Abstract
A growing body of evidence suggests that age-related hearing loss is related to changes in older adults’ memory. We test the hypothesis that the association is due to social disengagement following the onset of perceived hearing loss. At ages 65 (2004) and 72 years (2011), 3,986 participants from the Wisconsin Longitudinal Study (WLS) self-reported on hearing problems and several types of social engagement and completed three tests of memory. We estimated fixed effects regression models. Perceived hearing loss was related to significant decline in memory. Declines in frequency of in-person social contact were also associated with declining memory, but there was no evidence of a mechanism wherein reductions in social engagement explained the association between perceived hearing loss and memory decline. We conclude that self-reported hearing loss and social disengagement are likely independent risk factors for memory loss among older adults.
Keywords
Evidence has been building that age-related hearing loss is associated with decline in memory, as well as other cognitive functions such as processing speed, executive function, and visuospatial ability (Loughrey et al., 2018). Researchers have theorized reasons for the association including measurement error and common neuropathology, yet empirical study of potential mediators is in its nascence (Wayne & Johnsrude, 2015). Identification of mediators may inform the design of interventions to optimize both hearing ability and cognitive health in later life.
A long-standing proposition is that hearing impairment may jeopardize social relationships because it hinders effective communication (Strawbridge et al., 2000). Cross sectional research provides some foundation for this hypothesis; for example, hearing loss has been associated with social isolation among U.S. women in their 60s (Mick et al., 2014) and with loneliness among community-dwelling Canadian older adults (Mick et al., 2018). The evidence that social engagement benefits the maintenance of memory is stronger, with social disengagement being an established risk factor for cognitive impairment (Shankar et al., 2013).
Based on these research literatures, scholars have hypothesized that reductions in social engagement may partially account for associations between hearing loss and declining memory (Lin et al., 2013). One study has tested this hypothesis using structural equation analyses in a cross section of 164,770 members of the U.K. Biobank aged 40 to 69 years (Dawes et al., 2015). There was evidence that hearing loss was related to low cognitive performance and that independently, loneliness was also associated with low cognitive performance. However, loneliness did not explain the association between hearing and cognition.
We aim to expand upon the empirical evidence concerning social isolation as a potential mediator of the association between hearing loss and memory, using data from the Wisconsin Longitudinal Study (WLS). The longitudinal nature of these data allows for prospective measurement of the onset of perceived hearing loss, as well as change in memory over time. The WLS includes rich measures of various aspects of participants’ social lives, including perceived loneliness, extent of contact with friends and relatives, and presence of a close confidante. These measures allow for examining a variety of aspects of social relationships that could potentially account for associations between hearing loss and memory.
Method
Data
The WLS is a random sample of one third (N = 10,317) of graduates from Wisconsin high schools in 1957 (Herd et al., 2014). Participants gave verbal informed consent, and the institutional review board at the University of Wisconsin–Madison approved data collection (protocol number 2015-0955). The sample is non-Hispanic White, reflective of the population of Wisconsin at the time. Observations of memory, hearing, and social engagement were available at two time points for 3,986 participants.
Measures
Memory
At ages 65 (2004) and 72 years (2011), participants completed immediate and delayed word recall tests (Brandt et al., 1988) and a digit ordering test (Wechsler, 1997). We calculated the percentage of maximum possible score for each test and averaged the scores.
Perceived hearing
At both time points, hearing was measured using the Health Utilities Index (HUI) Mark 3 (Horsman et al., 2003). The test includes four self-reported, yes/no questions: During the past four weeks, have you been able to hear what is said . . . in a group conversation with at least three other people, without a hearing aid? In a group conversation with at least three other people, with a hearing aid? In a conversation with one other person in a quiet room, without a hearing aid? In a conversation with one other person in a quiet room, with a hearing aid?
We dichotomized the measure at each time point into those who had no perceived hearing problem and those who had any perceived hearing problem, regardless of hearing aid use and group setting.
Social functioning
There were four measures of social functioning at both time points. First, participants were asked, “Is there a person with whom you can really share your very private feelings and concerns?” Participants were categorized as having or not having a confidante. Second, participants answered two questions about loneliness which were standardized and averaged: “To what extent do you agree that you often feel lonely because you have few close friends with whom to share your concerns?” and “On how many days during the past week did you feel lonely?” Third, participants reported how many times during the past 4 weeks they had gotten together socially with (a) friends and (b) relatives (0 = never; 3 = more than twice a week). These two items were summed to create a measure of in-person social contact. Fourth, participants reported how many hours of telephone contact they had with friends and family in the past week. The strongest correlation among measures was between frequency of in-person and telephone contact (ρ = 0.21, p < .001).
Covariates
We controlled for self-reported health (0 = fair/poor; 1 = excellent/very good/good), cardiometabolic risk (0 = low risk; 1 = history of diabetes, heart problem, or stroke, or comorbid hypertension, hypercholesterolemia, and obesity), and depressive symptoms (Center for Epidemiological Studies-Depression scale; Radloff, 1977). In addition, other disabilities could confound the associations of interest by impairing social functioning. However, because these participants were so few (at baseline, four blind participants and 12 nonambulatory participants) and because only one of these 16 reported losing hearing function over the study period, we retained them in the analytic sample.
Analytic Strategy
Statistical approach
We estimated fixed effects regression models, which take the form
where Yij is memory score at wave i (i.e., age 65 or 72 years) for person j; α j captures the influence of unobserved factors related to person j that do not change over the 7-year period; Xijn denotes the values of observed factors for wave i for person j; and Yj and Xjn are person-level grand means that are subtracted from observation-specific values. A feature of this modeling strategy is that it subtracts away the effects of both observed and unobserved variables that do not change over time; that is, only change in a measure can account for change in memory. Model 1 included perceived hearing, as well as self-reported health, cardiometabolic risk, and depressive symptoms. Model 2 added the four measures of social isolation to test whether they reduced the size or significance of the association between perceived hearing and memory.
Study attrition and missing data
Correlates of attrition from the original 1957 sample of 10,317 included low educational attainment, poor health, and male sex (Herd et al., 2014). We used t tests to examine the available data for evidence of selective attrition with regard to perceived hearing and/or memory. Participants who left the study after 2004 were less likely than continuing participants to perceive hearing problems in 2004 (p < .01), but had poorer memory (p < .001) at age 65 years.
In addition, 1,698 active participants were excluded because they had incomplete memory tests at age 65 and/or age 72 years. Two-thirds of these incomplete scores were by design because the participant was in a random subsample that did not receive cognitive testing. We examined the information about memory that was available for the remaining third. At age 65, their memory scores did not differ from those of participants with complete scores, but at age 72 years, those with incomplete scores had poorer memory (p < .05) than those with complete scores.
Within the analytic sample of 3,986 participants, 79.95% were missing values on two variables or fewer. Telephone contact with friends and family at age 72 was the measure with the most missing data (19.07% of observations in 2011). We conducted multiple imputation by chained equations.
Results
Descriptive Statistics
Descriptive statistics for all study variables are presented in Table 1. At age 65 years, the average score on the tests of memory was 54.70% (measure ranges 0 to 100), and the average score had declined to 48.84% at age 72 (p < .001). Subjective hearing complaints in the analytic sample were uncommon, with 3.69% (n = 142) of participants reporting a problem at age 65, and 7.13% (n = 265) reporting a problem at age 72 years.
Average Scores for All Study Measures at Age 65 and 72 Years, Wisconsin Longitudinal Study, N = 3,986.
Note. Means and standard deviations shown for continuous measures; percentages shown for binary measures. All values reported prior to multiple imputation.
Fixed-Effects Models
The results of regression models are presented in Table 2. Model 1 indicated net of change in self-reported health, cardiometabolic risk, and depressive symptoms, people who acquired a perceived hearing problem experienced a significant decline in memory averaging 4.52 percentage points (p < .001). Model 2 showed that increasing one’s in-person social contact was associated with memory improvements: An increase of one standard deviation in face-to-face social contact was associated with an improvement in memory of 0.77 percentage points (p < .01). However, including the social measures did not alter the relationship between perceived hearing loss and memory. A series of additional sensitivity tests (not shown) showed no associations between perceived hearing loss and change in any of the four measures of social engagement.
Fixed Effects Regressions Indicating Associations Between Perceived Hearing Problems and Percentage Change in Memory Over Seven Years, Wisconsin Longitudinal Study, N = 3,986.
Note. Columns labeled B (SE) show unstandardized coefficients and standard errors. Columns labeled β show coefficients standardized in X.
p < .01. ***p < .001.
Discussion
Among a sample of nearly 4,000 White non-Hispanic older adults observed at age 65 and 72 years, perceived hearing loss was related to declines in memory. Increased in-person social contact was also associated with improvements in cognitive performance, but there was no evidence that changes in social engagement accounted for the association between perceived hearing loss and memory decline. These findings are consistent with the results of a prior study (Dawes et al., 2015), which similarly found that social isolation and hearing loss were distinct risk factors for poorer cognition.
These findings do not negate the importance of continued research and development of interventions concerning social engagement as a possible protective factor against cognitive decline. The results suggest, however, that targeting social engagement only among older adults who perceive hearing loss is unlikely to reduce their risk of memory loss. Although it seems intuitive that hearing loss would lead to social disengagement, the effect does not appear as widespread as one might anticipate (Mick et al., 2014, 2018). Many older adults may be able to adapt their social interactions to compensate for hearing loss, especially if loss is incremental (Pronk et al., 2014).
This study has several limitations. First, all participants were White, had completed high school, and lived in Wisconsin as adolescents. These sources of sample homogeneity preclude broader population inferences about perceived hearing loss, social disengagement, and memory. Second, the WLS currently includes only two waves of data on hearing and memory among young–old adults. A different pattern of findings might emerge within studies conducted over longer periods of time and with older persons. Finally, many studies measure hearing loss objectively (e.g., through pure tone tests), and estimate that up to two thirds of adults over the age of 70 years experience hearing loss as it is defined by the World Health Organization (Lin et al., 2011). In the WLS, where hearing loss was self-reported, prevalence was dramatically lower at less than 10% of participants. Many people who experience objective loss of some ability to perceive sound are not actively aware of that loss (Lin et al., 2011), such that objective hearing loss severity among reporters in the WLS was likely quite high.
Despite these limitations, this study contributes to research on hearing loss and social engagement as risk factors for memory decline. Our findings suggest that social disengagement does not explain the association between perceived hearing loss and memory loss. Continued empirical examination of causal processes across multiple domains of risk and protective factors for various components of cognition will help to strengthen and refine the evidence base for efforts to promote healthy cognitive aging across the life course.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical Approval
The Boston College Institutional Review Board (IRB) approved this study: Approval Number: 17.078.03-1.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by Grant R01AG057491 from the National Institute on Aging to the first two authors.
