This review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines, PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines, and the Arksey and O’Malley scoping review framework [21-23]. The protocol, including synthesis methods, was reregistered on Open Science Framework (OSF) before data collection [24]. The PRISMA 2020 and PRISMA-ScR checklists are provided in Checklists 1 and 2 [25].

A systematic literature search was conducted in Embase, the Cochrane Library, PubMed, and APA PsycNet, supplemented by manual screening of the reference lists of included studies and relevant reviews. The core search concepts included older adults, HL, auditory training, and cognitive training, with database-specific subject headings, keywords, Boolean operators, and search syntax adapted for each database. No restrictions on study design were applied during the initial search. Studies published in English and available through the final search date of July 30, 2026, were considered eligible. We also manually screened the reference lists of the included studies and relevant reviews to identify potentially eligible studies that may not have been retrieved through the database searches. The date of each database search and the complete database-specific search strategies, including all search terms, field specifications, and database-specific operators, are provided in Multimedia Appendix 1.

Studies were eligible for inclusion in the scoping review if they met the following criteria: participants were older adults, with or without HL. The review focused on ACT interventions, defined as those that explicitly integrate auditory processing training with cognitive components (eg, attention, working memory, executive function, or cognitive control) and have the primary aim of improving speech perception, auditory processing, or communication abilities in complex acoustic environments. No restrictions were applied regarding study design. Studies published in English before August 2026 were considered eligible.

For the subsequent meta-analysis, eligibility criteria were further refined. Only randomized controlled trials (RCTs) and quasi-experimental studies were included. Eligible studies were required to compare ACT interventions with a control condition, including education-based controls, waitlist controls, usual care, or traditional auditory rehabilitation. Studies were included if they reported quantitative outcomes related to auditory function (eg, speech perception, SIN performance, and auditory processing) and/or cognitive function (eg, attention, executive function, working memory, or global cognition), and provided sufficient data for effect size calculation (eg, means, SDs, and sample sizes).

Studies were excluded if they did not involve an explicit integrative auditory-cognitive intervention, focused solely on auditory training or on cognitive training without integration, lacked full-text availability (eg, protocols and conference abstracts), or did not provide sufficient data for meta-analysis.

Data were extracted independently by 2 reviewers (ML and CM) using a standardized form, with discrepancies resolved by consensus. Extracted data included study characteristics, participant demographics, hearing and cognitive status, ACT design, delivery mode, dosage, outcome measures, and key results. These data were synthesized descriptively to provide a comprehensive overview of the characteristics, intervention designs, and outcomes of ACT studies in older adults.

Two reviewers independently assessed the methodological quality of the included studies. For RCTs, the revised Cochrane risk-of-bias tool (RoB 2) [26] was used to evaluate bias across 5 domains: randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of the reported result. For quasi-experimental studies, the Joanna Briggs Institute (JBI) critical appraisal checklist for quasi-experimental studies was applied [27]. Disagreements between reviewers were resolved through discussion or consultation with a third reviewer. Among the 24 included publications, 3 articles reported the same underlying study populations as other publications and were considered multiple reports of the same study. Therefore, risk-of-bias assessment was conducted at the study level, including 16 independent RCTs assessed using RoB 2 and 5 quasi-experimental studies assessed using the JBI checklist.

The synthesis methods were preregistered before data collection on the OSF [24]. Findings were integrated through qualitative and quantitative synthesis. Content analysis was used to summarize nonquantitative findings, and a Sankey diagram was used to visualize outcome patterns. Meta-analyses were conducted in R (version 4.4.2; R Foundation for Statistical Computing) using the meta package. For continuous outcomes, intervention effects were calculated using change scores from baseline to postintervention. Mean differences (MDs) were used when outcomes were measured on the same scale, whereas standardized mean differences (SMDs; Hedges g) were used when the same construct was assessed using different scales. When change-score SDs were not directly reported, they were calculated from baseline and postintervention SDs assuming a correlation of 0.50. For studies reporting within-group change scores and Cohen d but not change-score SDs, the SDs were derived from the reported mean changes and Cohen d and were incorporated into the corresponding meta-analyses. Fixed-effects models were applied when heterogeneity was low (I²<50%), and random-effects models were applied when heterogeneity was moderate to high (I²≥50%) [28]. Heterogeneity was assessed using Cochran Q test and I². Forest plots were used to present pooled effect estimates and 95% CIs; the Egger test and funnel plots were used to assess potential publication bias when sufficient studies (≥10) were available.

A total of 688 records were identified. After removing 43 duplicates, 645 records were screened, of which 586 were excluded based on title and abstract. A full-text evaluation of the remaining 59 articles resulted in a further 35 exclusions (non-ACT interventions, reviews, unavailable texts, non-English papers, and single-modal auditory training), yielding 24 eligible articles. Among these, 3 articles were identified as multiple reports of the same intervention study, involving the same participants, group allocation, and intervention protocol. Therefore, the 24 included articles represented 21 independent studies. Eight studies were excluded from the meta-analysis for lack of data. Therefore, 16 reports corresponding to 14 studies were included in the meta-analysis. The screening process is shown in Figure 1.

The 21 included studies were published between 2007 and 2026, with 9 published in the past 5 years. Eleven were conducted in the United States, 2 in the United Kingdom, and 1 each in Switzerland, Japan, Singapore, India, Germany, Turkey, China, and Malaysia. Sixteen studies used RCT designs, and 5 used quasi-experimental designs. Sample sizes of intervention groups ranged from 11 to 63, and participant ages varied from 40 to 89 years. Participant demographic characteristics are summarized in Table 1.

Risk-of-bias assessments are presented in Multimedia Appendix 2. Among the 16 RCTs assessed using RoB 2, overall risk of bias was rated as low risk in 18.8% (n=3) of studies, some concerns in 75% (n=12), and high risk in 6.3% (n=1). The majority of concerns were related to the randomization process and selection of the reported results. Specifically, 31.3% (n=5) of studies were judged as having low risk, 62.5% (n=10) as having some concerns, and 6.3% (n=1) as having high risk regarding the randomization process. For deviations from intended interventions, most studies were rated as low risk (n=14, 87.5%). All studies showed low risk for the measurement of the outcome domain, whereas most studies had some concerns regarding selection of the reported results (n=13, 81.3%).

For the 5 quasi-experimental studies assessed using the JBI checklist, most studies demonstrated adequate methodological quality. However, several studies had potential concerns regarding selection and allocation procedures, temporal precedence, and control of confounding factors. Detailed assessments for each domain are presented in Multimedia Appendix 2 (JBI critical appraisal qualitative evidence syntheses).

Hearing characteristics were heterogeneous across studies. Four studies included participants with normal or near-normal hearing, 7 studies focused on mild HL, and 9 studies investigated mild-to-moderate HL. Five studies included mixed hearing profiles, combining participants with different degrees of hearing ability. Because several studies recruited heterogeneous samples, some studies were classified into multiple categories. HA use was uncommon; only 2 studies included HA users, whereas most studies enrolled nonusers.

Twelve studies included cognitively normal older adults, defined via cutoffs on Mini-Mental State Examination (MMSE; MMSE scores: ≥24‐25 or between 27 and 30)/Montreal Cognitive Assessment (MoCA; MoCA scores: ≥26 or ≥23) or IQ testing and a clinical interview (n=2; normal IQ). One study enrolled participants with educationally appropriate scores on the MMSE, one enrolled participants with subjective cognitive decline, and one included both cognitively normal and mild cognitive impairment groups. The remaining 4 studies did not report sufficient cognitive information.

Table 2 provides an overview of ACT delivery methods, modalities, content, and dosage across the included studies; detailed intervention and control characteristics, follow-up, and measurements are provided in Table S1, Multimedia Appendix 1.

This mixed synthesis is the first to summarize ACT in older adults. A total of 21 studies (24 articles) involving 1300 older adults from 10 countries (the United States, United Kingdom, Switzerland, Japan, Malaysia, Germany, Turkey, Singapore, India, and China) were included. This review comprehensively summarized key features of existing ACT interventions, including participants’ hearing characteristics, delivery modes, training content, frequency and dosage, as well as outcome-specific effects. Considerable heterogeneity was observed in target populations’ hearing status, and ACT was delivered via diverse modalities including computer-based, mobile app, VR, and face-to-face training. Training content fell into 3 categories: auditory training, auditory-cognitive integrated training, and cognitive training, with a median total training dosage of 10 (IQR 1.5-60) hours. Narrative evidence indicated consistent short-term benefits of ACT for SIN recognition, attention, and auditory memory, with some benefits maintained at follow-up. Meta-analyses further supported significant beneficial effects of ACT on SIN recognition, attention, overall memory, and global cognition, although substantial heterogeneity was observed for SIN, memory, and global cognition. In contrast, pooled effects were not significant for working memory, processing speed, self-reported hearing outcomes, or PTA, while evidence for auditory stream segregation was limited but consistently positive.

Current evidence shows that ACT has been applied to older adults across a range of hearing levels for both preventive and rehabilitative purposes. It has been used not only in those with mild-to-moderate HL to improve speech perception but also in older adults with normal audiometric thresholds who report subjective hearing difficulties to reduce listening effort and cognitive load. ACT has also been used as an adjunct to audiological rehabilitation before or after HA fitting, suggesting potential value in boosting auditory and cognitive outcomes. Most existing ACT studies have been conducted in older adults with normal cognition or mild cognitive impairment. Notably, Ferguson et al [32] reported that cognitively normal participants achieved greater ACT-related improvements in SIN perception, gap detection, and pitch pattern sequencing than did those with mild cognitive impairment. However, this between-group difference should be interpreted cautiously, as it may partly reflect regression to the mean or baseline differences rather than differential responsiveness to ACT. Thus, the extent to which baseline cognitive status moderates ACT effectiveness remains uncertain. For those with existing cognitive impairment, higher training dosage and broader multidomain interventions (eg, physical activity and psychosocial approaches) may warrant consideration [47], highlighting the need for tailored protocols based on baseline cognitive function.

To date, no standardized ACT protocols exist, with substantial variability in delivery modalities, content, and dosage. Computer-based and mobile app–based interventions are more accessible and flexible than face-to-face or VR training, supporting higher adherence through shorter, more frequent sessions, with a median total dosage of 10 hours. Integrated ACT appears superior to unimodal auditory or cognitive training alone. For example, Faust et al [18] found that separate auditory or visual cognitive training improved modality-specific functions, but simple coadministration did not produce synergistic benefits, emphasizing the importance of genuine integration (eg, combining speech perception with memory or attention demands) rather than parallel task delivery. Common integrated designs pair auditory stimuli with demands on short-term/working memory, attention, processing speed, or executive function. All included ACT interventions used adaptive difficulty. Although only one study [20] directly compared adaptive and fixed difficulty, adaptive protocols may better accommodate individual learning rates and heterogeneous baseline abilities in older adults, thereby improving engagement and training outcomes.

Regarding intervention effects, the most consistent improvement was observed in SIN recognition, with benefits maintained for 1 to 2 months postintervention [34,38], and short-term effects supported by a meta-analysis. A potential mechanism is that ACT may reduce excessive subcortical encoding of speech envelopes, strengthen top-down attentional control, and enhance functional connectivity between the prefrontal and auditory cortices, thereby improving temporal processing and SIN perception [14,15]. Nevertheless, the pooled estimate showed substantial heterogeneity (I²=87.2%), which may partly reflect differences in SIN assessment paradigms, including HINT (Hearing-in-Noise Test), WIN (Words-in-Noise test), QuickSIN (Quick Speech-in-Noise), speech-in-babble, and competing-speaker tasks, as well as differences in participants’ baseline hearing status and HA use. The pooled effect indicated a significant reduction in signal-to-noise ratio/speech reception threshold of approximately 1.23 dB following ACT (MD=−1.23, 95% CI −2.35 to −0.10 dB), suggesting a clinically meaningful improvement in speech recognition under noisy conditions [48]. However, the high between-study heterogeneity indicates that the magnitude of benefit varied considerably across studies. Subgroup analysis by HA use provided some additional insight into this heterogeneity. Among participants who did not use HAs, the pooled effect was significant and showed low heterogeneity (I²=9.3%), whereas the effect among HA users was not statistically significant and was accompanied by substantial heterogeneity (I²=92.0%). Although the test for subgroup differences was not statistically significant (P=.24), the marked difference in within-subgroup heterogeneity suggests that HA use may be one potential contributor to the variability in treatment effects across studies. This modest improvement may also have limited implications for real-world hearing function, consistent with the nonsignificant pooled effect for self-reported hearing outcomes (SSQ). By contrast, ACT showed minimal effects on pure-tone thresholds, consistent with the limited reversibility of age-related cochlear degeneration [49]. This may be because ACT primarily targets central auditory and cognitive processing rather than peripheral hearing sensitivity.

For cognitive outcomes, the meta-analysis showed significant pooled effects for global cognition, attention, and memory, whereas the effects for working memory and processing speed were not statistically significant. Among these outcomes, the evidence for attention was the most consistent, with a significant effect (SMD=0.50, 95% CI 0.23-0.77) and low heterogeneity (I²=29.5%); moreover, the certainty of evidence was rated as moderate. This finding suggests that ACT may have a relatively robust beneficial effect on attention, potentially because the integration of auditory processing with cognitive control, selective attention, and task monitoring repeatedly engages overlapping auditory-cognitive processes. Such integrated training may strengthen the ability to allocate and maintain attention to relevant auditory information, particularly under cognitively demanding listening conditions. The pooled effect for memory was statistically significant but accompanied by substantial heterogeneity, and neither subgroup analysis by memory domain nor leave-one-out sensitivity analysis identified a major contributor to this heterogeneity. Therefore, the memory findings should be interpreted cautiously, given the substantial between-study variability and uncertainty regarding the magnitude and generalizability of the effect. For global cognition, the significant pooled effect should also be interpreted cautiously because it was based on a small number of studies and was accompanied by substantial heterogeneity. Similarly, the nonsignificant findings for working memory and processing speed provide insufficient evidence to determine whether ACT produces consistent benefits in these domains, particularly given the limited number of studies and variability in outcome measures and intervention characteristics. Overall, the current evidence suggests that the cognitive effects of ACT may vary across domains, with relatively stronger and more consistent evidence for attention than for broader or more specific cognitive functions. Evidence regarding the longer-term maintenance of cognitive benefits remains limited, particularly because of the scarcity of follow-up assessments, heterogeneous measures, and small sample sizes. Further well-designed trials using standardized cognitive outcomes, larger samples, and longer follow-up periods are needed to clarify the magnitude, generalizability, and durability of these cognitive effects.

We note several limitations when interpreting both the included evidence and our synthesis. At the evidence level, most studies used passive or waitlist control groups, making it difficult to distinguish the specific effects of ACT from those of single-domain auditory or cognitive training. Commercial training platforms (eg, CogniFit) may offer limited customization for SIN tasks and limited crosslinguistic adaptability, which may reduce ecological validity and generalizability. Few studies considered stimulus delivery modes (eg, headphones vs loudspeakers), which may affect listening comfort and effort in older adults with HL. In addition, most studies enrolled cognitively normal participants; only one study directly compared responses between cognitively normal and cognitively impaired groups. Thus, the efficacy of ACT for older adults with cognitive impairment remains uncertain. Few studies used patient-reported real-life listening outcomes, such as subjective hearing difficulty and daily listening function, as primary end points. Given the importance of subjective hearing experience for communication confidence, social participation, mental health, and quality of life [50], future studies should incorporate more patient-reported measures and longer follow-up periods to assess the persistence and real-life relevance of training benefits.

Our review also has its own limitations. First, we restricted inclusion to English-language publications, which may have introduced language bias and led to the omission of relevant evidence published in other languages. Although the included evidence covered several non–English-speaking countries, including Japan, Germany, and Malaysia, relevant studies published in local languages may have been overlooked. This limitation may reduce the completeness and crosslinguistic representativeness of the evidence base and should be considered when interpreting the generalizability of the findings. Second, although meta-analyses were performed for key auditory and cognitive outcomes, pooled effects should be interpreted cautiously because of substantial heterogeneity and the small number of studies contributing to some outcomes. Variability in construct definitions, training dosage, delivery modes, and assessment time points may have reduced comparability across studies. Third, limited follow-up data prevented quantitative synthesis of maintenance effects across domains. Despite these limitations, this synthesis integrated narrative and meta-analytic evidence to characterize outcome patterns, quantify short-term effects, and identify key research gaps, thereby providing a basis for the standardization, refinement, and mechanistic investigation of ACT interventions for older adults.

This review systematically synthesizes the current evidence on ACT for older adults through a combined scoping review and meta-analytic approach. This review indicates that ACT may improve selected auditory and cognitive outcomes in older adults, particularly SIN perception, attention, memory, and global cognition. However, the certainty of the evidence was low for most outcomes, with substantial heterogeneity observed for several pooled effects. Evidence for changes in hearing thresholds, self-reported hearing function, working memory, and processing speed remains limited or inconclusive, and evidence for long-term maintenance is insufficient. These findings support the potential of ACT as a complementary approach to promote auditory-cognitive function in older adults, but do not yet establish its clinical effectiveness. Future trials should use standardized and adaptable ACT protocols, include patient-reported real-life listening outcomes, recruit more diverse samples with varied cognitive and hearing profiles, and incorporate longer follow-up periods. Establishing evidence-based protocols and clinically meaningful outcome benchmarks will be important for determining how ACT can be integrated into individualized hearing and cognitive health interventions.