AD-HEARING (ADherence to and adjustment of HEARING aids in clinical routine care as preventive dementia strategy: Improvement of cognition and well-being) is a monocentric, prospective, longitudinal, and quasi-experimental intervention study of hearing aid use under conditions of regular healthcare delivery in Germany. Starting in 2021, participants were recruited consecutively via various sources (active involvement of regional otolaryngologists and hearing care professionals as well as geriatric departments of other hospitals in the vicinity, including provision of flyers and posters for potential participants, intranet and social media announcements on the website of the University Medical Center (UMC) Göttingen, local press releases, and public dementia events). In-house recruitment at the Department of Psychiatry and Psychotherapy or the Department of Otolaryngology at the UMC Göttingen was also possible.

The potential participants had to meet the following inclusion criteria: (1) ability to give informed consent, (2) age ≥60 years, (3) speaking German fluently and being able to adequately perform the Freiburg Speech Test as a readout for ARHL and adequacy of hearing rehabilitation, (4) no dementia diagnosis of any cause prior to baseline, and (5) bilateral mild to moderate sensorineural hearing loss (HL) as an indication for a provision with hearing aids. HL was defined according to the recommendations of the World Health Organization (87), i.e., mean value of hearing thresholds at 0.5, 1, 2, 4 kHz (pure tone average, PTA-4) between 20 and 60 dB HL of the better ear. The participants with a conductive component of >20 dB HL at 0.5, 1, 2, or 4 kHz as well as the subjects for whom optimal hearing rehabilitation would probably have required ear surgery (e.g., insertion of an implantable hearing system) were excluded. The participants with a diagnosis of purely unilateral HL could also not qualify for this study. Participation was only possible for first-time hearing aid users (i.e., beginning of hearing aid use within the past 6 months prior to study participation).

The eligible subjects with a hearing aid prescription underwent baseline study procedures. Afterwards or in parallel, hearing aid provision as part of the routine healthcare in Germany (compensated by the individuals’ health insurance) was initiated by hearing care professionals. Approximately 6 months later, a second study visit (follow-up) was scheduled. For this follow-up visit, hearing aid users had to provide a fitting protocol for their hearing aids by their hearing care professional as an objective outcome for fitting quality. Following the German guidelines, an improvement of ≥20% in the Freiburg Speech Test in quiet was defined as adequate and successful hearing rehabilitation (88).

Participation was voluntary, and withdrawal was possible at any time. No financial incentives were offered for participation. All participants gave written informed consent. The study and all procedures involving human subjects were approved by the ethics committee of UMC Göttingen, Germany (#36/2/21), and pre-registered at the German Clinical Trials Registry (#DRKS00025111). All study procedures have been conducted according to the Helsinki Declaration of 1975, as revised in 2008.

Baseline and follow-up study parameters were collected in a time-period from May 28, 2021 to November 28, 2023, with an average number of 195 ± 3.7 days between baseline and follow-up assessment (median time-interval of 6.2 months). The predefined primary study outcomes comprised the parameters of (1) cognition [Consortium to Establish a Registry for Alzheimer’s Disease-Plus test battery; CERAD-plus (89), Chandler Score (90)] and psychological well-being, in particular (2) depression [Geriatric Depression Scale—15-item short form; GDS-SF (91, 92)], (3) social isolation [six-item short version of the Lubben Social Networking Scale; LSNS-6 (93, 94)], (4) psychological burden [General Symptom Index (GSI) T-values of the Symptom Checklist-90^®-Standard; SCL-90^®-S (95)], and (5) health-related quality of life [visual analogue scale of the EQ-5D (96)].

All exploratory secondary outcomes were mainly derived from the assessments on cognition and psychological well-being as listed above. For (1) cognition, the Mini Mental Status Test [MMST (97)] as part of the CERAD-plus battery (range: 0–30), the CERAD Memory Index (range: 0–100%) as well as the CERAD Memory Score (range: 0–41 (98), and the sum score of digit spans (forward and backward, range: 0–36) of the Wechsler Adult Intelligence Scale—4th edition [WAIS-IV (99, 100)] have also entered analysis (digit spans were used to additionally assess working memory). Normative values of CERAD-plus single test results and WAIS-IV digit spans were considered to differentiate between normal cognitive performance and MCI [according to the operationalized MCI criteria by Molinuevo et al. (101)]. At baseline only, the participants additionally gave estimations on their subjective cognitive performance level on a numeric rating scale (0 = “very bad” to 10 = “very good”; descriptive variable). As secondary outcomes for (4) psychological burden, the indices Positive Symptom Total (PST, T-values) and the Positive Symptom Distress Index (PSDI, T-values) of SCL-90^®-S were included.

Hearing aid-related healthcare outcomes comprised fitting protocols for hearing aids (baseline and follow-up) and the results from the International Outcome Inventory for Hearing Aids (IOI-HA, German version (102, 103); follow-up only). Additionally, the participants rated their subjective hearing ability on a numeric rating scale (0 = “very bad” to 10 = “very good”; descriptive variable). The fittings protocols contained results from audiometric assessments performed by hearing care professionals to detect ARHL and during the process of routine hearing aid provision and adjustment to assure successful hearing rehabilitation. PTA-4 served as a readout of ARHL and was determined as the mean of the hearing thresholds in dB HL for 0.5, 1, 2, and 4kHz frequencies of both ears. In the IOI-HA, the participants gave a self-report on successful hearing aid use. Item 1 denotes daily use of hearing aids during the past 2 weeks (ratings from 1 = “never” to 5 = “more than 8 hours”), and items 2–7 allow ratings for satisfaction with hearing aid use (ratings from 1 to 5 with varying wording anchors). A mean value ≥3 of the items 2–7 is regarded as general satisfaction with hearing aid use.

Procedures of hearing aid-related regular healthcare (audiometric assessments, prescription of hearing aids) were realized by otolaryngologists and hearing care professionals. Study procedures (psychometric assessments) were performed by trained and experienced neuropsychologists. It took approximately 60–90 min to complete all study-related baseline and follow-up procedures for each participant (including informed consent, medical history, and study assessments).

A total of 53 individuals were screened, of whom 31 with ARHL met all of the eligibility criteria described above and underwent baseline study procedures (inclusion between 05/2021 and 05/2023). Reasons for screening failure were (1) already being provided with/using hearing aids >6 months, (2) the ARHL was too mild for the patient to receive a hearing aid prescription, (3) subjective ARHL not confirmed by an otolaryngologist, and (4) age <60 years. After baseline assessment, one participant withdrew his consent, and two more were lost to follow-up. The final analysis sample with baseline and follow-up data comprised N = 28 participants (dropout rate of 9.7%; see Figure 1 ).

At follow-up, the final analysis sample was subdivided into two groups differing in successful hearing aid use. Post hoc classification was based on the rating of daily hearing aid use for the past 2 weeks [item 1 of the IOI-HA (102, 103)]. Such n = 17 were defined as hearing aid non-users (daily use ≤8 h) and n = 11 as hearing aid users (daily use >8 h, i.e., most time of the day; see Figure 1 ).

For data analysis, IBM SPSS Statistics 29 was used. Means with standard deviations (M ± SD), frequencies, Pearson correlations (r), and mean differences (M_Diff ) were calculated for descriptive representation, respectively. Multiple t-tests for independent samples and chi-square tests were used to test for significant differences between both subsamples (users vs. non-users) at baseline.

To analyze the predefined five primary outcomes (cognition, depression, social isolation, psychological burden, and quality of life), five general linear models (GLM) for repeated measures were created. Baseline and follow-up data were integrated as two-staged within-subjects factor, with both subgroups (users vs. non-users) as two-staged between-subjects factor. Besides both main effects, we tested for interaction effects to analyze possible different trajectories between both subgroups over time and conducted uncorrected pairwise comparisons between both groups for baseline and follow-up (alpha set at 0.05, two-sided).

Additionally, exploratory analyses on our secondary outcomes were conducted (please see Supplementary Material S1 and the results for further information).

Please see Table 1 for an overview of the baseline characteristics. Of N = 28 participants, 60.7% (n = 17) were female with a mean age of 72.21 ± 6.94 years, ranging from 60 to 86 years. The majority of subjects were married (n = 18, 64.3%). Mean education amounted to 15.88 ± 2.32 years. All participants were German native speakers. Medical history (past and current) mainly revealed cardiovascular diseases (n = 18, 64.3%; i.e., hypertension in most cases), followed by endocrinological diseases (i.e., thyroid dysfunction/hypothyroidism; n = 9, 32.1%), psychiatric disorders (i.e., depression, bipolar disorder, past post-operational delirium; n = 9, 32.1%), metabolic conditions (i.e., diabetes and hypercholesterolemia; n = 8, 28.6%), and neurological disorders (i.e., past traumatic brain injury, childhood epilepsy, seizure-free for years; n = 2, 7.1%). In single cases, the participants suffered from chronic obstructive pulmonary disease and polyneuropathy, rheumatism, and juvenile meningitis (summarized as “others”; n = 3, 10.7%). In three cases, malignancies were reported, all of them backdated for several years. According to their self-reported medical history, antihypertensives were mentioned as the most frequently prescribed concurrent medication (n = 18, 66.7%). Other regular drug use comprised l-thyroxine (n = 8, 29.6%), statins (n = 6, 22.2%), and antidepressants (n = 3, 11.1%).

With a mean of 33.34 ± 7.25 dB, pure tone average (PTA-4) was indicative of mild hearing impairment (i.e., 20–40 dB HL). While all participants had been diagnosed with at least mild ARHL, they rated their hearing ability in the middle range at baseline (11-point numeric rating scale: 0 = “very bad” to 10 = “very good”; 6.66 ± 1.69, ranging from 3 to 10).

The participants rated their subjective cognitive performance as 7.10 ± 1.96, i.e., in the upper range of the numeric rating scale ranging from 0 to 10. The MMST sum scores (28.96 ± 1.20) revealed mild cognitive deficits in n = 2 (7.1%, i.e., MMST ≤26), while n = 26 (92.9%, MMST >26) showed normal cognitive performance. Applying the normative cutoff >69 for cognitive performance of cognitively healthy older controls presented by Paajanen et al. (98), CERAD compound scores (CERAD Chandler Score: 82.34 ± 6.99; CERAD Memory Index (98): 92.10 ± 12.45) were mildly impaired in n = 2 (7.1%). According to the operationalized MCI criteria by Molinuevo et al. (101), cognitive performance in CERAD-plus and WAIS-IV digit spans at baseline had to be classified as MCI in n = 11 (39.3%) and as normal in n = 17 (60.7%). Patients with MCI differed significantly from those without MCI in the following cognitive parameters: CERAD Memory Score (18.93 ± 4.23 vs. 22.36 ± 1.77; p = .025); z-values of CERAD Figure Recall (-0.41 ± 1.20 vs. 0.96 ± 0.72; p = .004) and CERAD Figure Savings (-0.50 ± 0.93 vs. 0.49 ± 0.50; p = .006). In none of the cases an incident diagnosis of dementia had to be made.

Interpreting the GDS-SF sum scores (2.07 ± 2.73, range 0–11), n = 25 (89.3%) had no or not more than subsyndromal depression, n = 2 (7.1%) mild to moderate depression, and n = 1 (3.6%) severe depression. In n = 3 (10.7%) LSNS-6 sum scores (18.39 ± 4.95, range 5–24), the ratings characterized the subjects as socially isolated (cutoff for social isolation at 12 points). For the SCL-90^®-S GSI (48.54 ± 9.65, range 31–75), n = 2 (7.2%) showed T-values >60, i.e., above-average psychological burden. An EQ-5D rating on a visual analogue scale (0–100) reached 74.36 ± 18.69 (range 11–98), with n = 2 (7.1%) rating their health-related quality of life <50.

As to be expected, most primary outcomes for psychological well-being showed a strong covariation (please see Table 2 for correlations between primary outcomes at baseline). This mainly comprised positive correlations between symptom-oriented scales (e.g., GDS-SF and SCL-90^®-S GSI; r = .688, p <.001) and negative correlations between symptom-oriented vs. resource-based scales (e.g., SCL-90^®-S GSI and EQ-5D; r = -.573, p = .001).

Hearing aid users did not differ significantly from non-users in any of the sociodemographic and clinical baseline characteristics (please see Table 1 , all comparisons ns). Five participants stated at follow-up to have never used hearing aids (17.9%). Of those using hearing aids (n = 23), satisfaction with hearing aid use (quantified as mean of items 2–7 of the IOI-HA) was high in general (3.99 ± 0.56 for the final analysis sample) with significantly higher ratings of users (4.30 ± 0.36) compared to non-users (3.68 ± 0.55; p = .006). At follow-up, all participants who had provided a fitting protocol showed successful hearing aid provision, resulting in at least 20% improvement of the Freiburg Speech Test.

For primary outcome analysis, please see Figure 2 for the trajectories from baseline to follow-up for hearing aid users vs. non-users and Table 3 for all GLM with tests for (main) effects and pairwise comparisons. For (1) cognition (CERAD Chandler Score, Figure 2A ) and (3) social isolation (LSNS-6, Figure 2C ), we did not find any significant (main or interaction) effect or pairwise comparison between both groups at baseline or follow-up. For (2) depression (GDS-SF, Figure 2B ), a significant between-subjects effect was found (F(1, 26) = 4.94, partial η ² = 0.16, p = .035). Both at baseline (M_Diff = 1.91, p = .033) and follow-up (M_Diff = 1.83, p = .031), hearing aid users showed significantly fewer depressive symptoms. For both assessment time-points, this significant between-group differences remained at a stable level, resulting in a non-significant interaction effect (F(1, 26) = 0.01, ns). Although for (4) psychological burden (SCL-90(R)-S GSI, Figure 2D ), the interaction effect failed to reach significance (F(1, 26) = 1.30, ns), the descriptive analysis suggested that hearing aid users had a decrease of psychological burden from baseline (42.45 ± 7.70) to follow-up (39.09 ± 6.66). Furthermore, a significant between-subjects effect was found (F(1, 26) = 13.00, partial η ² = 0.33, p = .001), which was more pronounced in pairwise comparisons at follow-up (M_Diff = 12.56, p <.001) than at baseline (M_Diff = 10.02, p = .005) with less psychological burden for hearing aid users than in non-users. A similar pattern was found for (5) health-related quality of life (EQ-5D, Figure 2E ). The interaction effect failed to reach significance (F(1, 26) = 2.44, ns) as did the within-subjects effect (F(1, 26) = 0.36, ns), and the EQ-5D ratings varied between both subgroups between baseline and follow-up, with numerically higher quality-of-life estimations for users at both time-points. While at baseline the between-group difference did not reach significance (M_Diff = -8.70, p = .236), it did so at follow-up (M_Diff = -18.86, p = .005). Within the GLM, this was mainly expressed by a significant between-subjects effect (F(1, 26) = 5.50, partial η ² = 0.18, p = .027).

Data on exploratory secondary outcomes can be found in Supplementary Material S1 . The trajectories from baseline to follow-up for hearing aid users vs. non-users are visualized in Supplementary Figure S1 . Please see also Supplementary Table S1 for all GLMs with tests for main and interaction effects and pairwise comparisons. For (1) cognition parameters, MMST, CERAD Memory Index, CERAD Memory Score, and WAIS-IV digit spans neither significant (main or interaction) effects nor significant pairwise comparisons between the subgroups at baseline or follow-up were found (see Supplementary Figures S1A–F ). As an additional readout for (1) cognition, the chi-square tests did not reveal a significant deviation of frequencies of syndromal MCI diagnoses between hearing aid users vs. non-users, both at baseline (users: n = 3 (27.3%); non-users: n = 8 (47.1%); ns) and follow-up (users: n = 3 (27.3%); non-users: n = 7 (41.2%); ns). For (4) psychological burden, the additional outcomes PST and PSDI did not reach significance for interaction effects (F(1, 26) = 2.25 and 0.01, both ns, Supplementary Figures S1E,F ) as well. However, for PST, a significant between-subjects effect was found (F(1, 26) = 15.79, partial η ² = 0.38, p <.001), with hearing aid users experiencing significantly less psychological burden at baseline (M_Diff = 9.41, p = .033). This difference increased at follow-up (M_Diff = 12.87, p <.001; Supplementary Figure S1E ). A significant between-subjects effect was also detected for PSDI (F(1, 26) = 5.26, partial η ² = 0.17, p = .030). The hearing aid users showed less psychological burden in general, barely missing significance at baseline (M_Diff = 6.52, p = .057) but reaching significance at follow-up (M_Diff = 9.41, p = .033) due to a lowered standard deviation (see also Supplementary Figure S1F for 95% CIs).

Our study was deliberately designed to combine cognitive outcomes with different aspects of psychological well-being like depression, social isolation, psychological burden, and quality of life. For this purpose of a multifaceted approach, we have chosen established cognitive tests and psychometric instruments. It is also a clear strength of our study design to present a well-characterized sample of subjects with ARHL and to rely on audiometric data for the measurement of ARHL, fitting protocols to assure adequate and successful hearing aid rehabilitation and on psychometric data for hearing aid use (IOI-SH). With these procedures, it was able to limit self-report bias and ARHL severity bias. For hearing aid use, automated readouts of hearing aid logs could additionally improve the quality of our study design.

Most obviously, our sample size was too small to make definitive claims on the prospects of hearing aid use for cognition and psychological well-being. However, comparable prospective studies (70–72, 74–77) and even RCTs (78–82) share similar or even smaller sample sizes. Although this issue might have prevented the present study from achieving significant results, we found some mild positive effects on psychological burden and quality of life in hearing aid users. Furthermore, a follow-up period of 6 months of hearing aid use may be too short to obtain noticeable preventive effects on cognitive decline and dementia and is at best suitable to detect rather short-term changes in cognitive performance and psychological well-being. Importantly, follow-up in study designs like the present can easily be prolonged without ethical considerations because of not preventing participants eligible for hearing aids from their use. Although our study cannot compete with the standards of large-scale RCTs like the ACHIEVE study (83), the ethical concern has to be raised whether subjects of the control conditions may be prevented from using hearing aids for such a long time (i.e., 3 years in the ACHIEVE RCT). Even if it is unclear if hearing aid use is helpful in preventing cognitive impairment, decline, and dementia, it is a treatment which is definitely helpful in treating ARHL, combining cost-effectiveness in dementia prevention with quality-of-life gains (118)—and is yet undertreated. We therefore put up for discussion if non-randomized controlled trials or quasi-experimental studies like ours may be ethically more appropriate and comparably suitable to ultimately draw definite conclusions on the effects of hearing aid use for dementia prevention. For this purpose, (multiple) future long-term studies are needed—at best overcoming interfering factors as described above. Large samples in those trials will also enable to identify subgroups with more clear and substantial benefits for dementia prevention through the use of hearing aids.