The Science of Diabetes Self-Management and Care 2025, Vol. 51(5) 449 –461 © The Author(s) 2025 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/26350106251371080 journals.sagepub.com/home/tde
Abstract
Purpose: The purpose of the study was to examine the cross-lagged longitudinal associations between social isolation and cognitive function among older adults with diabetes and to investigate the mediating role of systemic inflammation in these associations.
Methods: Secondary data from the Health and Retirement Study were utilized across 3 waves (2006, 2010, and 2014). To examine the longitudinal relationships between social isolation and cognitive function, a cross-lagged panel modeling approach was employed, with particular attention to the mediating role of C-reactive protein. A sequential predictor procedure was used; initially, only social isolation and cognitive function were analyzed; subsequently, sociodemographic covariates were controlled for; and finally, health-related covariates were added.
Results: Data from 1336 older adults with diabetes in the United States were analyzed. In the partially adjusted model, reciprocal negative associations between social isolation and cognitive function were identified. However, these reciprocal negative associations were no longer statistically significant after accounting for health-related covariates. C-reactive protein did not serve as a mediator in the link between social isolation and cognitive function regardless of adjustments for covariates.
Conclusion: Given the reciprocal associations between social isolation and poorer cognitive function, a feedback loop may exist between these 2 factors. It is crucial to identify mediating mechanisms to disrupt this vicious cycle.
The prevalence of diabetes and dementia is simultaneously increasing dramatically in the global population due to rapid population aging. As of 2021, an estimated 38.4 million individuals in the United States are living with diabetes, with roughly 1.2 million adults newly diagnosed each year.1 Globally, there are about 10 million new cases of dementia each year, and the economic cost of dementia amounts to USD$1.3 trillion.2 Diabetes is associated with a heightened risk of cognitive impairment and dementia. Compared to individuals without diabetes, adults with diabetes face a 60% greater risk of developing all-cause dementia and are 1.5 times more likely develop Alzheimer’s disease.3,4 Given the vulnerability of individuals with diabetes to dementia, it is necessary to identify modifiable factors that can prevent and slow down the progression of cognitive impairment and dementia in individuals with diabetes.
Social isolation has gained attention as one of the modifiable factors contributing to cognitive impairment and dementia. An increasing number of studies have demonstrated a connection between social isolation and poorer cognitive function. A recent scoping review of 12 longitudinal studies reported a significant association between social isolation and cognitive decline.5 Specifically, social isolation was associated with a significant decline in episodic memory and mental status.6 Additionally, social isolation negatively predicted cognitive function over a 2-year follow-up period.7 Older adults with diabetes have additional risk factors contributing to social isolation, including poor physical health factors. For example, diabetes-related complications, such as retinopathy, neuropathy, and foot problems, may reduce physical mobility and increase the risk of social isolation.8
Despite increasing evidence that social isolation negatively affects cognitive function, empirical research on the underlying mechanisms is still lacking. A recent scientific statement by the American Heart Association (AHA) highlighted the effects of both objective and perceived social isolation on brain and cardiovascular health.9 This framework emphasizes the importance of identifying physiological, psychological, and behavioral pathways that mediate the relationship between social isolation, loneliness, and cognitive function. The AHA underscored the necessity for empirical studies to investigate these mediating pathways in the longitudinal association between social isolation and poorer cognitive function. Additionally, the AHA identified inflammatory processes as a key physiological mechanism that may explain the link between social isolation and poorer cognitive function.9
An important channel in disentangling the interplay between social isolation and cognitive function can be the role of the inflammatory response. Accumulating evidence supports the association of social isolation with systemic inflammation biomarkers. C-reactive protein (CRP) is widely recognized as an acute inflammatory biomarker, and it can offer a measure of systemic inflammation because it remains stable over time and shows fewer rapid changes in response to acute stressors.10 Evidence from a meta-analysis suggests that individuals experiencing higher social isolation tend to exhibit elevated CRP levels.11 Conversely, findings from another meta-analysis involving 41 articles revealed that greater social integration is associated with reduced CRP concentrations.12 Elevated CRP may be permeable to the blood-brain barrier and directly access the brain. This can alter brain structure, including white matter, gray matter, and hippocampal volume, and increase the risk of dementia.13 A meta-analysis found that elevated CRP levels are associated with a 1.45-fold increased risk of both all-cause dementia and Alzheimer’s disease.14 These findings support the hypothesis that systemic inflammation may serve as a mediating mechanism linking social isolation to cognitive decline.
Interestingly, emerging literature is beginning to question the directionality of the relationship between social isolation and cognitive function. A scoping review found that social isolation was related to cognitive decline and suggested that the relationship may be bidirectional.5 For instance, individuals with dementia often have less contact with others and experience reduced social engagement.15 Cognitive deficits may result in social withdrawal due to difficulties in maintaining social relationships. Specifically, individuals with Alzheimer’s disease are affected by neurodegeneration in the entorhinal cortex, the area in the brain responsible for the affiliation network, leading to emotional and social detachment from others and exacerbating social isolation.16 However, the majority of existing studies have examined this association in a unidirectional manner. Therefore, additional research is warranted to explore the potential bidirectional dynamics of the relationship.
Therefore, this study examined the reciprocal longitudinal association between social isolation and cognitive function among older adults with diabetes, focusing on the mediating role of systemic inflammation in a nationally representative sample in the United States. A cross-lagged panel model was applied to robustly assess these bidirectional associations. Understanding these mechanisms may guide the development of interventions to mitigate social isolation and prevent dementia in this vulnerable population.
Using secondary data, a longitudinal association study was conducted. This study utilized data from the Health and Retirement Study (HRS), which was collected over 3 waves in 2006 (T1), 2010 (T2), and 2014 (T3). The HRS is a longitudinal panel study that has conducted data collection since 1992 to investigate a nationally representative sample of older adults who are over 50. The HRS conducted in-person interviews with a randomly selected half of the participants and collected the data over the phone for the other half.17 Participants who were interviewed in person completed the Psychosocial and Lifestyle Questionnaire (PLQ). The PLQ is a self-administered survey regarding lifestyle, social relationships, family structure, and psychological situations. Additionally, nurses collected participants’ biomarker data during their home visits. These subsamples completed the PLQ and blood draws every 4 years. Three data sets collected in 2006, 2010, and 2014, including the PLQ, the HRS Venous Blood Study, and the RAND HRS Longitudinal File, were merged for analysis. Further information on the details of the HRS is available on the HRS website (https://hrs.isr.umich.edu). The Institutional Review Board reviewed and confirmed the exemption because the HRS data set is publicly available and de-identified.
Inclusion and/or exclusion criteria. Individuals with self-reported physician-diagnosed diabetes who were over the age of 50 were the main analytical unit in this study. Participants who completed a cognitive function assessment and at least 1 follow-up survey were included in the analysis. Participants who had dementia at baseline, as indicated by a Telephone Interview for Cognitive Status (TICS) score below 7, were excluded from the analysis. A TICS score ranging from 0 to 6 was indicative of dementia, scores between 7 and 11 suggested cognitive impairment without dementia, and scores from 12 to 27 were classified as normal cognitive function.18
Analytic sample. First, individuals with self-reported, physician-diagnosed diabetes at baseline were extracted for analysis (n = 1453). Then, those under 50 years of age (n = 13) and individuals with probable dementia at baseline (TICS < 7; n = 104) were excluded. The final sample comprised 1336 participants at T1, 1033 at T2 (reflecting a 22.7% attrition rate), and 780 at T3 (24.5% attrition rate).
Social isolation. Social isolation was operationally defined using the Steptoe Social Isolation Index, a 5-item measure originally developed for the English Longitudinal Study of Aging, with demonstrated validity in prior research.19 This index captures multiple dimensions of social connectedness, including (1) living alone or being unmarried, (2) infrequent contact (less than once per month) with one’s children, (3) infrequent contact with other relatives, (4) infrequent contact with friends, and (5) nonparticipation in social organizations, groups, or clubs on a monthly basis. Each criterion was dichotomously coded (0 = no, 1 = yes), and scores were summed to yield a total score ranging from 0 to 5. Higher scores indicate a greater degree of social isolation, reflecting the cumulative impact of limited social ties across these domains.
Cognitive function. Cognitive function was assessed using the TICS, administered as part of the HRS.20 The TICS is a widely used 27-item screening tool that provides a continuous measure of global cognitive performance.18 It evaluates multiple cognitive domains, including immediate and delayed word recall, serial 7 subtraction, and backward counting. Scores from all items were summed to produce a composite cognitive function score, with higher scores indicating better cognitive performance. The TICS has demonstrated strong psychometric properties, good test-retest reliability (r = .95),21 and excellent discriminative validity for detecting cognitive impairment and dementia (sensitivity = 88%, specificity = 87%).22
Systemic inflammation. Systemic inflammation was measured using serum CRP levels obtained from dried blood spot (DBS) samples collected during HRS interviews. CRP is a well-established biomarker of systemic inflammation and has been extensively validated as an objective, continuous measure in epidemiological research. Trained interviewers obtained DBS samples by filling up to 10 circles on 2 Whatman blood spot cards. The cards were then placed in a ventilated cardboard container designed to allow airflow from all directions, ensuring proper drying for at least 2 hours before shipping. DBS cards were stored at –70 °C before and after analysis. CRP concentrations were quantified using a high-sensitivity enzyme-linked immunosorbent assay.23 Prior research has shown high agreement between DBS-based CRP measurements and venous blood assays (correlation coefficients r = .99), supporting their validity and reliability for large-scale population studies.24 Due to skewness, CRP values were logtransformed for analysis.
Covariates. To reduce the potential impact of confounding variables, adjustments were made for sociodemographic and health-related covariates selected based on the previous studies.6,25 Sociodemographic variables included age, sex, race/ethnicity, educational attainment, and annual income. Health-related covariates included the number of comorbidities, depressive symptoms, and glycemic control. The number of comorbidities was calculated by summing self-reported diagnoses of 6 chronic diseases: stroke, chronic lung disease, heart disease, hypertension, arthritis, and cancer. Depressive symptoms were assessed using the 8-item Center for Epidemiologic Studies Depression Scale, which evaluates symptoms experienced over the past week. Internal consistency was high across all time points (Cronbach’s αs = .81, .79, and .80 for T1, T2, and T3, respectively). Glycemic control was measured with A1C, which reflects average blood glucose levels over the previous 2 to 3 months.26 A1C levels were obtained using DBS tests in the same way that CRP measures were obtained. Assay was conducted with the Bio-Rad Laboratories Variant II High-Pressure Liquid Chromatography System.23
All data analyses were conducted utilizing SPSS 28.0 and Mplus version 8.3. Before data analysis, we examined the data distribution, the missingness for each variable, and outliers. Because CRP levels did not follow a normal distribution, log-transformed CRP values were used for the main analysis. Categorical variables were summarized using frequencies and percentages, and continuous variables were summarized using means and standard deviations. Pearson’s correlation coefficient was calculated to identify the bivariate correlations between the main variables. A P value of <.05 was considered statistically significant.
This study used a 3-wave cross-lagged panel model (CLPM) under a structural equation modeling framework. The CLPM approach has been used to evaluate the reciprocal relationships between variables over time. It utilizes longitudinal data to understand the underlying processes of reciprocal causality between sets of constructs.27 The main variables were treated as manifest variables. To identify the reciprocal associations between social isolation and cognitive function, this study employed a sequential predictor procedure, evaluating 3 distinct models. Model 1, an unadjusted baseline model, included only the primary variables (ie, social isolation and cognitive function) without controlling for covariates. Subsequently, partially and fully adjusted models tested the same associations by incorporating covariates. Sociodemographic variables (ie, age, sex, race/ethnicity, education levels, and income) were added in Model 2. Health-related covariates (ie, comorbidities, depressive symptoms, and A1C levels) were added in Model 3. Sociodemographic covariates were modeled as time-invariant factors, and health-related covariates were treated as time-variant factors. Missing data were dealt with by full information maximum likelihood estimation. To identify the mechanisms underlying the social isolation/cognitive function interplay, CRP was incorporated into the CLPM, and longitudinal mediation tests were conducted. Model 4 was the unadjusted baseline model, excluding covariates. Sociodemographic covariates were added in Model 5, and health-related covariates were added in Model 6. Model fit was assessed using the comparative fit index (CFI) with a threshold of >.90, the root mean square error of approximation (RMSEA) with a cutoff of <.06, and the standardized root mean square residual (SRMR) with a standard of <.08, in accordance with the criteria suggested by Hu and Bentler.28
Table 1 presents the sample characteristics of this study. The mean age of the 1336 surveyed participants was 68.66 years at baseline (SD = 9.28). Of the participants, 55.0% were female. The majority of participants identified as White/Caucasian (76.6%), with a substantial proportion having attained an educational level below college (62.9%). Most participants’ annual income was in the low range ($0-$29 999; 45.9%). Because all participants included in the analysis had diabetes, the number of comorbidities was calculated excluding diabetes. Across the 3 waves, the mean number of comorbidities, measured on a scale ranging from 0 to 6, remained below 3. The average A1C levels were approximately 7%, indicating borderline levels of uncontrolled diabetes. The American Diabetes Association classifies an A1C of 7% or higher as uncontrolled diabetes.29
The mean level of social isolation was 1.28 (SD = 1.08) at baseline, and social isolation slightly increased in subsequent waves (mean = 1.43, SD = 1.12 at T2; mean = 1.52, SD = 1.13 at T3). The average CRP level was 5.47 mg/dL at baseline, indicating moderate elevation (from 1.0 to 10.0 mg/dL).30 The mean of cognitive function showed a decreasing pattern over the 3 waves (mean = 14.62, SD = 3.78 at T1; mean = 13.89, SD = 4.33 at T2; and mean = 13.68, SD = 4.24 at T3) but remained within the normal cognitive status range.25 The correlation matrix between social isolation, log-transformed CRP, and cognitive function is presented in Supplementary Table 1.
The results for the stepwise CLPM for the association between social isolation and cognitive function are presented in Supplementary Table 2. Figure 1 depicts the unadjusted CLPM (Model 1), with all path coefficient estimates standardized. Some model fit indices for Model 1 were good (CFI = .929, SRMR = .044), but the RMSEA of .143 indicated poor fit. Cognitive function had a significant negative lagged effect on social isolation (βcog T1→iso T2 = –0.100, P = .001; βcog T2→iso T3 = –0.099, P = .002). Social isolation at T2 negatively predicted cognitive function at T3 (βiso T2→cog T3 = –0.108, P = .000); however, the same association did not appear between T1 and T2 (βiso T1→cog T2 = –0.041, P = .098).
After controlling for sociodemographic covariates, all autoregressive and lagged path coefficients were significant. Model fit indices were acceptable (CFI = .916, RMSEA = .066, SRMR = .049). Cognitive function was significantly predictive of social isolation in the next wave (βcog T1→iso T2 = –0.102, P = .001; βcog T2→iso T3 = –0.100, P = .002). In turn, social isolation was associated with poor cognitive function in the next wave (βiso T1→cog T2 = –0.054, P = .025; βiso T2→cog T3 = –0.108, P < .001). In conclusion, a significant reciprocal relationship between social isolation and cognitive function was observed even after controlling for sociodemographic factors. The results for Model 2 are presented in Figure 2.
Health-related time-variant covariates (ie, the number of comorbidities, depressive symptoms, and A1C levels) and covariates were included in Model 2. In Model 3, no concurrent associations between social isolation and cognitive function were observed (see Figure 3). The negative lagged effects of cognitive function at T1 on social isolation at T2 remained (βcog T1→iso T2 = –0.083, P = .042), and the same association attenuated between cognitive function at T2 and social isolation at T3 after controlling for health-related covariates (βcog T2→iso T3 = –0.076, P = .062). Additionally, social isolation at T2 negatively predicted cognitive function at T3 (βiso T2→cog T3 = –0.129, P < .001), and social isolation at T1 did not have a significant lagged effect on cognitive function at T2 (βiso T1→cog T2 = –0.015, P = .641). Model fit indices for Model 3 were reasonable (CFI = .897, RMSEA = .065, SRMR = .053).
The results for the stepwise CLPM examining the mediating role of CRP on the relationship between social isolation and cognitive function are summarized in Supplementary Table 3. Figure 4 shows the unadjusted CLPM identifying the mediating effect of CRP. Social isolation did not predict CRP at any time point (βiso→CRP T2 = 0.024, P = .423; βiso T2→CRP T3 = 0.016, P = .630). However, social isolation at T2 and CRP at T2 separately significantly predicted cognitive function at T3 (βiso T2→cog T3 = –0.113, P < .001; βCRP T2→cog T3 = 0.099, P < .001). Interestingly, CRP at T2 had positive lagged effects on social isolation at T3 (βCRP T2→iso T3 = 0.071, P = .026), although the same association did not appear between T1 and T2 (βCRP T1→iso T2 = –0.003, P = .920). In conclusion, CRP did not mediate the longitudinal association between social isolation and cognitive function.
Similar to Model 4, Models 5 and 6 indicated that the mediating effect of CRP was not statistically significant in the relationship between social isolation and cognitive function whether controlling for covariates or not. Model 5 controlled for sociodemographic covariates (see Figure 5) and showed that social isolation was not significantly predictive of CRP in the next wave (βiso T1→CRP T2 = 0.026, P = .371; βiso T2→CRP T3 = 0.017, P = .614). Social isolation at T2 and CRP at T2 independently were associated with cognitive function in the next wave (βiso T2→cog T3 = –0.096, P < .001; βCRP T2→cog T3 = 0.081, P = .001). Although CRP at T2 predicted social isolation at T3 (βCRP T2→iso T3 = 0.071, P = .028), this lagged association was not observed at another time point (βCRP T1→iso T2 = –0.005, P = .855). The reciprocal association between social isolation and cognitive function remained significant after controlling for sociodemographic variables. Model 5 had a good fit to the data (RMSEA = .059, CFI = .904, SRMR = .052).
Model 6 controlled for both sociodemographic and health-related covariates (see Figure 6). Similarly, social isolation did not predict CRP in the next wave (βiso T1→CRP T2 = 0.049, P = .182; βiso T2→CRP T3 = –0.009, P = .825). However, social isolation and CRP at T2 independently predicted cognitive function at T3 (βiso T2→cog T3 = –0.133, P = .000; βCRP T2→ cog T3 = 0.076, P = .013). The lagged effects of CRP at T2 on social isolation at T3 became nonsignificant in Model 6 (βCRP T2→iso T3 = 0.036, P = .330). Thus, the study results did not support that CRP mediates the longitudinal association between social isolation and cognitive function regardless of adjusting for sociodemographic and health-related covariates.
This study examined the reciprocal longitudinal associations between social isolation and cognitive function in older adults with diabetes to simultaneously unpack the mediating role of systemic inflammation in this relationship. After controlling for sociodemographic covariates, findings revealed that higher social isolation predicted lower cognitive function, and conversely, lower cognitive function predicted greater social isolation among older adults with diabetes. However, these reciprocal longitudinal associations were attenuated when both sociodemographic and health-related covariates were included. Additionally, CRP did not exhibit a significant mediating effect in the negative reciprocal relationship between social isolation and cognitive function.
The significant contribution of this study resides in elucidating the reciprocal longitudinal associations between social isolation and poorer cognitive function among older adults with diabetes, even after controlling for sociodemographic covariates. This finding aligns with previous research on the bidirectional associations between social isolation and cognitive function in the general adult population.31 Specifically, the results are consistent with prior studies reporting that social isolation is associated with poor cognitive function.6,7 Similarly, the findings corroborate earlier studies demonstrating that poor cognitive function predicts higher social isolation.32,33 Collectively, these findings reinforce the assertion that social isolation negatively impacts cognitive function and vice versa, elucidating a vicious cycle that damages cognitive function over time.
However, when health-related covariates were included, the negative reciprocal longitudinal association between social isolation and cognitive function attenuated. This suggests that health-related covariates might act as potential underlying pathways explaining the longitudinal association between social isolation and poorer cognitive function. According to a scientific statement from the AHA, the potential mediating pathways linking social isolation and brain health may include cardiovascular risk factors, neuroendocrine axes, and depression.9 Healthrelated covariates, including comorbidities, depressive symptoms, and A1C levels, were included in the analysis because these factors may function as effective pathways linking social isolation and poorer cognitive function. These covariates may have contributed to attenuating the observed association between social isolation and cognitive function. Notably, the study population consisted of older adults with diabetes, who were particularly sensitive to the effects of A1C levels on poorer cognitive function. Specifically, previous research found that compared to those with A1C levels between 6.5% and 7.5%, individuals with A1C levels between 7.5% and 8.5% had a 1.33 times increased risk of dementia and those with A1C levels greater than 8.5% had a 1.54 times increased risk.34 To determine whether these health-related covariates uniquely affect older adults with diabetes, further studies are needed to compare the general older population with older adults with diabetes regarding the reciprocal longitudinal association between social isolation and poorer cognitive function while considering health-related covariates.
The findings did not elucidate the mechanism by which CRP mediates the longitudinal relationship between social isolation and cognitive function. In this study, social isolation did not predict CRP at any time point, a finding that is inconsistent with previous research.35,36 One possible explanation for this contradictory finding is the difference in follow-up intervals. Whereas this study evaluated the effects of social isolation on CRP after 4 years, the literature supporting the association between social isolation and systemic inflammation examined long-term periods exceeding 20 years. For instance, Matthews et al36 demonstrated a significant relationship between childhood social isolation and inflammation in adulthood. Similarly, Yang et al35 found that low levels of social relationships in childhood predicted an increased risk of inflammation in adulthood. Thus, the impact of social isolation on inflammatory responses may not be a short-term phenomenon but rather a long-term phenomenon observable over the life course. Another possible explanation is that the characteristics of diabetes may obscure the relationship between social isolation and inflammation. Individuals with diabetes typically exhibit higher levels of inflammation compared to the general population. Specifically, insulin resistance in individuals with type 2 diabetes may result not only in hyperglycemia but also in low-grade inflammation.37 Individuals with diabetes typically exhibit elevated levels of inflammation, which may obscure the effects of social isolation on cognitive function in this population. Collectively, these findings underscore the significance of longitudinal studies on social isolation and its associations with systemic inflammation over the life course, highlighting the need for further investigation targeting individuals with diabetes.
The results revealed a pattern in which the average score of social isolation slightly increased over successive waves. This finding aligns with previous research indicating that social isolation tends to become more severe with advancing age.38 Specifically, older adults age 80 years and older exhibited a significantly higher likelihood of experiencing severe social isolation compared to those ages 65 to 69 years.38 Additionally, this study demonstrated significant autoregressive paths between social isolation at T1, T2, and T3, indicating the extent to which social isolation tends to persist from one time point to the next. In other words, the severity of social isolation at one time point predicts its severity at subsequent time points. Older adults are more likely to experience social isolation due to additional risk factors, such as functional disability, sensory deprivation, loss of spouse, and retirement.39 Particularly, older adults with diabetes are more prone to experiencing the compounded effects of multiple risk factors influencing social isolation.8 This underscores the necessity for effective interventions to prevent the increasing incidence of social isolation among vulnerable populations. Hence, research and intervention development targeting populations at risk of social isolation, particularly diabetes, are essential.
A major strength of this study is the analysis of a large, nationally representative sample, allowing for the generalizability of findings across older adults with diabetes in the United States. To assess the directionality and potential causal pathways among the variables under investigation, a longitudinal design and cross-lagged panel modeling approach were utilized. Furthermore, we controlled for potential sociodemographic and health-related confounding factors, thereby limiting the possibility of residual confounding.
However, there are several limitations. First, because it is based on secondary data analysis, the selection of variables was restricted. The assessment of systemic inflammation biomarkers was limited to CRP levels. Future studies investigating the association between social isolation and cognitive function should consider other inflammatory biomarkers, including interleukin-6 and fibrinogen. Furthermore, it was not possible to determine whether participants were using medications that affect inflammation levels, such as antibiotics, or experiencing an acute illness at the time of CRP measurement. Future research should consider these potential confounding factors and undertake further investigations. Additionally, although systemic inflammation was considered a physiological mediator in the relationship between social isolation and cognitive function, stress-related mechanisms and psychological pathways may also play significant mediating roles in this association. Social isolation is regarded as a social stressor that can activate the hypothalamic-pituitary-adrenal axis and the sympathetic adrenomedullary system, thereby enhancing the inflammatory response.40 The effects of social isolation on health outcomes may be more pronounced through stress response rather than through the inflammatory process. Due to the inherent limitations of secondary data analysis, data on stress biomarkers could not be obtained or included in the analysis. However, more definitive results might be achieved if stress-related biomarkers, such as cortisol, are utilized. Thus, it is imperative to conduct research that incorporates biomarkers or psychometric measurements related to stress to obtain more comprehensive insights.
The present findings underscore the importance of addressing social isolation as a modifiable risk factor in the management of older adults with diabetes, with particular attention to the prevention of cognitive decline. Given the observed bidirectional association between social isolation and diminished cognitive function—although attenuated after adjusting for health-related covariates—health care providers are encouraged to integrate routine assessments of both social isolation and cognitive function into standard diabetes care. Early identification of atrisk individuals enables the implementation of targeted interventions, such as peer-support programs, community engagement initiatives, and cognitive training, to help disrupt the potential feedback loop between increasing social isolation and declining cognition.
Furthermore, the attenuation of these associations after controlling for comorbidities, depressive symptoms, and A1C levels highlights the need for a holistic, multidisciplinary approach that addresses the intertwined physical, psychological, and social determinants of cognitive health. Although systemic inflammation, as measured by CRP, did not mediate the relationship in this study, health care providers should be attentive to alternative physiological and psychosocial mechanisms, including other inflammatory biomarkers and stress-related pathways. Collaborative care models involving mental health professionals, primary care providers, and community organizations may be particularly effective in mitigating both social isolation and cognitive decline, ultimately enhancing long-term diabetes self-management and quality of life.
Evidence of negative reciprocal longitudinal associations between social isolation and cognitive function was found even after controlling for sociodemographic covariates. These findings suggest a detrimental feedback loop where higher levels of social isolation lead to poorer cognitive function, which subsequently exacerbates social isolation. However, this empirical study did not support the mediating role of systemic inflammation in the association between social isolation and poorer cognitive function. Future studies should consider other inflammatory biomarkers or stress-related physiological and psychological mediating factors as alternative mechanisms. Thus, further research is necessary to elucidate the underlying mechanisms between social isolation and cognitive function to interrupt the potential vicious cycle leading to cognitive impairment and dementia. Identifying these mechanisms will provide a foundation for developing novel interventions aimed at alleviating social isolation and preventing cognitive impairment and dementia.
The Health and Retirement Study is sponsored by the National Institute on Aging (Grant No. U01AG009740) and is conducted by the University of Michigan. The authors gratefully acknowledge the efforts of all study participants and staff who made this research possible.
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The author(s) received no financial support for the research, authorship, and/or publication of this article.
This study is a secondary data analysis. The Health and Retirement Study (HRS) has already been reviewed and approved by the University of Michigan Institutional Review Board (IRB), and all participants provided informed consent. The Ohio State University IRB reviewed and confirmed the exemption because the HRS data set is publicly available and de-identified (2024E0106).
Bohyun Kim https://orcid.org/0000-0002-9347-2730
Jie Hu https://orcid.org/0000-0003-4772-0315
The data used in the study was a national data set from the Health Retirement Study. The access of data requires approval and permission from the Health Retirement Study.
Supplemental material for this article is available online.
Centers for Disease Control and Prevention. National diabetes statistics report. Published May 15, 2024. Accessed May 28, 2025. https://www.cdc.gov/diabetes/php/data-research/index.html
World Health Organization. Dementia. Published March 13, 2025. Accessed May 28, 2025. https://www.who.int/news-room/fact-sheets/detail/dementia
Zhang J, Chen C, Hua S, et al. An updated meta-analysis of cohort studies: diabetes and risk of Alzheimer’s disease. Diabetes Res Clin Pract. 2017;124:41-47. doi:10.1016/j.diabres.2016.10.024
Chatterjee S, Peters SAE, Woodward M, et al. Type 2 diabetes as a risk factor for dementia in women compared with men: a pooled analysis of 2.3 million people comprising more than 100,000 cases of dementia. Diabetes Care. 2016;39(2):300-307. doi:10.2337/dc15-1588
Cardona M, Andrés P. Are social isolation and loneliness associated with cognitive decline in ageing? Front Aging Neurosci. 2023;15:1075563. doi:10.3389/fnagi.2023.1075563
Yu B, Steptoe A, Chen Y, Jia X. Social isolation, rather than loneliness, is associated with cognitive decline in older adults: the China Health and Retirement Longitudinal Study. Psychol Med. 2021;51(14):2414-2421. doi:10.1017/S0033291720001014
Evans IEM, Llewellyn DJ, Matthews FE, Woods RT, Brayne C, Clare L. Social isolation, cognitive reserve, and cognition in healthy older people. PLoS One. 2018;13(8):e0201008. doi:10.1371/journal.pone.0201008
Ida S, Murata K. Social isolation of older adults with diabetes. Gerontol Geriatr Med. 2022;8:23337214221116232. doi:10.1177/23337214221116232
Cené CW, Beckie TM, Sims M, et al. Effects of objective and perceived social isolation on cardiovascular and brain health: a scientific statement from the American Heart Association. J Am Heart Assoc. 2022;11(16):e026493. doi:10.1161/JAHA.122.026493
Marsland AL, Walsh C, Lockwood K, John-Henderson NA. The effects of acute psychological stress on circulating and stimulated inflammatory markers: a systematic review and meta-analysis. Brain Behav Immun. 2017;64:208-219. doi:10.1016/j.bbi.2017.01.011
Smith KJ, Gavey S, RIddell NE, Kontari P, Victor C. The association between loneliness, social isolation and inflammation: a systematic review and meta-analysis. Neurosci Biobehav Rev. 2020;112:519-541. doi:10.1016/j.neubiorev.2020.02.002
Uchino BN, Trettevik R, Kent de Grey RG, Cronan S, Hogan J, Baucom BRW. Social support, social integration, and inflammatory cytokines: a meta-analysis. Health Psychol. 2018;37(5):462-471. doi:10.1037/hea0000594
Hsuchou H, Kastin AJ, Mishra PK, Pan W. C-reactive protein increases BBB permeability: implications for obesity and neuroinflammation. Cell Physiol Biochem. 2012;30(5):1109-1119. doi:10.1159/000343302
Koyama A, O’Brien J, Weuve J, Blacker D, Metti AL, Yaffe K. The role of peripheral inflammatory markers in dementia and Alzheimer’s disease: a meta-analysis. J Gerontol A Biol Sci Med Sci. 2013;68(4):433-440. doi:10.1093/gerona/gls187
Hackett RA, Steptoe A, Cadar D, Fancourt D. Social engagement before and after dementia diagnosis in the English Longitudinal Study of Ageing. PLoS One. 2019;14(8):e0220195. doi:10.1371/journal.pone.0220195
Porcelli S, Van Der Wee N, van der Werff S, et al. Social brain, social dysfunction and social withdrawal. Neurosci Biobehav Rev. 2019;97:10-33. doi:10.1016/j.neubiorev.2018.09.012
Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JW, Weir DR. Cohort profile: the Health and Retirement Study (HRS). Int J Epidemiol. 2014;43(2):576-585. doi:10.1093/ije/dyu067
Crimmins EM, Kim JK, Langa KM, Weir DR. Assessment of cognition using surveys and neuropsychological assessment: the Health and Retirement Study and the Aging, Demographics, and Memory Study. J Gerontol B Psychol Sci Soc Sci. 2011;66(suppl 1):i162-i171. doi:10.1093/geronb/gbr048
Steptoe A, Shankar A, Demakakos P, Wardle J. Social isolation, loneliness, and all-cause mortality in older men and women. Proc Natl Acad Sci. 2013;110(15):5797-5801. doi:10.1073/pnas.1219686110
Brandt J, Spencer M, Folstein M. The telephone interview for cognitive status. Neuropsychiatry Neuropsychol Behav Neurol. 1988;1(2):111-117.
Seo EH, Lee DY, Kim SG, et al. Validity of the telephone interview for cognitive status (TICS) and modified TICS (TICSm) for mild cognitive imparment (MCI) and dementia screening. Arch Gerontol Geriatr. 2011;52(1):e26-e30. doi:10.1016/j.archger.2010.04.008
Manly JJ, Schupf N, Stern Y, Brickman AM, Tang MX, Mayeux R. Telephone-based identification of mild cognitive impairment and dementia in a multicultural cohort. Arch Neurol. 2011;68(5):607-614. doi:10.1001/archneurol.2011.88
Crimmins E, Faul J, Kim JK, Weir D. Documentation of DBS Blood-Based Biomarkers in the 2016 Health and Retirement Study. Survey Research Center, University of Michigan; 2020.
Crimmins E, Kim JK, McCreath H, Faul J, Weir D, Seeman T. Validation of blood-based assays using dried blood spots for use in large population studies. Biodemography Soc Biol. 2014;60(1):38-48. doi:10.1080/19485565.2014.901885
Yu K, Siang Ng TK. Investigating biological pathways underpinning the longitudinal association between loneliness and cognitive impairment. J Gerontol A Biol Sci Med Sci. 2023;78(8):1417-1426. doi:10.1093/gerona/glac213
Petersen JR, Omoruyi FO, Mohammad AA, Shea TJ, Okorodudu AO, Ju H. Hemoglobin A1c: assessment of three POC analyzers relative to a central laboratory method. Clin Chim Acta. 2010;411(23-24):2062-2066. doi:10.1016/j.cca .2010.09.004
Kenny DA. Cross-lagged panel correlation: a test for spuriousness. Psychol Bull. 1975;82(6):887-903. doi:10.1037/0033-2909.82.6.887
Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Modeling. 1999;6(1):1-55. doi:10.1080/10705519909540118
American Diabetes Association. What is the A1C test? Accessed May 28, 2025. https://diabetes.org/diabetes/a1c
Nehring SM, Goyal A, Patel BC. C Reactive Protein. StatPearls Publishing; 2024.
Qi X, Pei Y, Malone SK, Wu B. Social isolation, sleep disturbance, and cognitive functioning (HRS): a longitudinal mediation study. J Gerontol A Biol Sci Med Sci. 2023;78(10):1826-1833. doi:10.1093/gerona/glad004
Schnittger RIB, Wherton J, Prendergast D, Lawlor BA. Risk factors and mediating pathways of loneliness and social support in community-dwelling older adults. Aging Ment Health. 2012;16(3):335-346. doi:10.1080/13607863.2011.629092
Amano T, Morrow-Howell N, Park S. Patterns of social engagement among older adults with mild cognitive impairment. J Gerontol B Psychol Sci Soc Sci. 2020;75(7):1361-1371. doi:10.1093/geronb/gbz051
Wang K, Zhao S, Lee EKP, et al. Risk of dementia among patients with diabetes in a multidisciplinary, Primary Care Management Program. JAMA Netw Open. 2024;7(2):e2355733. doi:10.1001/jamanetworkopen.2023.55733
Yang YC, Boen C, Gerken K, Li T, Schorpp K, Harris KM. Social relationships and physiological determinants of longevity across the human life span. Proc Natl Acad Sci USA. 2016;113(3):578-583. doi:10.1073/pnas.1511085112
Matthews T, Rasmussen LJH, Ambler A, et al. Social isolation, loneliness, and inflammation: a multi-cohort investigation in early and mid-adulthood. Brain Behav Immun. 2024;115:727-736. doi:10.1016/j.bbi.2023.11.022
Sharif S, Van der Graaf Y, Cramer MJ, et al. Low-grade inflammation as a risk factor for cardiovascular events and allcause mortality in patients with type 2 diabetes. Cardiovasc Diabetol. 2021;20(1):220. doi:10.1186/s12933-021-01409-0
Cudjoe TKM, Roth DL, Szanton SL, Wolff JL, Boyd CM, Thorpe RJ. The epidemiology of social isolation: National Health and Aging Trends Study. J Gerontol B Psychol Sci Soc Sci. 2020;75(1):107-113. doi:10.1093/geronb/gby037
Donovan NJ, Blazer D. Social isolation and loneliness in older adults: review and commentary of a National Academies report. Am J Geriatr Psychiatry. 2020;28(12):1233-1244. doi:10.1016/j.jagp.2020.08.005
Cacioppo JT, Cacioppo S. Loneliness in the modern age: an evolutionary theory of loneliness (ETL). Adv Exp Soc Psychol. 2018;58:127-197. doi:10.1016/bs.aesp.2018.03.003
From College of Nursing, Inje University, Gimhae-si, Gyeongsangnam-do, Republic of Korea (Dr Kim); and College of Nursing, The Ohio State University, Columbus, Ohio (Dr Hu)
Corresponding Author: Bohyun Kim, College of Nursing, Inje University, C426, 197, Inje-ro, Gimhae-si, Gyeongsangnam-do 50834, Republic of Korea. E-mail: bhkim5854@inje.ac.kr
