Social participation trajectories and self-reported hearing impairment among Chinese older adults: Evidence from the Chinese longitudinal healthy longevity study over 10 years

Abstract

Social participation is an important aspect of healthy aging, yet its association with hearing impairment among older adults remains insufficiently explored. This study aimed to examine the social participation trajectories of Chinese older adults and their association with hearing impairment over 10 years. The study extracted 2154 Chinese older adults from the China Longitudinal Healthy Living Survey conducted between 2008 and 2018. The social participation and hearing impairment data used in this study were self-reported. We identified social participation trajectories using a group-based trajectory model and explored their association with hearing impairment using Cox hazard proportional regression. Subgroup analysis and interaction tests were further used to examine the potential influence of categorical covariates on the relationship between social participation trajectories and hearing impairment. We also performed 3 sensitivity analyses to determine the robustness of the results. Three social participation trajectory groups were identified: low-activity (N = 488, 22.66%), moderate-activity (N = 1298, 60.26%), and high-activity (N = 368, 17.08%). The median age of the participants was 73.00 (P₂₅: 69.00, P₇₅: 80.00) years. We documented that the incidence of hearing impairment was 20.84% during ten years of follow-up. After adjusting for covariates, we found that compared with the low-activity group, the hazard ratio of the moderate-activity was 0.54 (95% confidence intervals: 0.40–0.73), whereas that of the high-activity group was 0.17 (95% confidence intervals: 0.09–0.32). Interaction analyses showed that family annual income was a significant moderator (P for interaction <.05). The results of our study remained robust in 3 sensitivity analyses. Social participation in older adults can be classified into 3 distinct trajectory groups. Compared with the low-activity social participation trajectory, older Chinese adults in the moderate-activity and high-activity groups were associated with a lower risk of hearing impairment. Community-based strategies need to be developed to encourage older adults to maintain their social participation, which may help support hearing health.

1. Introduction

Aging is a global phenomenon with profound ramifications for societies. The proportion of the world’s population aged 65 years and over was 9.3% in 2020, expected to almost double by 2050.^[1] As the largest developing country and the world’s second-largest economy, China has emerged as the fastest aging rate in the world.^[2] One of the most significant challenges associated with an aging population is hearing impairment, and the severity of it increases with age.^[3] Hearing impairment affects more than 360 million people worldwide.^[4] At the personal level, hearing impairment is associated with cognitive decline^[5] and disability.^[6] At the socioeconomic level, hearing impairment increases the burden of health expenditure and healthcare system utilization.^[7] Given the heavy social and economic burden on individuals, families, communities, and countries of hearing impairment, it is imperative to discern the associated risk factors.

In the context of the hearing impairment burden among older adults, their health and well-being can benefit from behavioral interventions. Previous studies have explored behavioral risk factors of older adults’ hearing impairment, such as persistently short or long sleep duration^[8] and poor physical activity.^[9] In addition, as an important behavioral intervention to promote active healthy aging, social participation has attracted much attention.^[10] Based on the social relationship framework proposed by Kelly,^[11] social participation is also referred to as participation in social activities, which means engaging in life situations at the social level, different from social support and social networks.^[12] Social participation is not limited to activities that involve direct interaction with others but also includes a wide range of behaviors that individuals engage in during their social lives, even when done alone.^[13] However, older adults might not undergo the same patterns in the changes of social participation. Recent studies have increasingly adopted person-centered analytical approaches, such as latent class analysis, to capture the heterogeneity of social participation among older adults, identifying 4 distinct patterns, and these patterns may show differential associations with subsequent health outcomes, such as cognitive impairment.^[14] Therefore, we use group-based trajectory modeling to identify distinct longitudinal patterns of social participation, classifying participants into subgroups based on mutually exclusive characteristics and individual responses or patterns, thereby addressing the inherent heterogeneity present in the older population.

Social participation is hypothesized to impact health outcomes through 2 distinct mechanisms: direct effects and buffering effects.^[15] First, social participation co-occurs with physical activity, which has a direct benefit to individuals’ psychological well-being and physical health.^[16] Second, as a buffering effect, social participation can enhance social capital at the individual level, including access to additional resources and individual networks, thereby enhancing suitable coping mechanisms for stressful events and health-promotion mechanisms.^[17] The protective effects of high social participation on activities of daily living (ADL),^[18] cognitive function,^[19] and oral health^[15] have been proven in several studies. The association between hearing impairment and social participation has also been extensively explored in many studies, often focusing on hearing impairment as a predictor of social participation.^[20] Recent research has also expanded this framework by linking hearing-related participation restrictions to broader aging syndromes, such as physical, cognitive, and social frailty.^[21] Hearing impairment can lead to difficulties in communicating effectively in a diverse range of social environments, thereby reducing the frequency and willingness of individuals to participate in social activities.^[22] There are also studies showing that self-reported hearing loss is not associated with social participation, possibly because people may be able to understand each other through context or mutual understanding when talking about familiar topics with familiar friends, which can reduce an individual’s need to hear each word accurately.^[23] While the above literature explains social participation more in terms of social interactions with others, many studies incorporate activities done independently by individuals in constructing the variable of social participation.^[24] For example, activities such as reading or watching television do not rely primarily on auditory input, especially if captioning or visual aids are available. Even activities such as playing cards or mahjong, although sometimes performed in a group, do not necessarily require participants to clearly hear all verbal exchanges, as visual cues, gestures, and familiarity with the rules of the game provide sufficient context for participation. Given the broad nature of the definition of social participation, it should be considered that social participation may not only be a consequence of hearing impairment but may, in turn, play a role in the onset and development of hearing impairment. Consistent with this possibility, a large prospective cohort study from the UK Biobank reported that loneliness – a marker of reduced social participation – was associated with a higher risk of new-onset hearing loss.^[25] However, studies focusing on the association between social participation and hearing impairment among the Chinese older population are still lacking.

Therefore, the first aim of this study was to investigate the dynamic trajectories of social participation using nationally representative longitudinal data. The second aim was to provide a new perspective on the association between social behaviors and auditory health, identifying the association between social participation trajectories and hearing impairment in the specific context of China to promote an active aging process.

2. Methods

2.1. Study population

The data utilized in this study was derived from the Chinese longitudinal survey of healthy longevity (CLHLS, 1998–2018). To balance the age and sex of older adults, a multi-stage cluster random sampling method was adopted to survey the older adults who were ≥65 years old.^[26,27] Due to the large attritions in the follow-up waves resulting from the high mortality among the oldest old, the CLHLS were oversampled in their 80s, 90s, and 100s. Furthermore, half of the counties and city districts were chosen from 23 out of 31 provincial administrative regions in mainland China by adopting the random selection method. The sample of these provincial administrative regions accounts for almost 85% of the country’s total population.

The CLHLS is ongoing prospective, community-based cohort research focusing on demographic characteristics, socioeconomic conditions, lifestyle, personal health, psychological state, and cognitive function. The data was collected by the trained interviewers with structured questionnaires through face-to-face surveys. The CLHLS’s nonresponse rate was very low, and the data attrition average per wave was 4.85%.^[28] The CLHLS study was approved by the Biomedical Ethics Committee of Peking University (IRB00001052-13074). All participants included in the study signed a written informed consent form. Detailed information about the study design and the CLHLS’s data quality has been published.^[29]

This study abstracted the data of the last 4 waves of datasets, including the 2008 (baseline), 2011, 2014, and 2018 waves. There were 16,954 participants at baseline. First, we excluded participants under the age of 65 in the 2008 wave (N = 391). Second, based on the fitting requirements of the group-based trajectory modeling used in this study, we excluded individuals who completed fewer than 3 social participation surveys across the 4 waves (N = 11971). Third, participants with missing hearing impairment information from any wave (N = 2730) and participants with hearing impairment at baseline (N = 98) were excluded. Finally, a total of 2154 participants were included in this study (Fig. 1).

Figure 1.

The flow chart of sample selection in the study.

2.2. Measurements

2.2.1. Social participation

According to previous studies,^[30,31] social participation in this study refers to a broad range of activities involving engagement in life situations at the social level. The selected activities include “reading newspapers/books,” “playing cards/mahjong,” “watching TV/listening to the radio,” “housework,” “gardening,” “raising pets,” and “taking part in any social activity.” Answers ranged from 1 (never), 2 (not monthly, but sometimes), 3 (not weekly, but at least once a month), 4 (not every day, but at least once a week), and 5 (almost every day). The total score was 7 to 35, with higher scores representing higher activity of social participation.

2.2.2. Hearing impairment

Based on previous studies,^[8,32] hearing impairment was assessed using self-reported measures. Interviewees were categorized into 4 groups according to how well they could hear the questions asked by the interviewer. These groups included those who can hear and do not require a hearing aid, those who can hear but require a hearing aid, those who can partially hear and require a hearing aid, and those who cannot hear, even with a hearing aid. This categorization accounts for hearing aid use and its effect on hearing ability. Participants in the last 2 categories were determined as hearing impairment.

2.2.3. Covariates

Based on previous studies,^[33–35] we controlled for potential confounding covariates in the analysis, including demographic characteristics, socioeconomic conditions, lifestyle, personal health, psychological state, and cognitive function. The demographic characteristics included age, gender (male vs female), residential region (rural vs urban), ethnicity (Han vs other), and marital status (have a spouse vs have no spouse). The socioeconomic conditions included years of schooling (0 years, 1–6 years, and >6 years), family annual income (<30,000 yuan, 30,000–50,000 yuan, and >50,000 yuan), source of income (pension or salary vs other), and engaging in manual labor (yes vs no). Lifestyle variables included smoking status (yes vs no), alcohol consumption (yes vs no), and physical activity (i.e. purposeful fitness activities such as strolling, playing ball, running, and Chinese qigong) (yes vs no). Personal health included self-rated health status (1 = very good to 5 = very bad), ADL scores, and chronic multimorbidity (yes vs no) (see Supplementary Material, Supplemental Digital Content, https://links.lww.com/MD/R588).

2.3. Statistical analysis

First, we use group-based trajectory modeling to identify distinctive clusters of individual social participation trajectories and explore heterogeneity across groups. Using a general quasi-Newton procedure, group-based trajectory modeling predicted each group’s trajectory, estimated the probability that an individual is within each group, and assigned an individual to the group that they have the highest probability of being in. Many factors were taken into consideration when selecting the model, such as statistical measures (the p-values of model parameters and the confidence intervals (CI) of trajectory estimates) and visual inspection of predicted trajectories. The fitting process starts with fewer groups and higher-order functions for each subgroup. If the P-value or BIC of the high-order parameters model does not statistically significantly, remove them and continue fitting lower-order parameters. Repeat the order iteration until the P-value of the maximum order of all groups is <.05. Then, the optimal number of trajectories for this study was determined using the Bayesian information criterion (BIC) and average posterior probability (AvePP), where a smaller BIC implies a better model fit, and AvePP >0.7 indicates an acceptable model.^[36]

Second, we used descriptive variables according to trajectory group membership. The Kolmogorov–Smirnov test and Q–Q plots assessed the data for normal distribution, indicating that the continuous variables did not follow a normal distribution. Values are presented in median (P₂₅, P₇₅) for continuous variables or frequency (percentage) for categorical variables. To compare differences between groups, the Kruskal–Wallis test was used for continuous variables, and the chi-square test was used to compare categorical variables. In addition to statistical significance, effect sizes were further estimated to check for actual significance. For continuous variables, the effect size was reported as eta-squared η2 and estimated by the following criteria: a “very small” effect size of η² <0.01, a “small” effect size of 0.01 ≤η² <0.06, a “medium” effect size of 0.06 ≤η² <0.14, and a “large” effect size of η² ≥0.14. For categorical variables, the effect size was reported as Cramer V and estimated by the following criteria: a “trivial” effect size is Cramer V <0.10, a “small” effect size is 0.10 ≤Cramer V <0.30, a “medium” effect size is 0.30 ≤Cramer V <0.50, and a “large” effect size is Cramer V ≥0.50.^[37]

Third, using the extended Cox proportional hazards model, we estimate the hazard ratios (HRs) and 95% CI for hearing impairment by assigned social participation trajectories. Self-rated health status, depressive symptoms, and cognitive function were included as time-dependent covariates. Kaplan–Meier survival curves were used to examine the associations between social participation trajectories and hearing impairment. The statistical significance was calculated by the “Log-rank” test statistic. In the analysis, 3 models were employed: Model 1 was unadjusted; Model 2 was adjusted for age, gender, residential region, ethnicity, marital status, and years of schooling; Model 3 was adjusted for age, gender, residential region, ethnicity, marital status, years of schooling, family annual income, source of income, engaging in manual labor, smoking status, alcohol consumption, physical activity, sleep duration, self-rated health status, ADL scores, and depressive symptoms. To investigate potential effect modification, we performed a subgroup analysis and interaction test to explore the association between social participation trajectories and hearing impairment across different subgroups. All presented covariates in Model 3 were adjusted for, except for the stratified variable.

Finally, 2 distinct sensitivity studies were conducted to confirm the robustness of the results. In the first sensitivity analysis, missing data were imputed using multiple imputations by chained equations, and the imputed data were then used for analysis. In the second sensitivity analysis, we calculated the E-value to assess potential unmeasured confounding.

A 2-tailed P-value <.05 was considered to indicate a statistically significant difference. All analyses were performed with STATA version 16.0 (Stata Corp, College Station) and R statistical software.

3. Results

3.1. Social participation trajectory modeling

The fitting procedure commenced with the high-order single group and progressed towards the low-order multi-group, as outlined in Table 1. By comparing the BIC coefficient, we selected the fourth trajectories model. The AvePP values are all more than 0.8, and the proportion of subgroups within the group exceeded 5%.

Table 1.

Group-based trajectory modeling fitting information.

No	Order	BIC	AIC	aBIC	Proportion (%)	AvePP (%)
1	2	−25,917.19	−25,905.84	−25,919.95	100.00	100
2	2, 2	−25,288.35	−25,265.65	−25,293.87	62.77/37.23	91.19/87.35
3	2, 2, 2	−25,153.85	−25,119.80	−25,162.13	24.47/59.05/16.48	82.72/83.72/83.63
4	1, 2, 2	−25,151.26	−25,120.05	−25,158.85	22.66/60.26/17.08	83.28/83.83/83.97
5	1, 1, 2, 2	−25,153.20	−25,113.48	−25,162.86	13.97/36.35/41.23/8.45	76.39/67.15/75.16/83.00

aBIC = adjusted bayesian information criterion, AIC = Akaike information criterion, AvePP = average posterior probability, BIC = Bayesian information criterion.

Three distinct trajectories of social participation among Chinese older adults were identified (Fig. 2). The first trajectory group comprised 22.66% of the sample (N = 488) and was labeled “Low-activity,” with the lowest score and a consistent downward trend throughout the follow-up period. The second trajectory group comprised 60.26% (N = 1298) of the participants and was labeled as having “Moderate-activity” characteristics as a slightly rising trend from the first follow-up to the second follow-up and then showed a descending trend from the second follow-up until the final follow-up. The third trajectory group comprised 17.08% (N = 368) of the sample and was labeled as “High-activity,” with the highest score and characteristics as scores first significantly increased, then level off finally decreased to near approach initial level.

Figure 2.

Trajectories of social participation. The solid lines mean estimated values, and the dotted lines display the 95% CIs. CI = confidence intervals.

3.2. Baseline characteristics

The characteristics at baseline of the participants are shown in Table 2. The median age of the participants was 73.00 (P₂₅: 69.00, P₇₅: 80.00) years, and 53.34% of the participants were female. The proportion of individuals with hearing impairment was 20.84%. Compared to the low-activity group, the moderate-activity group exhibited a slightly lower median age of 73.00 (P₂₅: 69.00, P₇₅: 78.00), while the high-activity group had a median age of 70.00 (P₂₅: 67.00, P₇₅: 74.00). Age showed a statistically significant difference and a large effect size (η² = 0.152) between the 3 trajectory groups. Statistically significant differences and medium effect sizes were observed for ADL scores (η² = 0.116) and hearing impairment (Cramér’s V = 0.350). However, although statistically significant differences were found for self-rated health status (η² = 0.007), depressive symptoms (η² = 0.008), and cognitive function (η² = 0.023), gender (Cramér’s V = 0.160), residential region (Cramér’s V = 0.186), ethnicity (Cramér’s V = 0.064), marital status (Cramér’s V = 0.250), years of schooling (Cramér’s V = 0.256), family annual income (Cramér’s V = 0.088), source of income (Cramér’s V = 0.254), engaging in manual labor (Cramér’s V = 0.085), smoking status (Cramér’s V = 0.105), alcohol consumption (Cramér’s V = 0.115), physical activity (Cramér’s V = 0.165), the effect sizes were either small or very small, indicating their practical significance was limited. There were no statistically significant differences in sleep duration (η² = 0.001) and chronic multimorbidity (Cramér’s V = 0.047) (all P-value <.05). Table S1, Supplemental Digital Content, https://links.lww.com/MD/R588 presents the comparison of baseline characteristics between participants included and excluded (2154 of 14,800). The results show that during sample screening, while age (η² = 0.144) and ADL scores (η² = 0.111) may be moderately and above affected, most of the variables remained relatively balanced, i.e., did not have practically significant differences. It should be noted that excluded participants tended to be older and have poorer functional status, which may indicate that the analytical sample is more representative of relatively younger and functionally healthier older adults.

Table 2.

Baseline characteristics of the total sample and the sample by the different trajectory groups.

Characteristics	Social participation trajectories				P-value	Effect size
	Total (N = 2154)	Low-activity (N = 488)	Moderate-activity (N = 1298)	High-activity (N = 368)
Age (year), median (P25, P75)	73.00 (69.00, 80.00)	80.00 (75.00, 85.00)	73.00 (69.00, 78.00)	70.00 (67.00, 74.00)	<.001	0.152
Sleep duration, median (P25, P75)	8.00 (6.00, 9.00)	8.00 (6.00, 9.00)	8.00 (6.00, 8.00)	8.00 (6.00, 8.00)	.189	0.001
Self-rated health status, median (P25, P75)	2.00 (2.00, 3.00)	2.00 (2.00, 3.00)	2.00 (2.00, 3.00)	2.00 (2.00, 3.00)	<.001	0.007
ADL scores, median (P25, P75)	14.00 (14.00, 15.00)	15.00 (14.00, 18.00)	14.00 (14.00, 14.00)	14.00 (14.00, 14.00)	<.001	0.116
Depressive symptoms, median (P25, P75)	12.00 (11.00, 13.00)	12.00 (10.00, 13.00)	12.00 (11.00, 13.00)	12.00 (11.00, 14.00)	<.001	0.008
Cognitive function, median (P25, P75)	29.00 (27.00, 30.00)	28.00 (13.00, 30.00)	29.00 (9.00, 30.00)	29.00 (21.00, 30.00)	<.001	0.023
Hearing impairment, n (%)
Yes	449 (20.84)	223 (45.70)	212 (16.33)	14 (3.80)	<.001	0.350
No	1705 (79.16)	265 (54.30)	1086 (83.67)	354 (96.20)
Gender, n (%)
Male	1005 (46.66)	170 (34.84)	613 (47.23)	222 (60.33)	<.001	0.160
Female	1149 (53.34)	318 (65.16)	685 (52.77)	146 (39.67)
Residential region, n (%)
Rural	1882 (87.37)	448 (91.80)	1162 (89.52)	272 (73.91)	<.001	0.186
Urban	272 (12.63)	40 (8.20)	136 (10.48)	96 (26.09)
Ethnicity, n (%)
Han	2015 (93.55)	448 (91.80)	1211 (93.30)	356 (96.74)	.012	0.064
Other	139 (6.45)	40 (8.20)	87 (6.70)	12 (3.26)
Marital status, n (%)
Have a spouse	1332 (61.84)	202 (41.39)	841 (64.79)	289 (78.53)	<.001	0.250
Have no spouse	822 (38.16)	286 (58.61)	457 (35.21)	79 (21.47)
Years of schooling, n (%)
0 years	1023 (47.49)	328 (67.21)	625 (48.23)	70 (19.02)	<.001	0.256
1–6 years	833 (38.67)	141 (28.89)	523 (40.35)	169 (45.92)
>6 years	296 (13.74)	19 (3.89)	148 (11.42)	129 (35.05)
Family annual income (yuan), n (%)
<30,000	1718 (79.76)	411 (84.22)	1044 (80.43)	263 (71.47)	<.001	0.088
30,000–50,000	235 (10.91)	33 (6.76)	133 (10.25)	69 (18.75)
>50,000	201 (9.33)	44 (9.02)	121 (9.32)	36 (9.78)
Source of income, n (%)
Pension or salary	919 (42.66)	116 (23.77)	568 (43.76)	235 (63.86)	<.001	0.254
Other	1235 (57.34)	372 (76.23)	730 (56.24)	133 (36.14)
Engaging in manual labor, n (%)
Yes	1879 (87.35)	443 (90.78)	1145 (88.42)	291 (79.08)	<.001	0.085
No	272 (12.65)	45 (9.22)	150 (11.58)	77 (20.92)
Smoking status, n (%)
Yes	483 (22.42)	79 (16.19)	293 (22.57)	111 (30.16)	<.001	0.105
No	1671 (77.58)	409 (83.81)	1005 (77.43)	257 (69.84)
Alcohol consumption, n (%)
Yes	480 (22.28)	73 (14.96)	296 (22.80)	111 (30.16)	<.001	0.115
No	1674 (77.72)	415 (85.04)	1002 (77.20)	257 (69.84)
Physical activity, n (%)
Yes	766 (35.56)	131 (26.84)	445 (34.28)	190 (51.63)	<.001	0.165
No	1388 (64.44)	357 (73.16)	853 (65.72)	178 (48.37)
Chronic multimorbidity, n (%)
Yes	438 (20.33)	94 (19.26)	254 (19.57)	90 (24.46)	.097	0.047
No	1716 (79.67)	394 (80.74)	1044 (80.43)	278 (75.54)

ADL = activities of daily living.

3.3. Social participation trajectories and hearing impairment

Table 3 shows the associations between the social participation trajectories and hearing impairment by 4 Cox proportional hazards models. The HRs were significant in all the models (all P-values <.05). In the unadjusted model (Model 1), compared with the low-activity group, the HRs (95% CI) for hearing impairment was 0.31 (0.25, 0.37) for the moderate-activity group, and 0.07 (0.04, 0.12) for the high-activity group. After adjusting for age, gender, residential region, ethnicity, marital status, and years of schooling in Model 2, compared with the low-activity group, the HRs (95%CI) for hearing impairment was 0.49 (0.40, 0.60) for the moderate-activity group, and 0.13 (0.07, 0.23) for the high-activity group. After further adjustment for all variables in Model 3, the result was still statistically significant. The HRs for the moderate-activity and high-activity groups compared with the low-activity groups were 0.52 (0.27, 0.99) and 0.32 (0.11, 0.93), respectively.

Table 3.

Association between social participation trajectories and hearing impairment.

Model	Low-activity	Moderate-activity		High-activity
	HRs	HRs (95% CI)	P-value	HRs (95% CI)	P-value
Model 1	1.00	0.31 (0.25–0.37)	<.001	0.07 (0.04–0.12)	<.001
Model 2	1.00	0.49 (0.40–0.60)	<.001	0.13 (0.07–0.23)	<.001
Model 3	1.00	0.52 (0.27–0.99)	.049	0.32 (0.11–0.93)	.037

Model 1 was unadjusted.

Model 2 was adjusted for age, gender, residential region, ethnicity, marital status, and years of schooling.

Model 3 was adjusted for age, gender, residential region, ethnicity, marital status, years of schooling, family annual income, source of income, engaging in manual labor, smoking status, alcohol consumption, physical activity, sleep duration, self-rated health status, ADL scores, chronic multimorbidity, depressive symptoms, and cognitive function.

ADL = activities of daily living, CI = confidence intervals, HRs = hazard ratios.

Kaplan–Meier survival curves showed the cumulative risk of incident hearing impairment was markedly different among the social participation trajectories (log-rank test P <.001), with the minimum risk of hearing impairment in the high-activity group (Fig. S1, Supplemental Digital Content, https://links.lww.com/MD/R588). The risk table showed that over the follow-up period, the number of participants at risk decreased more sharply in the low-activity group compared to the moderate-activity and high-activity groups. This suggests a higher incidence of hearing impairment in the low-activity group, which aligns with the Kaplan–Meier survival curves.

3.4. Stratified and interaction analyses

The results of the subgroup analyses for 12 categorical variables are shown in Table S2, Supplemental Digital Content, https://links.lww.com/MD/R588. The interaction analysis results showed that family annual income was a significant moderator (P for interaction = .002), and social participation trajectories were negatively associated with hearing impairment among the family annual income subgroup. Whether in the moderate-activity group or the high-activity group, social participation trajectories were more negatively associated with hearing impairment among participants with a family annual income of 30,000 to 50,000 yuan (N = 235) and >50,000 yuan (N = 201) than among those with a family annual income <30,000 yuan (N = 1718).

3.5. Sensitivity analyses

In the first sensitivity analysis, results were robust after multiple imputations of all missing covariates (all P <.05) (Table S3, Supplemental Digital Content, https://links.lww.com/MD/R588). In the second sensitivity analysis, we generated an E-value to assess the robustness of unmeasured confounding. The E-values (2.61 and 6.84) indicated that there was unlikely to be an unmeasured confounding affecting the association between social participation trajectories and hearing impairment, regardless in the moderate-activity group or the high-activity group.

4. Discussion

To the best of our knowledge, this was the first study to investigate the association of longitudinal social participation trajectories and hearing impairment among Chinese older adults. Compared with the low-activity group, moderate-activity and high-activity groups were associated with a lower risk of hearing impairment. These associations persisted after adjusting for numerous confounding variables. Moreover, stronger negative associations between social participation trajectories and hearing impairment were detected among individuals with higher family annual incomes.

This study has revealed 3 distinct trajectories of social participation, namely, low activity, moderate activity, and high activity. Previous trajectory-based studies have similarly identified multiple patterns of social participation among older adults, most of which demonstrated an overall declining trend over time.^[31,38] The social participation trajectories of older adults in all of these studies above tended to decline over time, which is different from our study. First, the consistently low and declining pattern observed in the low-activity group raises concerns regarding potential vulnerability to adverse health outcomes. Individuals in this trajectory may experience limited opportunities to maintain or expand social engagement over time, which may be linked to reduced access to emotional support and coping resources.^[39] Previous studies suggested that social isolation and loneliness, which may result from limited social participation, were associated with adverse health outcomes, including increased risk of chronic conditions and mortality.^[40] Second, although both the moderate-activity and high-activity groups exhibited relatively higher levels of social participation during earlier follow-up periods, declines were observed across all trajectories in later stages. The initially higher engagement in the high-activity group may be related to greater social resources, healthier lifestyles, and broader social networks accumulated earlier in life. These factors may have facilitated continued participation during the early stages of retirement.^[41] However, age-related health changes and life transitions may gradually constrain sustained engagement over time.^[42]

This study also identified an association between social participation trajectories and hearing impairment, broadening the understanding of the association between social participation and older adults’ health outcomes. Our results revealed that social participation trajectories were negatively associated with hearing impairment among the whole older population. One possible explanation is that the consistently declining low level of social participation means that the low-activity group may lack sufficient social interaction and support from an early stage, which may trigger feelings of loneliness and social isolation,^[43] thereby inducing or exacerbating negative emotional states such as depression and anxiety.^[44] Previous research has revealed the bidirectional association between depression and anxiety disorders and sensorineural hearing loss.^[45] Another possible explanation is that positive social participation expands social support networks, providing access to more resources and support, including health information and medical resources.^[46] By accessing important hearing health care information about hearing conservation, treatment, and management of hearing loss, individuals can adopt positive health behaviors to manage their hearing problems better, thereby preventing the onset of hearing impairment. Although social participation in the high-activity group declined in the later stages over time, this prior participation advantage may have had a lasting protective effect on their hearing health since they had established stronger social support networks and access to health resources in the early stages, resulting in a lower risk of hearing impairment. Considering the negative effects of hearing impairment, encouraging older adults to participate in social life is important and urgently needed.

The mechanisms linking social participation trajectories to hearing impairment remain incompletely understood; however, several plausible pathways can be hypothesized based on existing evidence. First, a high level of social participation may improve immune system functions.^[47] Immune surveillance processes are known to be related to inflammatory activity through the activation of resident cells and the infiltration of circulating immune cells into the central nervous system.^[48] Gerontological research has suggested that chronic, age-related systemic inflammation is associated with the progression of age-related hearing loss, which may be linked to a higher likelihood of hearing impairment.^[49] Second, social participation has been associated with lower sympathetic nervous system activity and reduced stress-related hormonal responses, as well as more favorable cardiovascular profiles, including lower cardiovascular stress and improved vascular function.^[50] Such physiological states have been linked to better cochlear microcirculation and lower levels of oxidative stress, which may be relevant to auditory neural integrity.^[51] Importantly, these pathways should be regarded as hypothetical and interrelated explanatory frameworks rather than empirically tested mechanisms in the present study. Further longitudinal and experimental research is needed to clarify whether and how these proposed pathways may underlie the observed associations between social participation trajectories and hearing impairment.

There was a stronger negative association between social participation trajectories and hearing impairment among individuals with higher family annual incomes. One explanation could be that high-income individuals may be able to access higher-quality healthcare resources and health management services. For example, they are more likely to learn and implement more aggressive tertiary prevention and can more afford regular hearing health screenings. Similarly, this group makes it more possible to take proactive measures to prevent hearing impairment and to get timely treatment when minor hearing problems occur.^[52] Furthermore, low-income individuals have relatively limited access to social engagement opportunities, whereas higher-income individuals often have the financial means to engage in more intellectually and socially stimulating social activities, which may enhance cognitive reserve and promote overall neuroplasticity, thereby improving auditory processing.^[53,54] This finding emphasizes the importance of effectively reducing income inequality to promote hearing health.

Some limitations in this study have to be elucidated. First, the social participation and hearing impairment data were self-reported, which may introduce biases, such as recall bias and social desirability bias. Second, due to the CLHLS database’s limited data collection, the definition of social participation may not encompass all relevant aspects, such as using the internet. On the other hand, we do not have access to detailed information about the specific cause and duration of the hearing impairment and information related to the use of hearing aids. Future studies should consider these variables to provide a more comprehensive understanding of hearing-related outcomes. Third, individuals may withdraw from some social activities before they are formally diagnosed or report any significant hearing impairment. This “silent withdrawal” suggests that the observed reduction in social participation may not fully reflect the causal effect of social participation trajectories on hearing impairment but rather is partially influenced by early, undetected hearing loss. Fourth, while the revised definition of social participation captures the potential for social interaction across numerous activities, it is significant to acknowledge that the level of social engagement may differ depending on individual preferences or contexts. Future research could explore more context-specific measures to better capture how different activities facilitate social connections in different social environments. Fifth, although the reported Cronbach α values for the depression symptoms and cognitive function were above acceptable thresholds, there are potential limitations to measurement precision. This may introduce a degree of measurement error, which may affect the strength of the associations observed in the results of this study. Sixth, because excluded participants were older and had poorer functional status, the study sample may underrepresent the most functionally impaired older adults, which could limit the generalizability of the findings. Finally, although the calculated E-values demonstrated that unmeasured confounding had minimal impact on the interpretation of the results, the potential for bias cannot be completely ruled out.

5. Conclusion

This study identified 3 categories of social participation trajectories among Chinese older adults: low activity, moderate activity, and high activity. Our results also indicated a negative association between social participation trajectories and hearing impairment among older adults, particularly those with higher family annual incomes. From a public health perspective, these results suggest that supporting and promoting social participation among older adults may be relevant for healthy aging. In particular, efforts to enhance access to social resources and opportunities – especially for socioeconomically disadvantaged older adults – may help promote more equitable hearing health outcomes in later life.

Author contributions

Data curation: Xuange Sun.

Formal analysis: Yutong Sui, Xu Liu, Xuange Sun.

Project administration: Xuange Sun.

Supervision: Xuange Sun.

Validation: Xuange Sun.

Writing – original draft: Yutong Sui, Xuange Sun.

Writing – review & editing: Yutong Sui, Xu Liu, Xuange Sun.

Supplementary Material

medi-105-e48151-s001.docx ^{(153.5KB, docx)}