Associations Between Loneliness, Reading Ability and Episodic Memory in Non-Hispanic Black and White Older Adults

Abstract

Objective

Reading ability reflects a variety of beneficial life course exposures and may better index these exposures above and beyond education in racially diverse samples. Growing evidence suggests a negative impact of perceived loneliness on late-life cognitive health when parsing out the effect of other aspects of social relations. Few studies have examined how loneliness interacts with the reading ability or whether it operates differently in Black older adults who have higher dementia risk than Whites.

Methods

Participants in this cross-sectional study were drawn from the Washington Heights-Inwood Columbia Aging Project (n = 425 older adults, M_age = 74.23; 58% Black). Linear regressions estimated the main effects and interactions involving reading ability, loneliness, and race (non-Hispanic Black vs. non-Hispanic White) on episodic memory, controlling for age, sex/gender, and years of education. Subsequent models additionally controlled for income, employment status, depressive symptoms, disease burden, marital status, social network size, and number of social groups.

Results

Higher reading ability was associated with better memory, but loneliness was not associated with memory. The positive association between reading ability and memory was weaker among individuals with greater loneliness, and this interaction did not differ by race.

Conclusions

Loneliness may suppress the protective effect of higher reading ability on cognitive health among both Black and White older adults. Future longitudinal work is needed to clarify causal relationships among loneliness, reading ability, and memory decline.

Introduction

Given that longevity—the strongest predictor of memory decline and Alzheimer’s dementia (AD)—is increasing in the general population, better characterization of potentially modifiable factors for cognitive health is important. Reading ability is a proxy for a variety of life course exposures that could be beneficial in cognitive aging, but it is unclear whether the relationship between reading ability and cognitive function might be moderated by potentially modifiable late-life factors such as loneliness.

Although the number of years spent in school is one determinant of higher reading ability, many cognitively stimulating experiences that occur throughout the life course combine to promote higher reading ability. In addition, the benefits of more years of school for later-life cognitive ability differ across racial groups (Thomas et al., 2020). These differences are likely due, in part, to the racial differences in school quality for which reading ability can also serve as a proxy (Glymour & Manly, 2008). Thus, reading ability is a useful variable when studying the accumulated effects of life course exposures that may be beneficial for cognitive aging in diverse samples. Indeed, one comprehensive study on life experiences and cognitive trajectories in older adults reported that reading ability was associated with both better initial cognitive function and slower cognitive decline even after parsing out the effects of years of education, material resources, physical and cognitively stimulating activity, and biological factors (Brewster et al., 2014). These results indicate that reading ability captures additional resources beyond those that are commonly measured in cohort studies. As such, higher reading ability may help promote individuals’ cognitive flexibility in later life (Ardila et al., 2010). However, the greater flexibility associated with reading ability in identifying and executing different cognitive strategies may be negatively affected by late-life risk factors such as loneliness.

Loneliness is conceptualized as desiring companionship and the distress that accompanies a perceived lack of companionship (Hawkley & Cacioppo, 2010). Though loneliness has been shown to be related to objective aspects of the social network (Hawkley et al., 2008), such as the absolute count of network members (Katz et al., 2020) and number of social activities (Barnes, Mendes de Leon, Wilson, Bienias, & Evans, 2004), individuals with equivalent objective social network characteristics may not be similarly satisfied, leading to varying levels of loneliness. The socioemotional selectivity theory posits that with advancing age, older adults become increasingly selective of individuals with whom they interact in order to improve emotional well-being, which results in reduction of social network size in older age (Carstensen, 1995; Carstensen, Isaacowitz, & Charles, 1999). Although increased quality of interactions with preferred individuals in their social networks is a benefit of this increased intentionality in social interactions, older adults may also be more likely to experience loneliness due to unintentional reductions in social resources due to displacement that reduces engagement with preferred contacts (Savikko, Routasalo, Tilvis, Strandberg, & Pitkälä, 2005) and/or increased disability that can lead to mobility limitations (Mendes de Leon, Glass, & Berkman, 2003). Growing evidence suggests a negative impact of perceived loneliness on late-life cognitive health independent of objective aspects of social relations (Sutin, Stephan, Luchetti, & Terracciano, 2020). As a separate but related construct involving subjective emotional experience, loneliness may also operate independently of depression in its detrimental effects on cognitive health (Donovan et al., 2017).

Finally, loneliness in late life may also differ across racial groups. One study showed that older non-Hispanic Blacks are more likely to experience high levels of loneliness compared with older non-Hispanic Whites (Cole et al., 2007). However, it is not yet known whether there are racial differences in the impact of loneliness on the association between protective factors, such as reading ability and late-life cognition. This question is important given well-documented disparities in cognitive aging, with higher rates of incident dementia among non-Hispanic Blacks compared to non-Hispanic Whites (Mayeda, Glymour, Quesenberry, & Whitmer, 2016). In one study, another indicator of subjective emotional experience that shows a similar detrimental effect on memory in cognitive aging—depressive symptoms—was more strongly related to poorer performance on tasks of episodic memory and executive function in older non-Hispanic Black adults compared to older non-Hispanic White adults (Zahodne, Nowinski, Gershon, & Manly, 2014). Therefore, it may also be the case that the impact of loneliness differs by race. Few studies have examined race differences in the association between loneliness and cognition, and the results have been conflicting (Han et al., 2017; Sutin et al., 2020).

In this cross-sectional study, we aimed to examine whether the association between reading ability and episodic memory is moderated by loneliness and whether these associations differ by race. We hypothesized that worse reading ability and greater loneliness would be associated with worse episodic memory. Further, we hypothesized that loneliness would moderate the association between reading ability and episodic memory such that higher reading ability would be less strongly associated with better episodic memory among those reporting greater loneliness, as it may be the case that loneliness limits the beneficial effects of various resources on cognitive performance. Finally, we explored whether the impact of loneliness differs across non-Hispanic Blacks and Whites.

Methods

Participants were drawn from an ancillary study of psychosocial functioning (Zahodne et al, 2018) conducted with participants in the ongoing, longitudinal Washington Heights-Inwood Columbia Aging Project (WHICAP; Manly, Schupf, Tang, & Stern, 2005; Tang et al., 2001). WHICAP is a prospective, community-based study of aging and dementia in Northern Manhattan. Briefly, participants who were 65 years of age and older and living in northern Manhattan were identified via Medicare records and a commercial marketing company in 1992, 1999, and 2009. WHICAP participants are followed up at 18- to 24-month intervals during which bilingual staff administer a battery of cognitive, functional, and health measures in the participant’s preferred language (English or Spanish). WHICAP is managed by (Blinded for review) and received IRB approval from that institution.

Participants

Starting in 2013, active WHICAP participants recruited as part of the 2009 cohort were invited to join an ancillary study of psychosocial functioning. Given its reliance on self-report data, the primary exclusion criterion for participation in the ancillary study was a diagnosis of dementia according to DSM-III criteria via a consensus group of WHICAP investigators, comprising at least one neurologist and one neuropsychologist, at the most recent core WHICAP visit. Additional exclusion criteria for the current analyses were: diagnosis of dementia at the time of participation in the ancillary study (n = 4); cognitive tests not administered in English due to non-equivalence of English and Spanish reading measures; self-reported race and ethnicity other than non-Hispanic Black or non-Hispanic White race (n = 172); and missing data (n = 22). Race and ethnicity were determined via self-report consistent with the format of the 2000 U.S. Census. A summary of participant characteristics is provided in Table 1.

Measures

Memory composite

Cognitive functioning is assessed during every core WHICAP visit with a battery of paper-and-pencil tests that assess four cognitive domains, including episodic memory (Siedlecki et al., 2010). In the current study, episodic memory composite scores were calculated from scores on the Selective Reminding Test (Buschke & Fuld, 1974), as episodic memory has been found to be sensitive to detecting dementia development (Bäckman, Small, & Fratiglioni, 2001; Boraxbekk et al., 2015). Specifically, z-scores from immediate, delayed, and recognition trials were computed using means and standard deviations (SDs) from the larger WHICAP sample and were subsequently averaged.

Reading ability

Reading ability was assessed during the first core WHICAP visit with the reading ability subtest of the Wide Range Achievement Test-3 (WRAT-3-reading), whereby individuals are asked to pronounce individual words out loud. The final score is based on the number of words pronounced correctly and could range from 0 to 57.

Loneliness

Loneliness-short form from the NIH Toolbox-Emotion Module (Salsman et al., 2013) was used to measure loneliness and was administered during the ancillary study visit. Likert-type items were administered on a laptop. Final scores are unadjusted, standardized theta scores, which have a mean of 0 and a SD of 1. Higher theta scores indicate a higher subjective experience of loneliness.

Covariates

Primary models were controlled for age, sex/gender (reference category was male), and years of education (0–20). Education was chosen as a covariate because it could confound the relationship between reading ability and memory, in that, it may causally contribute to both. Subsequent models additionally covaried for employment status and income, as these variables could be confounders and/or mediators of the association between reading ability and cognitive performance. Joint monthly income (an ordinal variable ranging from 1 [$450 or less] to 12 [at least $4,000]) was modeled as a continuous variable, and current employment status was modeled as a dichotomous variable (unemployed/employed; reference is employed). In order to isolate any potential effects of loneliness from the effects of related health and psychosocial constructs, these subsequent models also controlled for social network size (total number of friends and family members reported by participant); number of social groups (e.g., union, social or recreational group, church-related group, child activity group [e.g., PTA, Boy Scouts], other group) that the participant reported attending at least once a month); depressive symptoms (Irwin, Artin, & Oxman, 1999; Center for Epidemiological Studies-depressive symptoms, CES-d, 10-item version; range: 0–10, where higher scores indicate more depressive symptoms); disease burden (the sum of self-reported medical diagnoses [hypertension, diabetes, stroke, arthritis, COPD, thyroid, liver, renal, ulcer, peripheral vascular disease, cancer, Parkinson’s disease, essential tremor, and multiple sclerosis]; range: 0–14); and marital status (married/not married; reference is married).

Data Analysis

Independent samples t-tests, chi-square tests, and regressions compared Black and White participants on the variables of interest (reading ability, loneliness, episodic memory) as well as covariates. A minimally adjusted regression model examined the main effects of reading and loneliness on episodic memory, covarying for age, sex/gender, race, and years of education. Separate models tested the two-way interactions of loneliness × reading ability, race × reading ability, race × loneliness, and the three-way interaction of reading ability × loneliness × race. Subsequent models included the complete covariate set. Pairwise deletion was used to handle the missing data throughout all analyses.

Results

Racial Differences in the Study Variables

As summarized in Table 1, non-Hispanic Black participants had more chronic diseases, larger social network size, lower income, fewer years of education, lower reading ability, and less loneliness than did White participants. Furthermore, fewer Black participants were male, married, and employed than were White participants. On average, there were no differences between Black and White participants on age, and depressive symptoms, and number of social groups. Table 2 depicts zero-order correlations among all variables of interest. Notably, correlations among all variables were small to medium, with adequate tolerance (all >.79) and VIF (all < 1.26), suggesting that multicollinearity was not a salient concern for the primary regression models.

In order to better characterize the race difference in loneliness, we examined a linear regression, parsing out the effects of social variables that were associated with loneliness, specifically, social network size, and number of social groups. Results indicated that after adjusting for social network size and number of social groups, there were no race differences in loneliness (B = −0.12, SE = 0.10, p = .272).

Race, Loneliness, Reading, and Episodic Memory

The first model examining the main effects of loneliness and WRAT, parsing out the effects of age, sex/gender, race, and years of education, showed that reading ability (B = 0.02, SE = 0.01, p < .001), but not loneliness (B = −0.02, SE = 0.04, p = .565), was associated with memory. Specifically, higher reading ability was associated with better episodic memory performance.

A second model, summarized in Table 3, indicated an interaction between loneliness and reading ability. When dissecting the interaction with loneliness as the moderator, higher reading ability was more strongly associated with better memory among individuals reporting lower levels of loneliness, as shown in Fig. 1. As can be seen in Table 4, after adding in other covariates, the interaction of reading ability and loneliness remained (B = −0.01, SE = 0.01, p = .022).

In a separate minimally adjusted model, an interaction of race and reading ability on memory was found (B = −0.06, SE = 0.02, p < .001), which suggested a weaker positive association between reading ability and episodic memory among non-Hispanic Blacks. Race-stratified models supported this finding (B_Black = 0.02, SE = 0.01, p = .01; B_White = 0.03, SE = 0.01, p = .017). Additional, separate models revealed that there was no two-way interaction between loneliness and race (B = 0.08, SE = 0.08, p = .313) and no three-way interaction between reading ability, loneliness, and race (B = −0.01, SE = 0.01, p = .139), respectively. Interaction findings were similar in the fully adjusted models with the additional covariates, with the exception of the three-way interaction. When the remaining covariates were added, the three-way interaction between loneliness, race, and reading ability became statistically significant (B = −0.01, SE = 0.01, p = .030). However, there was evidence of multicollinearity in this model, with several of the variables having very low tolerance values (<.04), and associations for some variables could not be estimated. Thus, we do not interpret findings for the three-way interaction in this fully adjusted model.

Discussion

The current study sought to clarify associations among loneliness, reading ability, and episodic memory in older Black and White adults. Results indicated a positive main effect of reading ability, but no main effect of loneliness, on episodic memory. Our hypothesis that there would be differences in the strength of the association between reading ability and cognition at different levels of loneliness was supported. Higher reading ability was less strongly associated with better episodic memory among individuals who reported greater loneliness. Finally, although the positive association between reading ability and episodic memory was weaker among non-Hispanic Blacks compared with non-Hispanic Whites, the modification of the association between reading ability and memory by loneliness did not appear to differ by race. These findings provide preliminary evidence that regardless of one’s race, the positive link between reading ability and memory may diminish in the presence of high levels of loneliness, even after parsing out the effects of other related factors such as depressive symptoms, social network size, and number of social groups. This finding is consistent with other studies emphasizing the affective experience of loneliness as one that is related to, but separate from, other objective measures of social support and social engagement (Sutin et al., 2020).

Loneliness, Reading Ability, and Episodic Memory

Although we did not find evidence for an association between loneliness and episodic memory, we did find that the association between reading ability and episodic memory varied by level of loneliness such that the positive association between reading ability and memory function was smaller at higher levels of loneliness. Positive associations between reading ability and memory are consistent with other studies that have also controlled for years of education (Manly et al., 2005). The current study extends these findings by suggesting that this positive association may be moderated by loneliness. Specifically, high levels of loneliness may limit individuals’ ability to utilize or benefit from resources like high reading ability to demonstrate higher performance on episodic memory tasks.

Racial Differences

Results indicating racial differences in reading ability such that Blacks had lower reading ability compared to Whites, which results in lower cognitive performance, are consistent with previous findings (Manly, Byrd, Touradji, Sanchez, & Stern, 2004; Manly, Jacobs, Touradji, Small, & Stern, 2002). However, our finding that Black participants reported lower loneliness than White participants contrasts with the findings from a previous study that found that more Blacks reported loneliness compared to Whites (Cole et al., 2007). Cole and colleagues (2007) and the current study used different approaches to characterizing loneliness, which may have contributed to the discrepant findings. Whereas the current study examined mean differences in loneliness across race, Cole and colleagues (2007) examined proportions of Black participants reporting high loneliness based on study-specific criteria that dichotomized levels of loneliness into high loneliness versus negligible levels of loneliness, after accounting for an index of social network size and frequency of contact.

Although inconsistent with Cole and colleagues (2007), the current finding that Blacks reported experiencing less loneliness compared to Whites is consistent with another study (Finlay & Kobayashi, 2018) that found a larger proportion of Blacks reported experiencing “no loneliness” (using a single-item measure asking whether participants felt lonely, not adjusting for any other indicators of social network characteristics) compared to Whites. Interestingly, the current study also found that Blacks reported larger networks compared to Whites, and the racial difference in loneliness was no longer significant after parsing out the effect of social network size. This pattern of results suggests that there may not be a racial difference in the subjective perception of loneliness when the number of individuals in the social network is held constant. Of note, the current finding that Blacks reported larger social networks than Whites conflicts with previous studies (Ajrouch, Antonucci, & Janevic, 2001; McPherson, Smith-Lovin, & Brashears, 2006), pointing to sample differences. In sum, conflicting findings regarding racial differences in loneliness and social network size emphasize the importance of more studies examining these differences, while accounting for the effects of other constructs that may be related to loneliness.

Our finding that there were no differences in the association between loneliness and episodic memory across race is consistent with the findings of Han and colleagues (2017), which found no race differences in the association between loneliness and cognition in a small, healthy sample of older adults (Han et al., 2017). Lack of evidence for a race × loneliness interaction in the current study is partially consistent with findings of a recent, larger study using a nationally representative sample that also used a comprehensive measure of loneliness and also accounted for other social, psychological, and demographic factors (Sutin et al., 2020). The authors of the Sutin and colleagues’ (2020) study found that loneliness was associated with an increased risk of dementia above and beyond social isolation and sociodemographic factors. However, similar to our findings, the association between loneliness and dementia risk did not differ across race. Thus, although there may be racial differences in loneliness, there do not appear to be racial differences in the impact of loneliness on dementia risk, episodic memory, or the association between reading ability and episodic memory in general samples.

Limitations and Future Directions

This study was not without its limitations. First, its cross-sectional nature prohibits a more conclusive interpretation regarding the direction of effects. Given that the experience of loneliness can persist over time (Cacioppo & Hawkley, 2009), and this persisting dissatisfaction has been associated with worse cognition over time (Tilvis et al., 2004), future longitudinal studies are needed to clarify the nature of associations between loneliness, reading ability, and episodic memory over time. Of note, Sutin and colleagues (2020) found that loneliness was related to subsequent risk of dementia over 6 years. Although Sutin and colleagues (2020) did not test whether lower cognition was associated with increases in loneliness over time, their findings nonetheless suggest an enduring effect of loneliness on cognitive function over time. Additional longitudinal research on interactions between loneliness and protective factors such as reading is needed.

Furthermore, the sample sizes for the individual racial groups were relatively small, which may have limited our ability to detect racial differences in the associations among loneliness, reading ability, and memory. A larger sample may permit more robust analyses to detect three-way interactions involving race, loneliness, and reading ability.

Our study was notable in that, rather than conflating loneliness with other psychosocial constructs, we examined the effects of loneliness before and after accounting for factors that may be associated with loneliness, such as social network size, number of social groups, and depressive symptoms. Furthermore, few studies examine these associations across Black and White older adults. Although our study did not find a three-way interaction between race, loneliness, and reading ability, there were significant differences between Black and White older adults in loneliness that may have implications for health.

Conclusion

In conclusion, our findings may suggest that high levels of loneliness may reduce the protective effects of high reading ability on memory in diverse older adults, though longitudinal data are needed. Future studies can continue to explore ways of maximizing older adults’ social well-being and cognitive functioning by considering resources that promote cognitive functioning and reduce perceptions of loneliness.

Acknowledgements

We acknowledge the WHICAP study participants and the WHICAP research and support staff for their contributions to this study. Portions of this manuscript were presented at the Annual Meeting of the International Neuropsychological Society in February 2019. This manuscript has been reviewed by WHICAP investigators for scientific content and consistency of data interpretation with previous WHICAP study publications. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Funding

Data collection and sharing for this project was supported by the Washington Heights-Inwood Columbia Aging Project (WHICAP; P01AG07232, R01AG037212, RF1AG054023) funded by the National Institute on Aging (NIA) and the National Center for Advancing Translational Sciences, National Institutes of Health (UL1TR001873) and through National Institute of Aging grant numbers K99/R00AG047963, R01AG054520, and R01AG054520S1.

Conflict of Interest

None declared.

References