kHz).
In terms of screening-test delivery, studies assessing a hearing questionnaire or single question involved self-administration (Table 2).3, 44, 73, 77, 78, 83–86, 93–95, 100, 102, 104 A variety of different personnel screened participants in studies examining handheld or mobile-based audiometric devices, including audiologists, speech language pathologists, primary care clinicians, research staff, and not further described “examiners.” Clinicians (i.e., neurologists, geriatricians, otolaryngologists) and audiologists administered clinical screening tools. Five of the studies did not indicate who administered the screener.42, 76, 81, 82, 84
Table 2
Characteristics of Studies Assessing Screening Test Accuracy (KQ 2).
Most studies included community-dwelling older adults enrolled from various outpatient clinical or community settings; four studies included adults who were in chronic care/rehabilitation facilities.81, 87, 92, 96 In addition, one study included cancer outpatients.99 In the 27 studies that reported sex of the participants, most included both males and females. Exceptions include two studies that were predominantly male92, 102 and one that was entirely female.76 Sample sizes analyzed varied from 30 to 4,906 participants, with a median of 107. Across the 28 studies that reported on the age of enrolled participants (mean, median, or range), the median age of participants was 69.3 years. Two studies3, 81 did not report age but had inclusion criteria limiting to older adults. Several studies included a minority of participants younger than 50 (the lower age boundary specified in our inclusion criteria), but in these studies the mean was at least 50, or we only included data for those who were age 50 or older. Of the 34 studies, only five44, 78, 79, 84, 86 reported on race or ethnicity; the percentage of participants who were white ranged from 0 to 100. Few reported any socioeconomic variables, and those that did reported the data in different ways: one study reported occupational classes,75 one study reported income,84 and six studies reported on education,82, 84–86, 95, 104 using different metrics. The majority of studies (k=17) were set in the United States.3, 44, 73, 76, 78–81, 83, 85, 91, 94, 96, 97, 100–103 The remainder were in Canada,92, 93 the United Kingdom,87–89 Australia,74, 75 other European countries;42, 77, 82, 90, 95, 98, 99, 104 and Asia.84, 86 We rated six studies as good quality74, 79, 82, 85, 93, 94 and the remainder as fair quality (Appendix D Tables 13-16). In the studies rated as fair quality, common sources of bias included unclear description of index test administration or interpretation, unclear patient selection (e.g., no description of whether a consecutive or random sampling was used, and no or unclear description of exclusion criteria related to comorbidity or symptom status).
Screening test accuracy results are organized by test category below. Many studies reported on the accuracy of screening tests to detect hearing loss defined by multiple thresholds (e.g., >25 dB, >40 dB) averaged over different frequencies; definitions of hearing loss also varied in terms of laterality (one or both ears affected). Detailed results, including all screening test cut points and hearing loss definitions based on PTA reported by included studies, are shown in Appendix E Tables 1 through 4. Table 3 summarizes results by test category and hearing loss severity.
Table 3
Summary of Accuracy for Included Screening Tests (KQ 2).
Single-Question Screening
Thirteen studies assessed the accuracy of single-question screening for detecting hearing loss (Appendix E Table 1).73–84, 86 The exact wording of the question varied slightly across studies (e.g., “Do you have a hearing problem now?” vs. “Do you feel you have a hearing loss?”). All studies indicated that an affirmative or “yes” response to the question was considered a positive screen; only one study noted that both affirmative and equivocal responses were considered a positive screen.81
For detecting mild hearing loss (>20 to 25 dB averaged over 3 to 4 frequencies), the pooled sensitivity based on 10 studies (12,637 participants) was 66 percent (95% CI, 58% to 73%) and the pooled specificity was 76 percent (95% CI, 68% to 83%) (Table 3; Appendix F Figure 1).73–82 The pooled sensitivity to detect moderate hearing loss (>35 to 40 dB averaged over 2 to 4 frequencies) based on six studies (8,774 participants) was 80 percent (95% CI, 68% to 88%) and the pooled specificity was 74 percent (95% CI, 59% to 85%) (Table 3; Appendix F Figure 2).74–76, 82–84 One additional study of 1,731 community-dwelling adults in Japan that did not report sufficient data to be included in pooled analyses found a sensitivity of 54 percent and a specificity of 78 percent for detecting mild hearing loss and a sensitivity 88 percent and a specificity of 67 percent for detecting moderate hearing loss.86
Screening Questionnaires
Eleven studies (reported in 12 articles) assessed the accuracy of screening questionnaires (Appendix E Table 2).3, 44, 75, 78, 83, 85, 86, 94, 95, 97, 100 Of these, eight studies assessed the accuracy of HHIE-S.3, 44, 75, 78, 83, 85, 86, 94, 100 Too few studies reported sufficient data to pool accuracy of the HHIE-S for detecting mild hearing loss (>25 dB at 2 to 4 frequencies). Across four studies (7,194 participants), sensitivity of HHIE-S using a cut point of score >8 ranged from 34 to 58 percent, and specificity ranged from 76 to 95 percent.75, 78, 85, 86 For detecting moderate hearing loss (>40 dB at 2 to 4 frequencies), the pooled sensitivity of HHIE-S using a cutoff score of >8 based on five studies (2820 participants) was 68 percent (95% CI, 52% to 81%) and pooled specificity was 79 percent (95% CI, 69% to 86%) (Table 3; Appendix F Figure 3).3, 44, 75, 83, 85
Two additional screening questionnaires were evaluated in one study each, the Hearing Self-Assessment Questionnaire (HSAQ)95 and the Revised Five Minute Hearing Test (RFMHT).97 For detecting mild hearing loss (>25 dB at 4 frequencies), the HSAQ had a sensitivity of 89 percent (95% CI, 78% to 96%) and specificity of 84 percent (95% CI, 72% to 92%) using a cut point of ≥15; sensitivity was slightly lower (76%) and specificity was slightly higher (96%) at a cut point of ≥19.95 The sensitivity of the RFMHT for detecting mild hearing loss was 80 percent and specificity was 55 percent.
Clinical Tests
Nine studies42, 73, 87–90, 98, 101, 102 evaluated the diagnostic accuracy of whispered voice, conversational voice, finger rub, watch tick, digits-in-noise (DIN), and WIN tests (Appendix E Table 3). Six of these studies42, 73, 87–90 assessed the accuracy of the whispered voice test at 6 inches and/or 2 feet using letters, words, or numbers, with different passing criteria, and one assessed the accuracy of the conversational voice test at 2 feet.87 For detecting mild hearing loss (>25 to 30 dB), pooled sensitivity of the whispered voice test was 94 percent (95% CI, 31% to 100%), and pooled specificity was 87 percent (82% to 90%) (5 studies; 669 participants) (Appendix F Figure 4).42, 73, 87–89 One study included in the pooled analysis reported on sensitivity and specificity of the whispered voice test when conducted by providers with different levels of experience and found variable results.88 A pooled analysis including data from experienced providers (vs. inexperienced/newly trained providers) was similar, but sensitivity was slightly higher (96%) and specificity was lower (79%). Sensitivity for detecting at least moderate hearing loss defined as >40 dB (3 studies; 296 participants) ranged from 30 to 60 percent and specificity ranged from 80 to 98 percent.73, 88, 90
Few studies assessed other clinical screening tests for hearing loss. One study (n=62) assessed the accuracy of the conversational voice test at 2 feet87 and reported low sensitivity (47%) and high specificity (100%) for detecting mild hearing loss. Watch tick and finger rub tests for detecting mild and moderate hearing loss were assessed in one study.73 Sensitivities were low for the watch tick and finger rub tests for detecting both mild (44% and 27%, respectively) and moderate hearing loss (60% and 35%, respectively); specificities were high for detecting both mild (100% and 98%, respectively) and moderate hearing loss (99% and 97%, respectively). Three studies98, 101, 102 assessed the accuracy of either DIN or WIN tests to detect mild hearing loss using different methods (Appendix E Table 3), including the U.S. National Hearing Test (a DIN telephone screening protocol developed for use within the VA)102 Sensitivity of the DIN or WIN ranged from 42 to 99 percent, with a median of 90 percent; specificity ranged between 24 and 98 percent, with a median of 86 percent.
Handheld or Mobile-Based Audiometric Devices
Ten studies evaluated the accuracy of various handheld audiometric screening devices (Appendix E Table 4).42, 44, 85, 90–93, 96, 99, 103
Five evaluated the AudioScope, a device that combines an otoscope with a portable audiometer to screen for hearing loss in the 0.5 to 4 kHz range. Two studies (215 participants) assessed the accuracy of the AudioScope to detect mild hearing loss (PTA thresholds of >25 to >30 dB); sensitivities ranged from 64 to 93 percent, and specificities ranged from 70 to 91 percent.85, 91 For detecting moderate hearing loss (≥40 dB), evidence from four studies (411 participants) found relatively high sensitivity (range: 94% to 100%) and variable specificity (range: 24% to 80%) for the AudioScope.42, 44, 85, 92 One study96 assessed the accuracy of both the AudioScope and a portable audiometer to detect moderate hearing loss (≥45 dB) in subpopulations defined by age decades, beginning with 50-year-olds through 90-year-olds. Across all age groups, sensitivities and specificities of both the AudioScope and portable audiometer for detecting moderate hearing loss were high (Appendix E Table 4). AudioScope sensitivities ranged from 85 to 90 percent, and specificities ranged from 89 to 94 percent. Similarly, sensitivities for the portable audiometer ranged from 88 to 94 percent, and specificities ranged from 90 to 94 percent.
Four studies assessed various tablet-based software audiogram apps designed for screening. Two studies by the same authors evaluated the accuracy of the uHearTM app in two separate cohorts of older adults with cancer undergoing a comprehensive geriatric assessment (78 participants) using different scoring methods to determine a positive screen.90, 99 Using a scoring method that defined a positive screening test result based on PTA ≥40 dB at 0.5, 1.0, or 2.0 kHz, sensitivity was high in both cohorts (100%) but specificity was relatively low (38% and 36%).90, 99 A revised scoring method to determine a positive screen was applied to both cohorts, defined as two or more nonconsecutive hearing grades below the moderate-to-severe threshold (≥56 dB) measured at five frequencies (from 0.5 to 4.0 kHz) in at least one ear.99 Based on this method, sensitivity varied between the first and second cohorts (100% and 68%, respectively), and specificity was similar (89% and 87%).99
One study (33 participants)93 assessed two iOS apps, EarTrumpet and ShoeBox, and found that both had relatively high sensitivity (88% and 100%, respectively) and specificity (96% for both) for detecting moderate hearing loss.
One RCT (107 participants)103 assessed three different apps—EarTrumpet (n=35), Audiogram Mobile (n=37), and Hearing Test with Audiogram (n=35)—for their ability to detect mild hearing loss (≥ 20 dB hearing loss at frequencies ranging from 0.25 to 8.0 kHz) in either a clinic waiting area or a quiet exam room. Following pure-tone audiometry, patients were randomly assigned to receive only one screening app in both the waiting area and quiet exam room. The specific screening setting had a minimal effect on test accuracy. In a clinic waiting area, all three apps had relatively high sensitivity (100%, 88%, and 89%, respectively), but more variation was seen in their specificities (72%, 92%, and 68%, respectively). In a quiet exam room, sensitivity remained relatively high (96%, 85%, and 88%, respectively), and specificity remained variable (83%, 95%, and 69%, respectively).
KQ 3a. What Are the Harms of Screening for Hearing Loss in Adults Age 50 Years or Older?
KQ 3b. Do the Harms of Screening for Hearing Loss Differ for Subpopulations Defined by Age, Sex, Race/Ethnicity, Risk of Past Noise Exposure, or Comorbid Condition?
We identified no eligible studies that evaluated harms associated with screening for hearing loss in older adults.
KQ 4a. What Is the Efficacy of Interventions for Screen-Detected Hearing Loss in Improving Health Outcomes in Adults Age 50 Years or Older?
Summary
Six trials (reported in 8 articles) evaluated benefits of amplification compared with no amplification among populations with screen-detected or recently detected, untreated age-related hearing loss over 6 weeks to 4 months.105–112 In five trials reporting on the HHIE, four found statistically significant benefit in favor of hearing aids compared with no amplification, and one crossover RCT found no difference between groups.110 Three of the four trials that found statistically significant benefit enrolled veterans (two RCTs105, 108 and one nonrandomized trial107); the difference in HHIE score changes from baseline in all three trials was greater than the 18.7-point difference considered to represent a minimal important difference.113 One RCT enrolling community volunteers found higher HHIE score changes from baseline among groups receiving two different hearing aid interventions (−18.2 points and −12.3 points) than placebo (−5.5 points); although comparisons were statistically significant for either intervention vs. placebo (p<0.001), differences between groups did not meet the score change considered to represent a minimal important difference. Four studies reported on general QOL or function and other non-hearing-related health outcomes;105, 107, 110, 111 of these, one found significant benefit in favor of the intervention on the Short Portable Mental Status Questionnaire and Geriatric Depression Scale.105 No study examined the effect of interventions on the incidence of dementia or neurocognitive impairment. The results are most applicable to older male populations with improved access to screening and no-cost hearing aids, such as veterans’ groups.
Detailed Evidence
We identified six controlled trials comparing amplification with no amplification among older adults with screen-detected or recently detected mild to moderate hearing loss (Table 4). Three were included in the 2011 review for the USPSTF,105, 107, 110 and three are newly included.108, 111, 112 All studies were set in the United States; enrolled populations included veterans (3 studies)105, 107, 108 and community-dwelling older adults (3 studies).110–112 Across all studies, the mean age of enrolled populations ranged from 69 to 74 years. Five studies enrolled a majority of males (56% to 100%), and one enrolled mostly females (53%).111 In five studies, the baseline HHIE score ranged from 29 to 51 (indicating at least mild to moderate hearing-related handicap).114 Sample sizes ranged from 15 to 380 participants. Four studies reported on race; of these, two were predominantly white (95% and 98%)105, 112 and one was 40 percent white.111 Interventions included ALDs (3 studies)107, 110, 111 and traditional hearing aids (5 studies).105, 108, 110–112 Studies varied in terms of the detail provided about hearing aid features and how they were fitted (Table 5). In the three studies evaluating ALDs, one did not describe the device,107 one assessed a devise comprising a single earbud connected to a receiver via a cord,110 and one evaluated two different ALDs: one with a remote microphone and headphones with dials for volume and tone control and a Bluetooth-enabled, single-ear-worn device that can be paired with a smartphone.111 Three studies compared multiple interventions with a no-amplification control group, including one study that assessed provision of the same hearing-aid device via two different service delivery models (one group received fitting using best-practice services from audiologists and the other group self-selected their own pre-programmed aids in a model designed to simulate OTC purchasing)112 and two that assessed both hearing aids and ALDs.107, 110 Duration of followup ranged from 6 weeks to 4 months. All studies reported on at least one hearing-related QOL and/or function outcome, primarily the HHIE questionnaire; four studies also reported on non-hearing-related health outcomes.105, 107, 110, 111 All studies were RCTs; one study also included a nonrandomized comparison of veterans who received two types of hearing aids (via randomized allocation) and a no-amplification control group that was randomized separately (to no-amplification vs. an assistive listening device).107 All studies were rated fair quality. Common sources of bias included lack of blinding (in that only one study compared amplification with a placebo device112), small sample sizes, and/or select study populations with limited descriptions of their baseline characteristics, raising concern for potential selection bias.
Table 4
Characteristics of Randomized, Controlled Trials of Treatment for Hearing Loss.
Table 5
Hearing-Related Outcomes in Randomized, Controlled Trials of Treatment.
Hearing-Related QOL and Function
All studies reported on one or more hearing-related QOL and/or function measures (Table 5). Most (5 studies) reported on the HHIE (25-items, score range: 0 to 100); lower HHIE scores (0 to 16) indicate no hearing handicap and higher scores indicate mild to moderate (17 to 42) or significant handicap (≥43).114 Mean baseline HHIE scores ranged from 29 to 51 across all study arms and were slightly higher in three studies enrolling veterans eligible for free hearing aids (36 to 51) than studies enrolling community volunteers (25 to 29).110, 112 Overall, four trials found statistically significant benefit in favor of hearing aids compared with no amplification, and one crossover RCT found no difference between groups.110 Three of the four trials that found statistically significant benefit enrolled veterans (two RCTs105, 108 and one nonrandomized trial107); the difference in HHIE score changes from baseline in all three trials was greater than the 18.7-point difference considered to represent a minimal important difference.113 The two RCTs enrolling veterans (574 total participants) both found significantly larger changes in baseline HHIE scores among those receiving hearing aids than controls over followup durations of 4 months (−34 points vs. 0 points; p<0.0001105 and 10 weeks (−17.5 points vs. +1.8 points; p<0.01).108 Results from the nonrandomized trial enrolling veterans (n=60) were consistent with the two RCTs; change in mean HHIE scores from baseline was higher among those receiving either a conventional hearing aid (−17.4 points) or a programmable hearing aid (−31.1 points) than controls (−2.2 points), p<0.001 for both comparisons.107 However, in the same study, there was no significant difference between groups randomized to an assistive living device (−4.4 points) or control (−2.2 points) over 3 months.107 The one RCT set in a non-VA setting that found benefit recruited participants via community advertisements at one academic institution (n=163) who were randomized to one of three arms: audiology-based best-practice fitted hearing aid, self-selection of the same hearing aid that was preprogramed and designed to simulate OTC purchasing, or a placebo device (fitted by an audiologist).112 At 6 weeks, the HHIE score change from baseline was higher in the audiology-based hearing aid group (−18.2 points) and OTC hearing aid group (−12.3 points) than placebo (−5.5 points); although comparisons were statistically significant for either intervention vs. control (p<0.001), differences between groups did not meet the score change considered to represent a minimal important difference (−18.7 points).112 The one study that did not find significant between-group differences in HHIE scores was a crossover RCT (n=80) enrolling community volunteers; participants were allocated to each arm for 6 weeks; there were no significant between-group differences in mean changes from baseline HHIE scores associated with hearing aids (−5.2 points), assistive listening devices (−3.2 points), hearing aids combined with assistive listening devices (−4.1 points), and no amplification (−2.2 points).110 This study also provided HHIE score changes from baseline reported by a significant other; authors only reported mean changes from baseline that were slightly larger for hearing aids (−7.5 points), assistive listening devices (−4.4 points), and hearing aids combined with assistive listening devices (−9.5 points) than no amplification (−1.4 points).110 No baseline scores, measures of variance, or significance of between-group differences were reported.
For other hearing-related outcomes, few studies reported on the same measure (Table 5). Two trials reported on the revised QDS.107, 111 One was a nonrandomized trial enrolling veterans (n=60); changes in mean score from baseline were larger among groups receiving a standard hearing aid (−0.70 points) and programmable hearing aid (−0.86 points) than controls (−0.05 point) (p=0.01), but there was no difference between the assistive listening device (+0.03 points) and no amplification controls (−0.05).107 The second trial (n=15) found a slightly larger change from baseline scores among the intervention group than controls (−5.9 vs. −2.1) but did not comment on whether the change was statistically significant.111 Two trials enrolling veterans reported on the Abbreviated Profile of Hearing Aid Benefit (APHAB), a 24-item scale used to measure self-rated communication ability.107, 108 One RCT (n=380) found larger reductions in mean scores among the hearing aid group vs. controls at 10 weeks (mean change in baseline APHAB score: −29.5 vs. +4.2; p<0.01).108 Results from the nonrandomized comparison were similar, showing larger reductions in mean APHAB scores from baseline to 3 months among those receiving standard hearing aids (−7.7 points) and programmable hearing aids (−16.3 points) than controls (−2.7 points), (p=0.01); however, changes in scores were not significantly different among groups randomized to an assistive listening device vs. control (−2.7 vs. −6.4 points).107 One trial reported on the WHO-DAS II (n=380) and found significantly larger changes from baseline scores among the hearing aid group than controls at 10 weeks (−2.9 vs. 3.2; p<0.01).108 Finally, one crossover RCT (n=15) measured changes in mean HHIE-S scores (15 items, score range: 0 to 40); at 3 months, changes in baseline scores were slightly larger among those receiving the intervention vs. controls (−8.5 vs. 0.03, respectively).111 Authors did not provide measures of variance or comment on statistical significance; however, the magnitudes of the differences between groups (and change from baseline scores) do not meet the 10-point change considered to be a clinically meaningful difference.115
Other Health Outcomes
Four studies reported on at least one general (non-hearing-related) health outcome, including general measures of QOL, cognitive function, social isolation, and depression (Table 6).105, 107, 110, 111 No outcome measure was assessed by more than one study. Three studies reported outcomes but did not provide numerical results110 or did not report sufficient information to determine whether differences between groups were significant.107, 111 One RCT enrolling veterans (n=194) found statistically significant benefit among those receiving hearing aids vs. controls on measures of cognitive function (Short Portable Mental Status Questionnaire, difference between groups in change from baseline: −0.28 points [95% CI, 0.08 to 0.48]; p=0.008) and depression (Geriatric Depression Scale, difference between groups in change from baseline: −0.80 points [95% CI, 0.09 to 1.51]; p=0.03), but no significant difference between groups on a measure of general health and function (Self Evaluation of Life Function scale, difference between groups in change from baseline: −1.9 points [95% CI, −0.1 to 4.0]; p=0.07).105
Table 6
General Health-Related QOL and Function and Other Health Outcomes in Randomized, Controlled Trials of Treatment.
KQ 4b. Does the Efficacy of Interventions for Screen-Detected Hearing Loss Differ for Subpopulations Defined by Age, Sex, Race/Ethnicity, Risk of Past Noise Exposure, or Comorbid Condition?
No subpopulation analyses were reported by the included studies.
KQ 5a. What Are the Harms of Interventions for Screen-Detected Hearing Loss in Adults Age 50 Years or Older?
KQ 5b. Do the Harms of Interventions for Screen-Detected Hearing Loss Differ for Subpopulations Defined by Age, Sex, Race/Ethnicity, Risk of Past Noise Exposure, or Comorbid Condition?
No eligible studies reporting on harms were identified.
- Results - Screening for Hearing Loss in Older Adults: An Evidence Review for the...Results - Screening for Hearing Loss in Older Adults: An Evidence Review for the U.S. Preventive Services Task Force
Your browsing activity is empty.
Activity recording is turned off.
See more...