A cross-sectional study with a stratified whole-group random sampling method was conducted in Urumqi, Xinjiang, China. First, the nine municipal districts of Urumqi were divided into district stratification, and three municipal districts were randomly selected using the random number table method. Next, one community in each municipal district was randomly selected using the same method. Three communities were finally selected as representative residents in this study. Participants were recruited from community health centers between January and July 2022, and face-to-face interviews with participants and on-site data collection were conducted. As a result of the aging process, it is possible that older individuals experiencing hearing loss may encounter challenges in accurately responding to inquiries owing to their diminished auditory capabilities. Thus, all investigators were professionally trained. When participants had hearing difficulties, investigators minimized the effect of hearing difficulties on question answering by slowing down the speed of speech, increasing the volume of the voice, or using the text form so that the participants could understand the items of the questionnaires clearly and answer them accordingly. In addition, to avoid language and cultural differences, the investigators included health workers and medical students who were familiar with the ethnic languages. Inclusion criteria were as follows: (1) current patients aged ≥65 years; (2) living in a community for >6 months; (3) volunteering to participate and providing informed consent for this study; and (4) clear consciousness and no language impairment. Exclusion criteria were as follows: (1) patients with physician-diagnosed dementia, mental or severe physical illness that prevents them from performing cognitive assessment tests; (2) hospitalized and elderly patients who are bedridden for long periods.

The N:q ratio is considered a simple and reliable rule to estimate the minimum sample size of a structural equation model because it takes into account the complexity of the model to be estimated, in which N is the number of samples and q is the parameter to be estimated in the model. Bentler and Chou (1987) state that a ratio of at least 5:1 between the sample size and the parameter being estimated is required to ensure that the estimate of the parameter is plausible, and a ratio of at least 10:1 is required to ensure the validity of the significance test (Bentler and Chou, 1987). Jackson (2003) suggests that 20:1 is a more desirable ratio (Jackson, 2003). In this study we used a ratio of 20:1 to determine the minimum sample size, where 22 parameters (coefficients of 6 paths, factor loadings between 5 observed and latent variables, measurement errors of 9 observable variables, and structural errors of 2 endogenous latent variables) needed to be estimated in this study, and a sample size of at least 440 is recommended. Recruiting a total of 646 participants, of whom 12 had dementia, 2 had missing data on both SRH and hearing function, 3 had missing data on SRH, 1 had missing data on hearing function, and 4 had missing data on cognitive function. By including 624 participants, the ultimate study sample met the minimum requirements for sample size.

The Medical Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University approved this study (Ethics Approval No. K202009-05). All participants provided written informed consent.

For data processing and statistical analysis, IBM SPSS 27.0 was used. Continuous variables were assessed for normality using the Kolmogorov-Sminov test. Normally distributed data are expressed as mean ± standard deviation(SD), whereas non-normally distributed data are expressed as median with an interquartile range of 25–75% in parentheses. Categorical variables were described using the number of cases and percentages. Chi-square tests and Fisher’s exact tests were used to compare sociodemographic and health-related characteristics between participants with and without cognitive impairment. Differences in MMSE scores across sociodemographic and health-related characteristics were compared using the Mann–Whitney test and the Kruskal-Wallis test. Spearman’s correlations were conducted to test the relationships between hearing loss, SRH scores, depressive symptom scores, anxiety symptom scores, and cognitive function scores. Statistical significance was defined as a value of p of <0.05.

IBM AMOS version 26.0 was used to construct the structural equation model, and since the hearing function is a dichotomous variable, we processed it as a dummy variable before incorporating it into the study model, used the maximum likelihood method for parameter estimation, and evaluated the model fit using the goodness-of-fit test. The following criteria were used to judge the goodness of fit of the model: the chi-squared divided by the degrees of freedom (χ²/df) ratio was less than 3, Goodness of Fit Index (GFI), Adjusted Goodness of Fit Index (AGFI), Comparative Fit Index (CFI), Normed Fit Index (NFI), and Tucker-Lewis Index (TLI) were greater than 0.90, and the root mean square error of approximation (RMSEA) was less than 0.08; The stability of the mediating effects and parameter estimates were examined using the Bootstrap method (Barber and Thompson, 2000). Bootstrapping is a type of repeated sampling in which a random sample of the same size is drawn with a return, and does not require the assumption of normality of the sampling distribution. We used 5,000 bootstrap resamples to calculate 95% bias-corrected confidence intervals (Preacher and Hayes, 2008). If the upper and lower limits did not include 0, it was statistically significant.

Robustness tests were conducted to determine the validity of the study model. The first method is, the alternative modeling method. Model 4 of the SPSS Macro Process was used to test for mediating effects (Hayes and Preacher, 2014), and variables that differed significantly in the chi-square test and Fisher’s exact test were used as control variables. The second method, changing the sample composition. Expanding the sample size fills in missing data values through multiple interpolations. Reducing the sample size tested the male sample by stratifying for gender.

Data collected through participants’ self-reports may be subject to common method bias, and this bias may lead to misinterpretation of the study results. To rule out common method bias, Harman’s one-factor test was utilized (Podsakoff et al., 2003). This revealed that the amount of variance explained by the first common factor accounted for was 20.89% (<40%), indicating that there was no substantial common method bias in the study’s data.

Participants were 624 community-dwelling older adults with a mean age of 72.42 ± 6.07 years were included, of which 151 (24.2%) had cognitive impairment and 249 (39.9%) had hearing loss. In addition, only 9 of 249 older adults with hearing loss wore hearing aids (3.61%), and 33.7% of all older adults with hearing loss had cognitive impairment. Sociodemographic and health-related characteristics of participants grouped by cognitive function are shown in Table 1. We found significant differences between the different cognitive function groups in terms of gender, age, number of falls, nutritional status, depressive symptoms, anxiety symptoms, SRH, and hearing function (p < 0.05), whereas the differences were not statistically significant regarding ethnicity, education level, living alone, monthly household income, smoking history, drinking history, and IADL condition (p > 0.05). A comparison of the MMSE scores of older adults with diverse characteristics revealed that differences in MMSE scores were statistically significant in terms of age, education level, living alone, monthly household income, nutritional status, number of falls in the previous year, hearing function, depressive symptoms, anxiety symptoms, and SRH (p < 0.05), whereas the differences were not statistically significant regarding gender, ethnicity, smoking history, drinking history, and IADL condition (p > 0.05) (Supplementary Table S1).

Descriptive statistics and Spearman’s correlations for the main study variables are shown in Table 2. Hearing loss was found to have a positive connection with anxiety (r = 0.167, p < 0.01) and depression (r = 0.195, p < 0.01). SRH (r = −0.350, p < 0.01) and cognitive function (r = −0.164, p < 0.01) were both adversely linked with hearing loss.

The hypothesized models were tested for goodness-of-fit, and the results of the goodness-of-fit tests for the serial mediation model 1 and serial mediation model 2 were consistent with the following results: χ²/df = 2.662, GFI = 0.979, AGFI = 0.958, CFI = 0.967, NFI = 0.949, TLI = 0.948, and RMSEA = 0.052, all of which met their respective criteria. Table 3 shows the standardized estimates of the paths in each serial mediation model. The results of the path analyses showed that hearing loss had a negative effect on SRH, cognitive function, and a positive effect on depressive anxiety symptoms. SRH had a negative effect on depressive anxiety symptoms (See Figure 2 for serial mediation model 1 and Figure 3 for serial mediation model 2).

The study examined the relationship between SRH, depressive anxiety symptoms, and cognitive function by analyzing two different causal orders. Table 4 shows the bootstrap results for indirect effects. The results of the serial mediation analysis showed that the total indirect effect of SRH and depressive anxiety symptoms was −0.115 (95% CI: −0.168 to −0.070), and the total effect of the model was −0.221 (95% CI: −0.307 to −0.132), with the total indirect effect accounting for 52.04% of the total effect of the model. In serial mediation model 1, the indirect effect path (HL → SRH → DAS → CF) had an indirect effect of −0.029 (SE = 0.013, 95%: −0.058 to −0.007). In the serial mediation model 2, the indirect effect paths (HL → DAS → SRH → CF) had an indirect effect of −0.016 (SE = 0.006, 95%: −0.030 to −0.005). In the relationship between hearing loss and cognitive functions, SRH and depressive anxiety symptoms co-played a serial mediating role. This suggests that loss of hearing function leads to an increased risk of poor SRH or depressive anxiety symptoms, which in turn leads to a decline in cognitive function.

First, the alternative modeling approach. The independent mediating roles of SRH, depression, and anxiety between hearing loss and cognitive functioning were tested, controlling for gender, age, number of falls, and nutritional status variables. The results showed that the mediating effects of SRH, depression, and anxiety were all held after controlling for confounders (see Supplementary Table S2 and Figure 1). Second, the sample composition was changed. The model was retested using the new sample after filling in missing values (sample size = 634) and the male sample (sample size = 268). The results showed that the model fit was more favorable to the data, and the results of the mediation analyses were generally consistent with the original model (see Supplementary Table S3).

Our findings showed that hearing loss significantly and negatively affected cognitive function in older adults. We also found that 33.7% of older adults with hearing loss suffered from cognitive impairment. This result indicates that hearing loss poses a serious threat to the cognitive health of older adults. Multiple hypotheses have been used to explain the mechanisms linking hearing loss and cognitive decline. For example, the common etiology hypothesis suggests that the association between hearing loss and cognitive decline may be due to common neurodegenerative disorders and there is no clear causal relationship between the two (Uchida et al., 2019). According to the sensory deprivation hypothesis, auditory decline leads to a reduction in the transmission of auditory information to the brain and a long-term lack of a sufficient amount of perceptual stimulation, causing atrophy in areas of the auditory system, resulting in the onset of cognitive decline (Slade et al., 2020). According to the information degradation hypothesis, patients with hearing loss tend to listen hard, which can overconsume the brain’s cognitive resources to offset the absence of auditory information, and in the long term, cognitive resources will be over-utilized, leading to cognitive decline (Powell et al., 2021). Although the exact underlying mechanism between hearing loss and cognitive decline remains unclear, current research suggests that restoring auditory input improves auditory neural function and cognitive function in hearing loss patients (Karawani et al., 2018). Hearing loss interventions can prevent or delay 8% of dementia (Livingston et al., 2020). Therefore, the management of hearing in older adults should be enhanced to improve hearing function and reduce cognitive deterioration and other adverse effects.

To the best of our knowledge, few studies to date have used SRH as a mediator to investigate the relationship between hearing loss and cognitive function. Our study demonstrated that hearing loss can indirectly affect cognitive function in older adults through SRH. Although there is a high prevalence of poor SRH in older adults with hearing loss, the underlying mechanisms remain unclear. Most studies support that the nature of hearing loss (congenital or acquired) and the degree to which hearing loss causes limitations in daily activities affect SRH (Anderle et al., 2023). Hearing loss in older adults not only affects their daily communication and interactions but also leads to other health problems such as impaired lower limb function, frailty syndrome, and IADL disability (Yévenes-Briones et al., 2021). Thus, older people with hearing loss may perceive significant changes in their lifestyle, activities, and social life due to physical limitations, and they may be aware of their health decline compared to their past or peers, thus making a poor evaluation of their health (Pinto et al., 2016). Hogan et al. (2015) argues that when an individual’s coping mechanisms are not well-suited to the demands of their environment, they are more likely to experience increased stress and diminished well-being, ultimately leading to a lower SRH (Hogan et al., 2015). Whitmore et al.’s (2023) study suggests that SRH may initially decrease when individuals experience a disease adversity, such as a new diagnosis. However, as resources are activated and older adults recover or adapt to adversity, SRH may return to its original levels or even increase Furthermore, Whitmore’s study indicates that individuals with congenital hearing impairment are not necessarily associated with poor SRH. This is because they develop alternative communication mechanisms apart from speaking and hearing. Moreover, after receiving hearing aids and/or rehabilitation, they develop coping strategies, such as sign language and oral-facial reading, to integrate into the community (Anderle et al., 2023). All of this evidence suggests that individuals who are unable to cope with the changes brought about by hearing loss and lack appropriate coping strategies, such as seeking professional help or using hearing aids, are more likely to experience poor SRH. Working memory refers to an individual’s ability to temporarily store and manipulate information while performing a cognitive task, and a poor SRH is associated with poorer overall cognition and working memory, which leads to decreased cognitive function (Aguiñaga et al., 2023). Thus, more emphasis should be placed on improving the poor SRH in older adults, which could be a novel approach to preventing or delaying cognitive decline.

Results from our study imply that older adults suffering from hearing loss are more likely to experience depression and anxiety, which raises the likelihood of cognitive deterioration. Hearing loss has historically been related to poor mental health (Bigelow et al., 2020). 1hearing loss and depressive anxiety symptoms. For instance, reduced auditory input leads to dysfunctional emotional processing circuits that are dysfunctional key limbic structures responsible for emotion and behavior, which can lead to impaired perception and misclassification of emotional responses (Zinchenko et al., 2018). Decreased amygdala and hippocampal responsivity to emotional sounds in patients with hearing loss is another potential mechanism, which can contribute to the occurrence of depression (Husain et al., 2014). Disrupted connections in the amygdala also exacerbate anxiety associated with hearing loss (Tang et al., 2020). Behavioral mechanisms have also been used to explain the relationship between hearing loss and depressive anxiety symptoms. Due to the negative impacts of hearing loss, such as social isolation, loneliness, and limited mobility, it can raise the risk of depressive anxiety symptoms (Sharma et al., 2021). Evidence suggests that anxiety and depression have severely adverse consequences on the transient memory domain of cognitive capability in patients with hearing loss, thereby accelerating cognitive decline (Andries et al., 2023). Additionally, anxiety and depression accelerate the rate of atrophy in the frontal and temporal lobes of the brain, respectively, facilitating the progression of MCI toward Alzheimer’s disease (Mah et al., 2015; Sacuiu et al., 2016). Thus, reducing anxiety and depressive symptoms in older adults with hearing loss possibly protects their cognitive function to some extent, which warrants further exploration in the future.

A series of mediating effects of SRH and depressive anxiety symptoms provide new perspectives on the mechanisms through which hearing loss affects cognitive function. In serial mediation model 1, hearing loss initially caused poor SRH and subsequently increased the risk of depressive anxiety symptoms, which may have led to a more rapid cognitive decline. In addition to the high burden of hearing loss and its adverse influence on the quality of life, hearing loss affects patients’ SRH and psychosocial (Jayakody et al., 2022). Patients with hearing loss may have poor SRH due to limited social participation and disability in daily activities in the absence of hearing aids (Anderle et al., 2021). A Brazilian study found that the odds of reporting poor SRH in peers who perceived hearing loss as a health problem were 3.72 times higher compared with peers who failed to perceive hearing loss as a health problem (Guia et al., 2018). Poor SRH can lead to a lack of perceived self-worth and meaning in an individual’s life (Vogel et al., 2021). Therefore, patients with hearing loss are more likely to exhibit ‘disengaged coping’ to escape the stress and exhaustion related to social interactions, as well as the embarrassment of displaying hearing difficulties before other people, which implies avoiding addressing hearing loss by opting out or withdrawing, for instance, withdrawing from social gatherings or pretending to hear during conversations (Heffernan et al., 2016). This negative coping can exacerbate mental tension, undermine mental health, and lead to the onset of anxiety and depression. Depression and anxiety are clinical markers that can help identify early signs of cognitive decline (Perin et al., 2022).

In serial mediation model 2, hearing loss initially leads to depressive anxiety symptoms and subsequently increases the risk of poor SRH, which can lead to a decline in cognitive function. If individuals with hearing loss adopt appropriate coping mechanisms (e.g., care-seeking behaviors, hearing rehabilitation training, wearing hearing aids, and seeking social support), the negative psychosocial and health consequences of hearing loss can be mitigated (Wells et al., 2020). In this case, hearing loss patients have a low risk of depression and anxiety and tend to make a positive assessment of their health. Moreover, individuals with good SRH perform better in executive function, working memory, and global cognition than individuals with poor or fair SRH (Aguiñaga et al., 2023). This study found that SRH and depressive anxiety symptoms are interrelated, regardless of the direction in which they flow between hearing loss and cognitive function, both of which have a serial mediation effect, shedding light on the mechanisms through which hearing loss impacts cognition. These findings suggest that older individuals should prioritize not only the management of hearing loss but also the management of SRH and mental health.

This study innovatively used SRH and depressive anxiety symptoms as mediating variables and confirmed that both SRH and depressive anxiety symptoms partially mediated the relationship between hearing loss and cognitive function in community-dwelling older adults. These findings may help identify people at high risk of cognitive impairment and ultimately prevent and treat this disease through the management of SRH and depressive anxiety symptoms in older adults with hearing loss. According to the data, the prevalence of hearing loss among Chinese adults over 60 years old is 58.85%, but the rate of hearing aid acquisition is only 6.5% (Gong et al., 2018), which is lower than the 10% in Japan and 17.4% in South Korea (Moon et al., 2015; Sugiura et al., 2022). Reliable evidence suggests that the use of hearing aid devices in patients with hearing loss reduces the risk of cognitive decline by 19% (Yeo et al., 2023). In this study, only 3.61% of older adults with hearing loss had access to hearing aids, well below the national level and in other Asian countries. We suggest that older adults with hearing loss should be promptly identified and managed to improve the utilization of hearing aid devices. Future studies should also develop cognitive function assessment tools that are more applicable to older adults with hearing loss. Community and medical personnel should regularly assess SRH and the mental health of older adults, especially those with hearing loss, which may facilitate the early detection of cognitive problems. Previous research found that the number and quality of social networks can buffer the adverse impacts of poor SRH on mental health (Windsor et al., 2016), and therefore older adults should be encouraged to engage in more social group activities regularly to build a good social support system.

Although this study has some theoretical and practical implications, some limitations exist. First, this cross-sectional study did not fully elucidate the causal relationships among hearing loss, SRH, depressive anxiety symptoms, and cognitive function. Longitudinal studies with large sample sizes are needed to comprehensively explore the causal relationships among the investigated variables. Second, hearing loss was determined by self-report in this study. Although self-reported hearing loss is a brief and highly effective indicator in epidemiological studies, there is a need to jointly use self-reported hearing loss with hearing status measured by pure-tone audiometry for screening and research (Louw et al., 2018). Finally, the study variables in this study were surveyed by questionnaire and may be subject to reporting bias.