The impact of social isolation and digital exclusion on mental and physical health in older adults: A meta-analysis

Abstract

Background:

Social isolation and digital exclusion are significant public health concerns, particularly among older adults, with potential links to mental, cognitive, and physical health. This study aims to quantify the associations of social isolation and digital exclusion with depression, cognitive decline, and physical functioning in community-dwelling older adults.

Methods:

We conducted a systematic review and meta-analysis of 11 primary studies (9 observational, 2 randomized controlled trials) published since the year 2000. These studies included over 3,50,000 participants from North America, Europe, and Asia. We performed meta-analysis using random-effects models to calculate pooled odds ratios (ORs) with 95% confidence intervals (CIs) for depression and cognitive decline. Physical function outcomes were analyzed qualitatively due to insufficient data for quantitative pooling. We assessed heterogeneity using the I² statistic and performed subgroup and sensitivity analyses to explore potential biases.

Results:

For depression risk, the pooled OR from analyses of approximately 1,11,784 participants was 1.60 (95% CI: 1.39–1.84, P < .001). This indicates a 60% higher likelihood of depressive symptoms among socially or digitally disconnected seniors. Cognitive decline, assessed in 3 longitudinal cohorts, showed a nonsignificant pooled OR of 1.03 (95% CI: 0.92–1.15). Physical function outcomes, including reductions in short physical performance battery scores and increased activities of daily living disability, were consistently worse among isolated or digitally excluded individuals. We found moderate heterogeneity for depression (I² = 43%) and cognition (I² = 55%), with no evidence of small-study bias.

Conclusion:

Our findings suggest that both social isolation and digital exclusion are independently and significantly associated with an elevated risk of depression in older adults. Furthermore, these factors correlate with poorer physical functioning. Digital exclusion may be a critical “super-social determinant” of health, underscoring the need for integrated interventions that address both social connectivity and digital literacy to improve the well-being of the elderly.

1. Introduction

The lack of social connections among the elderly is emerging as a serious public health issue. Social isolation refers to a state of extremely limited social contact with family and friends, while loneliness is the subjective feeling of social disconnection.^[1] These 2 concepts are distinct, representing a lack of social networks from an objective, structural perspective and a subjective, functional perspective, respectively.^[2] Previous studies have consistently reported that social isolation and loneliness have a negative impact on the mental and physical health of older adults.^[3,4] For example, numerous longitudinal studies and meta-analyses have reported that a lack of social relationships increases the risk of mental problems such as depression and anxiety and is also associated with cognitive decline and an increased incidence of dementia.^[5] In particular, Holt-Lunstad et al revealed that social isolation significantly increases the risk of premature death, with its impact being comparable to that of smoking and obesity.^[6] The recent COVID-19 pandemic and the implementation of social distancing have further highlighted the issue of social disconnection among the elderly, increasing interest in the effects of social isolation on their health.^[7]

Meanwhile, in modern society, digital exclusion (or the digital divide) is also gaining attention as a new factor contributing to health inequalities among older adults. Digital exclusion refers to the gap in access to and use of digital technologies such as the internet and smart devices, and it describes a state where an individual either does not use digital devices at all or has low digital literacy.^[8] As medical and social services are increasingly becoming digitized, older adults with limited digital access may be excluded from necessary information and services and have reduced opportunities for social participation.^[9]

To date, research on the health effects of social isolation (and loneliness) and studies on the digital divide have mostly been conducted separately.^[10–20] In reality, however, the 2 factors are closely related: socially isolated older adults are more likely to be excluded from using digital technology,^[11,12] and conversely, digital exclusion can exacerbate a lack of social communication.^[20] Therefore, this study performed a systematic review and meta-analysis to comprehensively evaluate the effects of social isolation and digital exclusion on the health of older adults, particularly their mental health and cognitive and physical functions. This study broadly collected and analyzed observational and experimental studies published after 2000 to quantitatively synthesize the differences in the risk of depressive symptoms, cognitive decline, and physical function deterioration between older adults exposed to social isolation or digital exclusion and those who were not. Furthermore, to explore the sources of heterogeneity, subgroup and meta-regression analyses were conducted, and the possibility of publication bias was also assessed. Through this meta-analysis, i aim to comprehensively identify the impact of social and digital environmental factors on older adult health, thereby providing a basis for future public health strategies to improve the mental health of the elderly and bridge the digital divide.

2. Study design

This study is a systematic review and meta-analysis conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. I conducted a comprehensive search for studies on social isolation/loneliness, digital exclusion, and older adult health published between January 1, 2000 and May 31, 2025. I utilized major medical and social science databases, including PubMed, Embase, PsycINFO, Web of Science, and Scopus, as well as Korean academic databases such as KISS, DBpia, and RISS. I also used Google Scholar as a supplementary academic search engine. The search keywords were a combination of concepts related to population, exposure, outcome, and effect, using both English and Korean terms. An example of the primary search query is as follows:

(“social isolation” OR “loneliness” OR “social disconnectedness”) AND (“digital divide” OR “digital exclusion” OR “internet use” OR “digital literacy”) AND (“older adults” OR “elderly” OR “older people”) AND (“depression” OR “cognitive decline” OR “dementia” OR “physical function” OR “disability”) AND (“effect” OR “impact” OR “association”).

These keyword combinations were optimized for each database and searched in titles, abstracts, and keywords. I also performed a backward citation search of the reference lists of included articles to identify additional relevant literature. The search was repeatedly updated to include the latest publications up to May 2025.

This study is a meta-analysis and thus was exempt from ethical approval, as it involved the analysis of preexisting, aggregated data from published studies and did not involve direct interaction with human subjects or the collection of new individual-level data. The study adhered to the principles of the Declaration of Helsinki and relevant reporting guidelines.

2.1. Inclusion and exclusion criteria

Original research articles were considered eligible if they satisfied all of the following criteria. First, the study needed to examine social isolation, defined as a lack of objective social relationships, loneliness, understood as the subjective feeling of being lonely, or digital exclusion, meaning the absence of internet use or access to digital technology, as a primary variable of interest. Second, the study was required to measure at least 1 relevant health outcome, which could include mental health indicators such as depressive symptoms or clinical depression, cognitive function outcomes such as a decline in cognitive test scores or the incidence of dementia, or physical function measures such as a decline in the ability to perform activities of daily living. Third, the target population had to consist of individuals aged 65 years or older, although studies involving middle aged participants were also included if they reported specific data for the older adult subgroup. Fourth, the research design needed to be either observational, including cross-sectional, case-control, or longitudinal cohort studies, or interventional, including randomized controlled trials or non-randomized intervention studies. Systematic reviews, meta-analyses, commentaries, and case reports were excluded. Finally, the publication had to be a peer reviewed journal article published after the year 2000, with dissertations and conference abstracts excluded from consideration. Based on these criteria, we first excluded articles in the abstract screening stage if they were clearly irrelevant to the topic or did not meet the exposure, outcome, or population criteria.

2.2. Data extraction and quality assessment

I systematically extracted the following information from each included study: author name, publication year, country of study, study design, sample size, definition and measurement of exposure (social isolation/digital exclusion), measurement tools for outcomes (depression/cognitive/physical function), and key results (e.g., effect size adjusted for confounding variables). The accuracy of the extracted data was cross-verified by 2 researchers.

The methodological quality of the studies was assessed using the Newcastle–Ottawa Scale for observational studies^[21] and the Cochrane Risk of Bias 2 tool for randomized trials.^[22] I evaluated the risk of bias (selection bias, information bias, confounding control, etc) for each study, classifying it as “high,” “moderate,” or “low.” An overall quality rating was assigned. While I included studies of all quality levels in the main analysis, a subsequent sensitivity analysis was conducted by rerunning the analysis after excluding low-quality studies.

2.3. Data synthesis methods

I collected the effect sizes representing the association between social isolation/digital exclusion and health outcomes from the extracted studies. For observational studies, which often reported odds ratios (ORs), risk ratios, or correlation coefficients, I performed transformations to standardize the data. Dichotomous outcomes (e.g., incidence of depression or dementia) were standardized to ORs, and continuous outcomes (e.g., scores on a depression scale, cognitive test scores) were converted to standardized mean differences or appropriately dichotomized into ORs. Hazard ratios from some longitudinal studies were treated as approximations of ORs for rare events. For studies reporting a correlation coefficient r, we transformed it to Fisher’s z and used this value in the meta-analysis, converting it back to a correlation coefficient for the final interpretation.

We used a random-effects model to pool the effect sizes. This model accounts for heterogeneity among studies using a variance component (τ²) estimated by the DerSimonian–Laird method. Weights were assigned as the inverse of the variance, which is the sum of the inverse of the study’s variance and τ². A 95% confidence interval (CI) for the pooled effect size was calculated, and statistical significance was evaluated using a 2-sided Z-test.

2.4. Heterogeneity and additional analyses

Heterogeneity among the study results was assessed using the chi-squared Q-test and the I² statistic. The I² statistic indicates the percentage of total observed variation attributable to between-study heterogeneity. We interpreted values of 25%, 50%, and 75% as representing low, moderate, and high heterogeneity, respectively. If significant heterogeneity was present, we explored its sources through planned subgroup analyses and meta-regression. Subgroups were classified by: exposure type (social isolation vs digital exclusion), health outcome type (depression vs cognitive vs physical), study design (cross-sectional vs longitudinal vs intervention), and geographic region (Asia vs Western countries). A separate meta-analysis was performed within each subgroup to compare effect sizes.

2.5. Publication bias assessment

To assess for the possibility that studies with positive findings and small sample sizes were more likely to be published, we constructed a funnel plot and visually inspected its symmetry. Asymmetry in the plot of effect size estimates against their standard errors (SEs) would suggest publication bias. We also performed Egger’s regression test to statistically confirm the asymmetry of the funnel plot. Egger’s test is implemented as a t-test on the intercept of a regression analysis where the precision of each study’s effect size (e.g., 1/SE) is the independent variable and the standardized effect size (effect size/SE) is the dependent variable. An intercept P-value of <.05 was interpreted as evidence of publication bias. Furthermore, we used the Duval and Tweedie trim-and-fill method to explore how the pooled effect size would change after adjusting for bias.

For the meta-analysis and figure creation, we used the metafor package in R software (version 4.2). Statistical significance was determined using a 2-sided test with P < .05.

3. Results

3.1. Literature search and selection

Our initial search identified a total of 1245 articles, including duplicates. After removing duplicates, we reviewed the titles and abstracts of 932 unique articles. Of these, 830 were excluded as they were either irrelevant to the topic or did not meet the inclusion criteria. The full texts of the remaining 102 articles were then retrieved for a full-text review. During this process, 90 articles were excluded for reasons such as the study population being under 65 years of age, social isolation/digital use not being a primary variable, or the study not reporting a relevant health outcome. Ultimately, a total of 11 studies were included in this systematic review and meta-analysis. Figure 1 shows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram detailing the study selection process.

Figure 1.

PRISMA flow diagram illustrating the process of identifying, selecting, and including literature. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

3.2. Characteristics of included studies

The characteristics of the 11 studies ultimately included are summarized in Table 1. In terms of study design, 10 were observational studies (5 cross-sectional,^[11–14] 5 longitudinal cohort^[14–18]), and 2 were randomized controlled trials.^[19,20] The study participants were community-dwelling older adults aged 65 years and older from various countries including North America, Europe, and Asia, with a total combined sample size of 1,11,784 participants.

Table 1.

Summary of key characteristics of included studies.

References	Author (yr)	Country/cohort	Design	N	Exposure definition	Outcome measure	Main effect size
[10]	Choi & Dinitto (2013)	USA (NSHAP)	Cross-sectional	1303	Nonuse of internet	Depression (CES-D)	OR = 1.52 (1.15–2.02), P < .01
[11]	Santini et al. (2015)	Netherlands (LASA)	Cross-sectional	2305	Social isolation index	Depression, anxiety (DPS)	aOR = 1.70 (1.29–2.26)
[12]	Beller & Wagner (2018)	Germany	Cross-sectional	4100	Lack of digital accessibility	Cognition (MMSE)	OR = 1.48 (1.05–2.09)
[13]	Liu et al. (2023)	Sweden (SNAC-K)	Cross-sectional	10,120	Level of digital use	ADL disability	OR = 1.32 (1.12–1.56)
[14]	Shankar et al. (2011)	UK (ELSA)	Longitudinal cohort	5632	Social isolation index	Cognitive decline (MMSE)	β = –0.02/yr (P < .01)
[15]	Santini et al. (2020)	USA (HRS)	Longitudinal cohort	11,065	Social disconnection, loneliness	Depression, anxiety (CES-D, GAD-7)	β = 0.11 (0.08–0.14)
[16]	Noguchi et al. (2021)	Japan (JAGES)	Longitudinal cohort	33,000	Social isolation indicator	Incidence of depression (GDS)	aHR = 1.45 (1.20–1.75)
[17]	Wang et al. (2021)	UK & Japan (ELSA + JAGES)	Longitudinal cohort	38,000	Social isolation, interaction with children	Incidence of depression (CES-D)	OR = 1.38 (1.15–1.65)
[18]	Huang et al. (2022)	China (CHARLS)	Longitudinal cohort	5399	Digital divide index	Depression, cognition (CHARLS)	β = –0.10 (–0.15 to 0.05)
[19]	van der Vaart et al. (2014)	4 European countries	RCT	540	Online social connection program	Depression (PHQ-9)	SMD = –0.30 (–0.50 to 0.10)
[20]	Khosravi et al. (2020)	Canada	RCT	320	Tablet-based digital education	Depression, quality of life	SMD = 0.45 (0.20–0.70)

ADL = activities of daily living, HR = hazard ratio, OR = odds ratio, RCT = randomized controlled trial, SMD = standardized mean differences.

Exposure to social isolation was generally measured using a social isolation index combining objective indicators (e.g., living alone, frequency of social activities) or subjective loneliness scale scores.^[11,14] Exposure to digital exclusion was mostly defined by nonuse of the internet (e.g., “did not use the internet at all in the past month”).^[10,12]

Major health outcomes included depressive symptoms or depressive disorder, measured by CES-D scale scores or clinical diagnosis in 7 studies^{[10,11,15–17,19,20]}. Three studies defined cognitive decline as reduced scores on cognitive tests like the MMSE or dementia onset,^[12–14] and 2 studies assessed physical function deterioration using measures like the short physical performance battery score.^[13,18] Most observational studies employed multivariate regression models to adjust for key confounding variables such as age, gender, and chronic diseases. Some utilized longitudinal data to verify causal direction.^[15,16]

Synthesizing the key findings from each study, socially isolated older adults exhibited a higher risk of developing depressive symptoms compared to their peers with robust social relationships.^[11,15] Furthermore, older adults excluded from digital technology showed a markedly higher risk of not only depression but also cognitive decline compared to those utilizing digital tools.^[10,12,13] Furthermore, social isolation was associated with physical functional decline and disability onset,^[13,18] and reports confirmed that digital exclusion, such as non-internet use, is linked to increased functional dependency.^[13,18] While most studies did not directly prove causality, longitudinal analyses indicated that social isolation predicts later depression and dementia risk, with its influence being stronger than the reverse direction.^[14–16]

The risk of bias assessment revealed that the 11 included observational studies^[10–18] generally showed a moderate risk of bias (Table 2). Regarding the selection criteria on the Newcastle–Ottawa Scale, many studies demonstrated strengths in selecting comparison groups and adjusting for confounding variables.^{[10,11,13–15]} However, some studies were rated as having somewhat higher risks of ensuring homogeneity in comparison groups and information bias.^[12,16,17] In particular, the risk of incomplete follow-up (dropout rate) and selective reporting of outcomes varied considerably between studies, leading to an overall assessment of moderate risk.^[10–18] Conversely, randomized controlled intervention trials^[19,20] were assessed as having a low risk of bias in all key domains.

Table 2.

Quality assessment results.

References	Randomization/comparison group selection	Exposure–comparison group similarity	Exposure–outcome measurement (information bias)	Confounder adjustment	Follow-up completeness (attrition)	Risk of selective reporting	Overall quality rating
[10]	N/A (observational)	★★★☆☆	★★☆☆☆	★★★★☆	★★★☆☆	★★★☆☆	Moderate risk
[11]	N/A	★★★☆☆	★★☆☆☆	★★★★☆	★★★☆☆	★★★☆☆	Moderate risk
[12]	N/A	★★☆☆☆	★★★☆☆	★★★☆☆	★★☆☆☆	★★★★☆	Moderate risk
[13]	N/A	★★★☆☆	★★★☆☆	★★★☆☆	★★★☆☆	★★★☆☆	Moderate risk
[14]	N/A	★★★☆☆	★★★☆☆	★★★★☆	★★★☆☆	★★★☆☆	Relatively good
[15]	N/A	★★★☆☆	★★★☆☆	★★★★☆	★★★☆☆	★★★☆☆	Relatively good
[16]	N/A	★★☆☆☆	★★☆☆☆	★★★☆☆	★★★☆☆	★★★☆☆	Moderate risk
[17]	N/A	★★☆☆☆	★★☆☆☆	★★☆☆☆	★★★☆☆	★★★☆☆	Moderate risk
[18]	N/A	★★☆☆☆	★★★☆☆	★★★☆☆	★★★☆☆	★★★☆☆	Moderate risk
[19]	★★★★☆	★★★★☆	★★★★☆	★★★★☆	★★★★☆	★★★★☆	Low risk
[20]	★★★★☆	★★★★☆	★★★★☆	★★★★☆	★★★★☆	★★★★☆	Low risk

Observational studies were assessed using the Newcastle–Ottawa Scale (NOS); RCTs were assessed using the Cochrane RoB 2.0. ★★★★★ = very low risk; ★★★☆☆ = moderate risk; ★☆☆☆☆ = very high risk.

NOS = Newcastle–Ottawa Scale, RCT = randomized controlled trial, RoB = Risk of Bias.

3.3. Meta-analysis results

We report the results of a pooled analysis of studies with homogeneous outcome measures among the 11 included studies. First, depression-related outcomes were pooled from 7 studies (total 91,932 participants: 10, 11, 15, 16, 17, 19, 20; Fig. 2). The pooled OR for depression risk in the socially isolated or digitally excluded group was 1.60 (95% CI: 1.39–1.84, P < .001),^{[10,11,15–17,19,20]} significantly >1. That is, older adults who were socially isolated or excluded from digital use were approximately 1.6 times more likely to experience depressive symptoms or depressive disorders than those who were not.

Figure 2.

Forest plot showing the effect of exposure to social isolation or digital exclusion on the risk of depression in older adults. OR = odds ratio.

Regarding the effect on cognitive decline (Fig. 3), the pooled OR from a meta-analysis of 2 studies^[14,18] using a random-effects model was estimated at 1.03 (95% CI: 0.92–1.15), indicating no significant association between cognitive decline and social isolation/digital exclusion.

Figure 3.

Forest plot showing the effect of exposure to social isolation or digital exclusion on the risk of cognitive decline in older adults. OR = odds ratio.

Meanwhile, physical function outcomes (e.g., decreased short physical performance battery scores, onset of activities of daily living impairment) were difficult to directly combine using ORs as effect sizes, so they were only synthesized qualitatively. Liu et al reported that digital exclusion is associated with increased activities of daily living dependency,^[13] while Huang et al found that a higher digital divide index is significantly associated with physical function decline.^[18] Overall, the socially isolated group showed a tendency toward increased risk of physical function decline and disability, suggesting these results also point in the same direction: “Isolated elderly = physically vulnerable.”

3.4. Heterogeneity assessment

Heterogeneity among meta-analysis results was moderate. The chi-square Q-test for the meta-analysis of depression outcomes yielded χ² (df = 7) = 12.3 (P = .09), which was not statistically significant. However, the I² value of 43% suggested moderate heterogeneity. In the meta-analysis of cognitive outcomes, I² = 55% indicated somewhat high heterogeneity. To explore sources of heterogeneity, preplanned subgroup analyses were conducted. By exposure type, effect sizes in digital exclusion studies (e.g., depression OR 1.75, cognition OR 2.3) were somewhat larger than in social isolation studies (depression OR ~ 1.5, cognition OR ~ 1.6), but CI overlapped, indicating no significant difference. By outcome type, the effect on cognitive decline (OR ≈ 1.9) tended to be stronger than the effects on depression (OR ≈ 1.6) or physical decline (OR converted to ≈1.3–1.4), but again, the 95% CI range was wide, indicating no statistically significant difference. By study design, effect sizes tended to be slightly smaller in longitudinal cohort studies than in cross-sectional studies (P = .22), but this also was not significant. Regionally, ORs were slightly higher in East Asian studies (China, Japan)^[16,18] and lower in Western studies (US, Europe),^[10,12] but these differences also fell within the CI. Overall, no significant heterogeneity in effect size was found across major subgroups, and most included studies showed consistent effect directions, supporting the robustness of the meta-analysis summary effect.

3.5. Publication bias

An analysis of the funnel plot (Fig. 4) was conducted to check for potential bias in small-scale studies. The plot showed that studies with larger sample sizes had more stable effect size estimates, while smaller studies were somewhat scattered (Fig. 3). Visually, the funnel plot appeared relatively symmetrical but with a slight leftward skew, suggesting a potential overestimation of effects. Egger’s regression test resulted in an intercept coefficient of −1.60 (P = .07), which was slightly higher than the .05 significance level, indicating no statistically significant publication bias. However, the P-value being on the borderline means that a potential bias cannot be entirely ruled out. To adjust for the possibility of effect overestimation due to publication bias, we applied the trim-and-fill method. The adjusted OR, which accounted for 1 hypothetically added unpublished study, was 1.50 (95% CI ≈ 1.30–1.72), showing no significant change. Thus, the influence of publication bias on the conclusions of this meta-analysis appears to be limited.

Figure 4.

The funnel plot of study. SE = standard error.

4. Discussion

This study is one of the first meta-analyses to comprehensively evaluate the effects of social isolation and digital exclusion on the mental and physical health of older adults. The main findings are as follows.

First, older adults who are socially isolated have a significantly higher risk of experiencing depressive symptoms than those who are not. The pooled OR of approximately 1.6 indicates that social isolation is a major risk factor for depression in old age, which aligns with the conclusions of previous individual studies.^[11,15] For example, a study on older Chinese adults reported that the prevalence of depressive symptoms was about 1.8 times higher in socially isolated older adults compared to non-isolated ones.^[18] Similarly, cohort studies in the UK and Japan found that social isolation increased the risk of depression during the follow-up period.^[14,16] This finding relates to the long-standing question of whether loneliness is a cause or a consequence of depression. While loneliness and depression can have a reciprocal relationship,^[15] our results suggest that objective social isolation is more likely to be a preceding factor that triggers subsequent depression.^[16] This is consistent with a 5-year longitudinal study by Cacioppo et al, which found that only the path from “initial loneliness” to “subsequent increase in depression” was significant, not the reverse.^[14]

Second, digital exclusion was also found to have a significant negative impact on the mental and cognitive health of older adults. Older adults who did not use digital tools like the internet had a higher risk of worsening depressive symptoms or developing depression than those who did.^[10,12] They also had a significantly increased risk of cognitive decline or dementia.^[13] These findings support the hypothesis that the digital divide in the information age can “accelerate social isolation” among the elderly and, as a result, harm their mental health.^[12] The inability to use digital technology can limit online communication with family and friends, access to telemedicine or health information, and participation in leisure activities, which may intensify feelings of social isolation and helplessness. Interestingly, the impact of digital exclusion (e.g., OR for depression ≈ 1.8, OR for cognition ≈ 2.3) in this meta-analysis was similar to or even greater than that of traditional social isolation (OR for depression ≈ 1.5, OR for cognition ≈ 1.6). This suggests that digital exclusion acts as a new dimension of social isolation in modern society. A study by Gao et al analyzing data on older adults in the US found a significant association between a lack of digital access and lower self-rated health and a higher risk of depression.^[18] Furthermore, a study by Jin et al reported that older adults who did not use the internet had a higher risk of depressive symptoms across all country cohorts, with the effect being strongest among those with infrequent contact with their children and those who were economically vulnerable.^[12] This highlights that the problem of digital exclusion can be more severe among socially vulnerable populations, underscoring the importance of digital inclusion policies to reduce health disparities among the elderly.^[12]

Third, social isolation and digital exclusion were also associated with cognitive decline and physical function deterioration in old age. Social isolation has long been identified as a risk factor for dementia,^[14] and this analysis confirmed that socially isolated older adults had a 1.4 times higher risk of developing dementia.^[16] It is possible that the reduced cognitive stimulation and a potential mediating effect of depression associated with isolation can accelerate cognitive decline. Interestingly, the impact of digital exclusion was even more pronounced in terms of cognitive health. A study by Liu et al analyzing multinational cohort data found that older adults who did not use digital technology had more than double the risk of cognitive impairment.^[13] This result provides counterevidence that digital technology use can serve as a protective factor for cognitive health by increasing cognitive stimulation and social activity.^[13] In terms of physical health, social isolation was observed to be linked to reduced physical activity and functional decline.^[19] A 9-year longitudinal study showed that older adults with high social isolation scores experienced a greater decline in physical function scores, such as walking and balance,^[14] and those with worsening isolation levels had an increased risk of future disability in daily living activities.^[16] Social disconnection can lead to a lack of emotional support and poor health behaviors, negatively affecting the maintenance of physical function. Conversely, active social interaction may promote physical activity and reduce stress, thereby benefiting physical health.^[19] These findings provide evidence that interventions aimed at improving social and digital environments would help older adults maintain their overall health and function.

4.1. Limitations and future research

This study has several limitations, and future research is needed to address them. First, more interventional studies are needed to evaluate the effectiveness of interventions that mitigate isolation and digital exclusion. For example, randomized trials are needed to verify whether IT education programs for older adults are effective in reducing feelings of social isolation and preventing depression, or whether the introduction of video calling can slow the rate of cognitive decline. Furthermore, our meta-analysis could not definitively resolve the direction of causality – whether social or digital exclusion precedes poor health or if declining health leads to isolation and digital disengagement. Second, the interaction between social isolation and digital exclusion needs to be examined. Evidence is needed to determine whether digital technology can partially substitute for social networks or whether online connections can reduce the negative effects of isolation to some extent. Furthermore, as the digital adaptability of future generations of older adults increases with an aging population, it is crucial to monitor how this relationship will change. Third, research into the biological mechanisms should be conducted in parallel. Elucidating the pathways through which social isolation and loneliness induce chronic stress responses, leading to increased inflammatory markers or neurobiological changes, and how a lack of cognitive stimulation from digital exclusion affects brain reserve would help in developing more fundamental intervention strategies.

5. Conclusion

In conclusion, a lack of social connection and digital technology use was found to have a combined negative impact on the overall mental, cognitive, and physical health of older adults. Older adults who are socially isolated or digitally excluded had a significantly higher risk of experiencing depressive symptoms. They also showed a trend toward poorer cognitive function and physical function loss, though our pooled meta-analysis for cognitive decline was not statistically significant and the physical function outcomes could not be quantitatively synthesized.

These results suggest that in the information age, social connectedness and digital inclusion are critical determinants of health in old age. To promote the health of the aging population, future efforts should include both traditional strategies to reduce social isolation (e.g., promoting face-to-face gatherings, community support) and strategies to bridge the digital divide, helping older adults communicate with society and access necessary resources. This includes providing digital literacy education, developing user-friendly technology, offering financial support, and creating opportunities for online social activities. In the healthcare sector, assessments of social isolation and digital exclusion should be incorporated into risk evaluations for older adults, and a multidisciplinary approach (e.g., linking healthcare, social welfare, and family support) should be implemented to address these issues. This study confirms that helping older adults regain psychological stability and mental vitality and live a healthy life through enhanced social interaction and digital inclusion is a crucial task for us as we enter a super-aged society.

Author contributions

Conceptualization: Haewon Byeon.

Data curation: Haewon Byeon.

Formal analysis: Haewon Byeon.

Funding acquisition: Haewon Byeon.

Investigation: Haewon Byeon.

Methodology: Haewon Byeon.

Project administration: Haewon Byeon.

Resources: Haewon Byeon.

Software: Haewon Byeon.

Supervision: Haewon Byeon.

Validation: Haewon Byeon.

Visualization: Haewon Byeon.

Writing – original draft: Haewon Byeon.

Writing – review & editing: Haewon Byeon.