Bridging the digital health divide: digital endowment, informal social participation, and health inequality among older adults

Abstract

Background

Digital technology, as a new quality productive force, offers novel approaches to actively address population aging. The development and enhancement of digital endowment among older adults not only affects their physical and mental health but also serves as a crucial driver for advancing the construction of a digital society and mitigating health inequality.

Methods

Based on data from the 2020 China Health and Retirement Longitudinal Study (CHARLS), this research examines the impact of digital endowment on health inequality among older adults from a micro-individual perspective, along with the mediating role of informal social interaction.

Results

First, digital endowment demonstrated a significant positive association with health status and a consistent negative relationship with the degree of health inequality among older adults. These results were corroborated by multiple robustness tests. Second, the connection between digital endowment and health disparities displayed structural heterogeneity, with stronger negative associations observed among women, individuals with primary education, and older adults residing in central China. Third, mediation analysis revealed that self-support and family intergenerational support played significant mediating roles. The self-support pathway was linked to reduced health disparities, while the family intergenerational support pathway exhibited an opposing pattern. Peer support showed no statistically significant mediating effect.

Conclusion

Guided by the principle of digital inclusion, efforts should focus on equalizing digital infrastructure construction and providing age-appropriate digital skills training to enhance the digital endowment of older adults. Emphasis should be placed on leveraging peer influence and encouraging digital reverse mentoring to build online-offline social networks. Health forums and mutual aid communities can strengthen the accumulation of informal social support for older adults. Concurrently, differentiated intervention strategies for urban/rural areas and regions should be implemented, alongside promoting the interconnectivity and sharing of digital health resources. This will systematically bridge the digital divide among older adults, accelerate the holistic development of Digital China, and ensure universal access to digital dividends.

Clinical trial number

Not applicable.

Introduction

In recent years, structural demographic contradictions centered on population aging have become increasingly prominent worldwide. According to the 2024 National Aging Development Bulletin released in August 2025, China’s population aged 60 and above reached 310.31 million by the end of 2024, accounting for 22.0% of the total population [1]. This proportion is projected to exceed 30% by 2035, marking China’s transition into a super-aged society. The older adult population exhibits distinct characteristics including empty-nesting, functional disability, and multi-morbidity, contributing to a substantial rise in socioeconomic burdens associated with aging-related health challenges [2]. Amid the era of longevity, health inequality within the older adult population has emerged as a critical concern, posing a major obstacle to achieving China’s “Healthy Aging” strategic objectives [3, 4]. Attaining health equity constitutes a national strategic priority for universal health. To address this, China has implemented multiple initiatives to reduce health disparities. For instance, the “Healthy China 2030” Blueprint (2016) explicitly calls for “significantly improving health equity,” while the Healthy China Action Plan (2019–2030) (2019) targets the “fundamental achievement of health equity by 2030.” Nevertheless, influenced by individual characteristics, healthcare resource allocation, regional economic development, and public policies, health inequality among older adults persists across age groups and social strata [5]. Under the goal of common prosperity, both the health status of residents (including the older adults) and the fairness of its distribution require further improvement.

With the development of network communication technologies and the advent of the digital era, using the Internet to participate in social life, engage in interactions, pursue leisure activities, and seek information has become a new norm in daily life. According to the 55th Statistical Report on Internet Development in China, China had 156 million internet users aged 60+ by 2024-end, representing 49.9% of its older adult population [6]. Digital-health integration now drives innovation in medical services, unlocking new possibilities for older adults’ lives. Yet older adults face greater challenges accessing digital resources than other groups, causing a new social inequality: the “digital divide” [7]. This gap is particularly pronounced among the older adults due to limited internet access, complexity of digital devices, and slower adoption of new technologies [8]. To bridge the digital divide for seniors, China has issued policy documents including the Implementation Plan for Effectively Addressing Challenges Faced by Seniors in Using Smart Technologies (Guo Ban Fa [2020] No. 45), Guidelines on Promoting the Healthy Development of Elderly and Childcare Services (Guo Ban Fa [2020] No. 52), and Opinions on Developing the Silver Economy to Enhance Elderly Well-being (Guo Ban Fa [2024] No. 1), all emphasizing the elimination of digital barriers for older adults, advancement of smart elderly care solutions, and leveraging digital means to improve health outcomes and overall well-being in aging populations.

Although existing research has extensively examined digitalization’s relationship with health outcomes among older adults, most scholars have operationalized this concept through unidimensional behavioral measures such as internet usage frequency [9–12], specific application use [13, 14], mobile device adoption [15], or social integration [16]. This simplified measurement approach fails to capture the multidimensional nature and structural impact of digitalization in older adults’ lives, particularly limiting our understanding of its underlying mechanisms in shaping health equity. To address this limitation, this study introduces the comprehensive concept of “digital endowment”, which we define as the totality of digital resources and capabilities that individuals possess and utilize in the digital era, encompassing access, skills, and application. This concept finds its theoretical roots in human capital theory [17], conceptualizing digital competence as a form of capital that can be accumulated through investment and yields returns in well-being. From a sociological perspective, digital endowment influences individuals’ and groups’ positions and relationships within digital society [18]. Those with substantial digital endowment enjoy advantages in information access and resource allocation, consequently shaping social structures and developmental trajectories in the digital realm. Among older adults, disparities in digital endowment extend far beyond mere device ownership, arising from complex factors including connectivity barriers, skill deficiencies, cognitive differences, and the digital divide itself. These variations profoundly reshape health information acquisition, healthcare service accessibility, health management efficacy, and social participation quality, establishing digital endowment as a crucial yet underexplored perspective for understanding health inequality in this population [18, 19].

Current research shows no consensus on whether digitalization reduces health inequality. Some scholars argue it primarily benefits high-socioeconomic-status groups, thereby exacerbating health disparities [20, 21]. Others contend that digital development narrows gaps in healthcare access and health information acquisition between privileged and disadvantaged groups, ultimately reducing health inequities [3]. As noted in the UN Human Development Report, the internet creates a two-tiered system: one offering low-cost, high-speed services to affluent, educated groups; the other burdened by time, cost, and information barriers for those relying on outdated resources [22]. While digitalization may improve older adults’ health, it risks reinforcing existing health inequalities [3, 23]. Meanwhile, informal social support (non-institutional assistance from family, friends, and neighbors) remains a protective factor for the health of older adults. Unlike younger groups, older adults prioritize strong ties with family and friends over weak-tie networks [24]. The informal support formed through older adults’ personal networks is fundamentally crucial, though its effects are uneven. Older adults with limited informal social connections typically have smaller networks and fewer resources [25]. Digital tools may expand some older adults’ social networks and support efficiency, while simultaneously worsening social isolation and support loss for others due to the digital divide. This creates a vicious cycle of digital disadvantage and social deprivation [26]. The impact of digital technology on older adults extends beyond the transformation of external social networks. It may also foster health self-management efficacy by empowering individuals, thereby emerging as another critical pathway influencing health inequality.

Against the backdrop of accelerating aging and rapid digitalization, investigating whether digital endowment exacerbates pre-existing health inequalities among older adults in China holds significant relevance. This study leverages 2020 China Health and Retirement Longitudinal Study (CHARLS) data to examine the mechanisms and pathways through which digital endowment affects health inequality in older adults from a micro-level perspective. Our analytical approach proceeds as follows: First, we construct a multidimensional digital endowment index system covering access, skills, and application to empirically examine its overall association with health status and health disparities among older adults. Second, we conduct an in-depth analysis of heterogeneity across the four dimensions of gender, education level, urban-rural residence, and geographic region. Finally, we develop a mediation model to test the intermediary roles of both the internal empowerment pathway of self-support and the external support pathway of informal family and peer interactions within the association mechanism. The research aims to provide novel theoretical perspectives for understanding health disparities among older adults in the digital era, while offering evidence-based support for formulating precisely targeted digital inclusion policies for aging populations.

Literature review and research hypotheses

Amid rapid digitalization, scholars have extensively explored how enhancing older adults’ digital endowment can bridge the digital divide, improve health outcomes, and enable fuller participation in digital benefits.

Conceptualization and measurement of digital endowment

Digital endowment extends human capital theory into the digital era, integrating multiple theoretical traditions while anchoring in individual-level capital theories. The concept draws inspiration from resource endowment theory at a macro level, where digital infrastructure disparities create a “digital resource endowment” across regions (Heckscher & Ohlin, 1993). At the micro level, it builds on human capital theory (Schultz, 1961; Becker, 1975) [27, 28] and capital pluralism (Bourdieu, 1987) [17]. While traditional human capital encompasses knowledge, skills, and health, in digital society, the ability to access, master, and effectively use digital technologies has become a crucial new form of “digital human capital”. This form of capital can be accumulated through investment (such as learning and practice) and can significantly enhance individual well-being (including health outcomes) [29]. Bourdieu’s capital theory expands capital forms to include economic, cultural, and social capital. Digital endowment can be viewed as a composite that integrates multiple forms of capital: digital access (possessing devices and connectivity) carries economic capital attributes; digital skills (knowledge and abilities) represent cultural capital; while digital application (using networks to maintain social connections) can create and transform social capital. This study operationalizes digital endowment as the comprehensive digital resources and capabilities that individuals possess and can mobilize in the digital era, encompassing access, skills, and application dimensions. It emphasizes digital endowment as a form of micro-level individual capital that can be accumulated, invested in, and yield returns, rather than as a macro-level geographical endowment.

This study systematically differentiates digital endowment from related but distinct concepts. The digital divide operates at the macro level, describing structural inequalities in digital access, use, and benefits across social groups [30]. In contrast, digital endowment functions at the micro level, measuring an individual’s digital resource level. These concepts maintain a causal relationship: uneven digital endowment distribution constitutes the micro-level cause of the macro-level digital divide. Digital literacy/skills, while related to digital endowment, differ in scope. Digital literacy refers specifically to cognitive and technical capabilities for effective technology use [31]. Within our framework, digital skills represent one core dimension of digital endowment, which encompasses broader aspects including digital access and application. Compared to digital capital, digital endowment shows different emphasis. Digital capital treats digital resources as tools for social advancement, emphasizing competition and social reproduction [32, 33]. While digital endowment forms the foundation for digital capital, this study conceptualizes it primarily as a resource reservoir affecting health and well-being, focusing on direct welfare effects rather than social competition.

Digital endowment and elderly health inequality

Digital endowment and elderly health

Older adults exhibit greater health vulnerability, often experiencing conditions such as disease susceptibility, cognitive decline, and physical frailty, resulting in generally poorer health compared to younger and middle-aged populations. Existing research has provided valuable insights into the relationship between digitalization and health among older adults, with most findings supporting a health-enhancing perspective that digital technologies can promote both physical and mental well-being. According to health promotion theory, residents in areas with higher internet penetration tend to possess stronger psychological, physiological, and medical decision-making capabilities [23]. The use of smart wearable devices enables older adults to maximize the benefits of these technologies in health management, thereby improving their health outcomes [34]. Digital technologies, particularly internet-based platforms, positively influence disease prevention among older adults by providing accessible health consultation services [10]. Activity Theory suggests that digital communication technologies serve as channels for social participation, helping to counter social disengagement in later life [35]. Studies indicate that maintaining connections with family and friends through the internet can effectively enhance mental health among older adults [36]. Theories of Resocialization emphasize that the internet facilitates successful adaptation to both aging and digital society, stimulating individuals’ agency in the resocialization process [37]. Through various functions including health consultation, cognitive training, interpersonal communication, and recreational activities, digital technologies contribute to improved psychological health and cognitive functioning among older adults. Additionally, smart devices provide valuable assistance to those with mobility limitations [38]. Grounded in these theoretical perspectives, we posit that digitalization is associated with the health status of older adults.

Consequently, we propose:

Hypothesis 1

Digital endowment is positively associated with health status among older adults.

Digital endowment and elderly health inequality

Substantial digital divides exist within the older adult population, as unequal opportunities and capabilities to use the internet lead to varying degrees of health benefits obtained through digital means. Such differential access contributes to disparities in healthcare utilization and consequently exacerbates health inequality among older adults [39]. According to the life course perspective, contemporary Chinese older adults have experienced both the planned economy era and economic reforms. The long-term cumulative effects from their earlier life stages create stronger heterogeneity within this age group compared to others, forming the fundamental source of disparities in accessing internet-based health promotion benefits [40]. Due to Matthew effects, older adults with higher education and urban residence tend to accumulate digital skills more easily, enabling more efficient utilization of health resources to amplify health advantages, thereby potentially widening health disparities [20]. As an emerging social determinant of health, the unequal distribution of digital endowment (digital divide) significantly influences both health outcomes and health equity [39].

In the digital economy era, digital technologies demonstrate inclusive and equitable characteristics that can effectively mitigate the negative impact of socioeconomic factors on health inequality [41]. Digital technologies partially overcome physical barriers such as geographical distance, transportation limitations, and mobility constraints, making basic healthcare and health management support more accessible to disadvantaged groups. For older adults with relatively scarce resources, improvements in digital endowment may yield relatively greater marginal health benefits, potentially alleviating health inequality within this population [23]. Research indicates that digital development significantly improves living conditions in rural areas through income enhancement and healthcare service improvement [42]. Jia and colleagues demonstrated that internet-based healthcare services help alleviate gender-based inequality in health opportunities in rural contexts [43]. Simultaneously, social interactions and healthcare resources accessed through mobile devices help alleviate depressive symptoms and loneliness among older adults, gradually narrowing health disparities between older adults and the general population [44].

Overall, digital technologies possess non-rivalrous and non-excludable characteristics. The network-driven flow of information and dissemination of health-related content effectively meets the needs of vulnerable groups, contributing to the reduction of health inequality across different populations. Based on this theoretical and empirical foundation, we propose Hypothesis 2.

Hypothesis 2

Digital endowment is negatively associated with health inequality among older adults.

Dual-pathway mechanisms linking digital endowment and health inequality: informal social engagement and self-support

External pathway: informal social engagement, digital endowment, and health inequality among older adults

Social support functions dually by alleviating burdens and strengthening resources, helping individuals navigate difficulties while enhancing their coping capacities. Research indicates disparities in social support represent a significant manifestation of digital inequality [45]. Informal support fulfills critical needs for emergency assistance, emotional exchange, and psychological security, forming the primary psychological anchor for older adults. Only through the organic integration of formal and informal support can a reliable security system for high-quality living be established for the aging population [24]. Deeply influenced by Confucian culture, China prioritizes interpersonal relationships, forming a unique relationship-based social fabric where social capital predominantly resides in familial and friendship networks. Network enhancement theory further posits that the internet helps older adults maintain and expand social connections, boosting participation, enriching social capital, and ultimately improving life satisfaction and health [46]. Existing literature demonstrates that informal support systems mitigate uneven digital empowerment across groups through digital access, skills, and integration. Vulnerable older adults compensate for digital or health literacy deficits through social ties, thereby mediating health inequality reduction [18]. This study examines the role of informal support in the relationship between digital endowment and health inequality among older adults through two specific dimensions: family support and peer support.

(1) Intergenerational Family Support

Intergenerational relationships are defined as vertical kinship ties (including adoptive relations) among family members. Within China’s Confucian-influenced East Asian context, adult children constitute the core social support network for middle-aged and older adults, rendering intergenerational family support the predominant form of elderly social sustenance [47]. Offspring-to-parent support encompasses three dimensions: emotional comfort, financial provision, and daily caregiving. Synthesizing existing research, digitalization exerts dual effects on older adults intergenerational support: health enhancement and perceptual shift. First, the health enhancement effect manifests as internet usage facilitating social engagement, recreational activities, and health knowledge acquisition among the older adults, thereby promoting self-health management and improving physical/mental well-being [48]. Simultaneously, digital connectivity transcends spatial barriers, redefining interpersonal bonds and intergenerational communication. This increases online interaction frequency between parents and children, enhancing emotional support for the older adults [49]. Consequently, expenditures on elderly care services may decline, potentially reducing upward financial transfers and time-intensive caregiving burdens on offspring [50]. Second, the perceptual shift effect reflects that internet exposure gradually reshapes consumption behaviors and attitudes among the older adults, increasing acceptance of formal care services. This could diminish their reliance on filial caregiving as a preferred aging strategy [51]. Yu et al. integrated intergenerational support into the framework of health opportunity inequality, revealing bidirectional reinforcement between overall older adults’ health and intergenerational support. Financial transfers from offspring were identified as a driver of health opportunity disparities [52]. Divergent theoretical expectations from Resource Substitution Theory versus Resource Reinforcement Theory indicate the internet’s health impact varies across older adults based on preexisting resource endowments [53]. Thus, elevated digital endowment may reduce intergenerational family support, thereby further exacerbating health inequality within the elderly population. This leads to Hypothesis 3:

Hypothesis 3

The relationship between digital endowment and health inequality is mediated by reduced family intergenerational support, which exacerbates health disparities.

(2) Peer support

Peer support refers to knowledge sharing and experiential transfer among older adults with similar ages, living environments, life experiences, and values, emphasizing interactions within peer groups [54]. The Social Escort and Socioemotional Selectivity Theory posit that older adults’ social networks gradually constrict inward from peripheral ties [46]. The Social Compensation Effect Theory highlights digital tools’ capacity to maintain existing social relationships and fostering communication and cooperation among individuals, thereby establishing stronger connections. Within the paradigm shift toward networked individualism, information technologies enable the older adults to expand their social networks outward from inner circles. Empirical studies indicate older adult users of short-form video platforms maintain cross-regional peer networks more effectively, counteracting the decline of peer-based social connections while significantly increasing interaction frequency and network density [55, 56]. Expanded peer networks enhance belongingness and health information accessibility. This peer-derived social embeddedness not only provides emotional sustenance but also reduces information asymmetry, improving the efficiency and willingness to adopt health behaviors [57]. Structural health disparities often originate from resource disadvantages and social exclusion among vulnerable groups. Digitally facilitated peer networks and group belongingness counteract marginalization by providing substitutive social resources to information-disadvantaged older adults, thereby attenuating health disparities [58]. Consequently, we propose Hypothesis 4:

Hypothesis 4

Digital endowment is associated with reduced health inequality through expanded peer support among older adults.

Internal pathway: self-support, digital endowment, and health inequality among older adults

Self-support denotes a state of sufficient personal income and independent living capacity. It functions as an intrinsic motivational factor, enabling individuals to recognize their strengths and advantages to inspire initiative and action when confronting adversity [24]. Research indicates that when the older adults successfully accomplish tasks through digital means, it enhances their self-identity, sense of control over life, and information agency, thereby demonstrating stronger self-support capacity [59, 60]. According to the New Human Capital Theory and the Uses and Gratifications Theory, utilizing digital skills increases the accumulation of new human capital, consequently strengthening behavioral confidence [61]. The Health Belief Model posits that perceived capability and self-cognition directly influence the adoption of health behaviors. Enhanced digital endowment thus boosts the older adults’ self-drive, internalizing it into tangible health actions [62]. Within traditional frameworks of health inequality, disadvantaged populations frequently experience poorer health outcomes due to deficits in resources, cognition, and capabilities. As an intrinsic regulatory mechanism, self-support motivates individuals to proactively acquire external resources and resist risks, thereby buffering the deterministic impact of social structures on health outcomes. This mitigating effect is particularly pronounced among vulnerable groups [62]. Consequently, self-support partially offsets deficiencies in socioeconomic resources, structurally alleviating health inequality. Based on this theoretical foundation, we propose the following hypothesis 5:

Hypothesis 5

Digital endowment is related to lower health inequality through enhanced self-support capabilities in older adults.

The conceptual pathways linking digital endowment to old adults’ health inequality are illustrated in Fig. 1.

Fig. 1.

Impact pathways of digital endowment on old adults’ health inequality

Econometric models and data description

Data source

This study employs data from the 2020 China Health and Retirement Longitudinal Study (CHARLS). CHARLS is jointly administered by Peking University and Wuhan University. Its sample covers over 10,000 households across 150 county-level districts and 450 village-level units in China. This provides strong representation of China’s elderly population [63]. The questionnaire encompasses individual demographics, family structure and economic support, health status and healthcare utilization (including medical insurance), employment and retirement status, pension security, as well as income, expenditure, and asset profiles. It contains variables on old respondents’ general information and physical/mental health. The survey also includes questions on residents’ internet usage patterns. These are particularly relevant to this research. The analytical sample was restricted to individuals aged ≥60 years. We excluded respondents under age 60. After removing cases with missing or invalid variables, the final sample contains 7904 valid observations. This study examines theoretical relationships between variables rather than producing population descriptions. Following the methodological approach outlined by Solon et al. [64], we use unweighted estimation since our model fully controls for sampling-related covariates including region, residence, age, and gender.

Variable selection

Dependent Variable

(1) Health Level

This study examines how digital endowment broadly influences older adults’ overall health. Given that health is a multidimensional construct encompassing physical, psychological, and cognitive domains, relying on a single health dimension cannot fully capture the comprehensive health benefits of digital technologies or adequately reflect health inequality. Therefore, we developed a composite health index that better aligns with our theoretical objectives by integrating multiple health aspects. Following Ouyang et al. [65] and leveraging CHARLS data availability, we constructed a health indicator system covering five dimensions: self-rated health, disability status, mental health, cognitive function, and chronic conditions. We adopted the entropy weight method from Zhang et al. (2010) [66] as a methodological reference, as their systematic evaluation framework provides a clear, established approach for composite index construction. Our research context shares similarities in needing to synthesize multiple health indicators with different measurement units into a unified index. While we followed their computational framework, we specifically recalculated entropy values and indicator weights using our sample data to ensure appropriate weighting scheme for this study. The procedure involves: normalizing individual indicators, calculating their entropy values and weights, then computing composite scores using the linear weighted aggregation formula. The calculation formula is as follows:

1

2

3

4

Where i denotes individuals, j denotes individual health indicators, N represents sample size, h_ij indicates the j-th health indicator, X_ij denotes the normalized score of the j-th health indicator for individual i, Q_j signifies the entropy value of the j-th health indicator, and W_j represents the weight of the j-th health indicator. Health indicator definitions and entropy weight method weights are shown in Table 1.

Table 1.

Measurement indicator system for health status

Health Indicator	Indicator Definition	Weight
Self-Rated Health	Assessed based on the question: “How would you rate your current health status?”Responses of “Very Good”, “Good”, or “Fair” are assigned a value of 1; responses of “Poor” or “Very Poor” are assigned a value of 0.	13.02%
Disability Status	Measured using the Activities of Daily Living (ADL) scale and the Instrumental Activities of Daily Living (IADL) scale. ADL items include: dressing, bathing, eating, getting in/out of bed, toileting, and controlling bowel and bladder. IADL items include: doing housework, cooking, shopping, making phone calls, taking medication, and managing finances. An individual is classified as non-disabled (assigned 1) if reporting “No Difficulty” for all items; otherwise, they are classified as disabled (assigned 0).	20.55%
Mental Health	Measured using the Center for Epidemiological Studies Depression Scale (CES-D). The scale comprises 8 items assessing negative affect and 2 items assessing positive affect. Total Score Range: 0–30. A higher score indicates a higher level of depressive symptoms. Individuals scoring ≥10 are classified as having depressive symptoms (assigned 1); those scoring < 10 are classified as having intact mental health (assigned 0).	19.74%
Cognitive Function	Assessed using the modified Telephone Interview for Cognitive Status (TICS) survey. The TICS evaluates domains including general knowledge, calculation ability, memory, recall, and visuospatial ability. Total Score: 31 points. Participants are classified as having cognitive impairment (assigned 1) based on education-adjusted cut-offs: Illiterate < 17 points, Primary School < 20 points, Middle School or above < 24 points. Participants scoring at or above these cut-offs are classified as cognitively intact (assigned 0).	25.30%
Chronic Conditions	Assessed based on the question: “Have you ever been diagnosed by a doctor with any of the following chronic diseases?” (List of 15 chronic diseases provided). Individuals reporting no chronic diseases are assigned 0; those reporting one or more chronic diseases are assigned 1.	21.37%

(2) Health Inequality

This measure uses the relative deprivation index of older adults’ health levels as the metric for health inequality degree. According to relative deprivation theory, individuals within a specific group exhibit a negative correlation between their health level and health disadvantage. Older adults with poorer health experience higher relative deprivation in accumulated health disadvantages, indicating greater health inequality [67]. The Kakwani relative deprivation index satisfies dimensionless, normalization, and transfer invariance requirements while overcoming the Gini coefficient’s limitation in additive decomposability [68]. Consequently, this study employs the Kakwani relative deprivation index to calculate individual-level health inequality among the older adults. The formula is defined as follows:

Y represents the older adults with sample size n, sorted in ascending health level order. The distribution function of health for this older adult population is:

5

like that:

6

In Equation (2), denotes the relative deprivation index of the j-th older adult relative to the i-th individual’s health. Summing over j, then dividing by the mean health level of the entire older adult population, yields the health relative deprivation index RD_i for the i-th individual, representing their health inequality level.

7

In Equation (3), represents the mean health level of all older adult in population Y, denotes the mean health level of samples in Y exceeding , and signifies the proportion of samples in Y with health levels exceeding relative to the total sample size.

Explanatory Variables

Digital Endowment: A comprehensive framework measuring older adults’ overall digital resources, encompassing digital access (availability of devices and connectivity), digital skills (knowledge and capability to use technologies), and digital application (breadth and depth of technology use in daily life). This study defines digital endowment through these three dimensions, with specific measures detailed in Table 2 based on data availability. Given the different measurement units and value ranges across dimensions, we applied Min-Max normalization to standardize scores from each dimension to a 0–1 scale. The composite digital endowment index was constructed by summing these normalized scores, eliminating scale differences and ensuring balanced contributions from all dimensions.

Table 2.

Measurement indicator system for digital endowment

Dimension	Indicator Definition	Rationale
Digital Access	Based on the response to “Can you access broadband internet where you currently live?”. Answers are scored as: “Yes” = 1, “No” = 0.	This represents the most fundamental infrastructure access condition and forms the basis of digital endowment.
Digital Skills	Based on survey items regarding “What do you generally use the internet for?”: “Watch videos” (Yes = 1, No = 0); “Read news” (Yes = 1, No = 0); “Chat” (Yes = 1, No = 0); and “Do you use WeChat?” (Yes = 1, No = 0). Total of 4 items, each worth 1 point, yielding a score range of 0–4.	These activities (information acquisition, entertainment, communication) better reflect older adults’ core ability to actively use the internet to meet basic needs, with less interference from external economic and social factors.
Digital Application Breadth	Based on: 1) “What do you generally use the internet for?”: “Play games” (Yes = 1, No = 0); “Manage finances and other activities” (Yes = 1, No = 0). 2) “Which of the following devices do you use to access the internet?”: “Desktop computer” (Yes = 1, No = 0); “Laptop computer” (Yes = 1, No = 0); “Tablet computer” (Yes = 1, No = 0); “Mobile phone” (Yes = 1, No = 0). Total of 6 items, each worth 1 point, yielding a score range of 0–6.	This dimension measures the diversity of applying digital technologies across different life scenarios. Using multiple devices and engaging in various activities indicates stronger transformative capacity of digital endowment.

Instrumental Variables

To address potential endogeneity, this study employs the 2014 provincial telephone penetration rate as an instrumental variable (IV) for individual digital endowment, following the historical IV approach [69]. This choice is justified on two grounds: First, the year 2014 preceded China’s massive mobile internet expansion (4 G licenses were issued in December 2013), making historical communication infrastructure a relevant predictor of contemporary digital capabilities. Second, as a predetermined “sunk cost,” this historical variable satisfies the exclusion restriction by influencing health outcomes only through shaping the digital environment. The data, extracted from the China Statistical Yearbook 2015, were matched with respondents’ provincial information in CHARLS 2020 for subsequent econometric analysis.

Mechanism Variables

(1) Informal support

Informal support, typically provided by relatives, friends, and neighbors, fulfills functions of emergency assistance, emotional exchange, and psychological security. Adopting Yang et al.’s framework [24], this study measures two dimensions: intergenerational family support and peer support.

Intergenerational family support includes offspring-provided daily care, financial support, and emotional comfort. Offspring daily care: “In daily life difficulties, do you have children (children, daughter-in-law/husband, grandchildren/grandchildren) to help you?” (1 = yes, 0 = no) [70]. Financial support: “In the past year, have you received regular financial support from your children?” (1 = yes, 0 = no). Emotional comfort: “How often you contact your children by phone, text, wechat, letter or email?” (From “almost never” to “almost every day”, assign a value of 1–9). Composite score sums all three components; higher scores indicate stronger support.

Peer support, following Bai Weijun et al. [71], measures social activity frequency: “Past month participation in: visiting friends; chess/card games; helping non-cohabiting relatives/neighbors; etc.” (8 activities). Score ranges 0–8 based on activity count; higher values indicate stronger peer support.

(2)Self-support

This study conceptualizes “self-support” as the enhanced internal psychological resources and health management efficacy that individuals develop through utilizing technological resources in the digital era. To measure this construct, we employ the CHARLS questionnaire item “Overall, are you satisfied with your life?” as a proxy variable. This indicator serves as a core measure of subjective well-being and global psychological welfare, effectively capturing individuals’ comprehensive and stable evaluation of their life quality. The improvement of digital endowment boosts older adults’ overall life satisfaction and sense of control by enhancing their information mastery, social connection maintenance, and access to life services. This enhanced psychological well-being constitutes a crucial internal resource that motivates individuals to adopt more proactive health behaviors, thereby positively influencing health outcomes. The item uses a five-point scale: “Not at all satisfied” = 1, “Not very satisfied” = 2, “Somewhat satisfied” = 3, “Very satisfied” = 4, and “Extremely satisfied” = 5. Higher scores indicate stronger self-support capacity, reflecting superior internal psychological resources.

Control Variables

Based on the social determinants of health model [72] and prior studies [18, 73, 74], regression analyses control for: age, gender, education, residence type (urban/rural), marital status, number of children, and region. Regional dummy variables (Eastern, Central, Western China) were included to control for time-invariant regional heterogeneity, such as disparities in policy and healthcare resources. This specification helps isolate the net association between digital endowment and health by comparing individuals within the same macro region. Variable descriptive statistics appear in Table 3.

Table 3.

Variable definitions and descriptive statistics

Variable Category	Variable	Definition and Measurement	Mean ± SD/n (%)
Dependent Variables	Health Inequality	Measured using the Kakwani index of relative deprivation to quantify health deprivation levels.	0.081 ± 0.083
	Health Status	Composite score calculated via the entropy weight method, incorporating five indicators: self-rated health, disability status, mental health, cognitive function, and chronic conditions.	1.571 ± 0.227
Independent Variable	Digital Endowment	Composite score from three dimensions (digital access, skills, and application breadth).	0.763 ± 0.764
Mediating Variables	Self-Support	Subjective wealth perception: Very difficult = 1; Somewhat difficult = 2; Easy = 3; Very easy = 4.	2.403 ± 0.769
	Intergenerational Support	Summed scores of offspring-provided support: daily living assistance, financial aid, and emotional comfort.	3.565 ± 2.477
	Peer Support	Number of peer social activities participated in (count).	0.723 ± 0.950
Control Variables	Age	Actual age (years).	67.819 ± 5.871
	Number of Children	Count of living offspring.	2.779 ± 1.342
	Gender	Male	4005 (50.67)
		Female.	3899 (49.33)
	Education Level	Illiterate	1942 (24.57)
		Primary school or below	3630 (45.93)
		Junior high school	1430 (18.09)
		Senior high school or technical secondary	762 (9.64)
		College or above	140 (1.77)
	Residence Type	Urban	2818 (35.65)
		Rural.	5086 (64.35)
	Marital Status	Currently married = 0	6341 (80.23)
		Not currently married	1563 (19.77)
	Region	Eastern China	2729 (34.53)
		Central China	2623 (33.19)
		Western China.	2552 (32.29)

Methodological models

Benchmark regression

This study first examines the impact of digital endowment on old adults’ health inequality. The benchmark model is specified as follows:

8

Here, represents health inequality level faced by the i-th old adult, denotes digital endowment of the i-th individual, signifies a set of control variables, , and are coefficients to be estimated, and is the random disturbance term.

Mediation effect model

Following Wen Zhonglin’s stepwise regression approach [75], this study establishes a mediation model with informal support as the mediating variable. This model is incorporated into the pathway through which digital endowment affects elderly health inequality. Specifically, two pathways require verification: the first is the effect of digital endowment on the mediating variable, and the second is the effect of the mediating variable on elderly health inequality. For the first path, the regression model is constructed as:

9

For the second path, the regression model is specified as:

10

Equations (9) and (10) constitute the mediation effect model, where represents the mechanism variable. This study operationalizes as self-support, intergenerational family support, and peer support respectively. , , and retain the same definitions as in Eq. (8). , , , , , , and denote parameters to be estimated, while and are random disturbance terms. Mediation effects are established if three conditions are simultaneously satisfied: first, achieves statistical significance in Eq. (9); second, achieves statistical significance in Eq. (10); third, the significance of from formula (8) to in formula (10) becomes insignificant, or is significantly smaller.

Empirical results analysis

Benchmark regression results

Table 4 presents the association between digital endowment and health outcomes among older adults. To enhance estimation robustness, we employed sequential regression modeling: Models (1) and (4) show the basic association without controls or regional dummy variables; Models (2) and (5) include individual-level control variables; Models (3) and (6) further incorporate regional dummy variables to control for geographical heterogeneity. Model (3) results indicate a significant positive association between digital endowment and health level (β = 0.021, p < 0.01) after controlling for covariates and regional characteristics, supporting H1’s expected direction. Model (6) shows a significant negative association between digital endowment and health inequality (β = −0.006, p < 0.01), consistent with H2’s expectation that higher digital endowment correlates with reduced health inequality. These findings collectively demonstrate that improved digital endowment among older adults associates with both enhanced individual health and reduced population-level health inequality.

Table 4.

Benchmark regression results of the associations between digital literacy, health status, and health inequalities among older adults

Dependent Variables	Health Status			Health Inequality
	（1）	（2）	（3）	（4）	（5）	（6）
Digital Endowment	0.033*** (0.003)	0.021*** (0.003)	0.021*** (0.003)	−0.011*** (0.001)	−0.006*** (0.001)	−0.006*** (0.001)
Age	/	−0.044*** (0.005)	−0.044*** (0.005)	/	0.015*** (0.002)	−0.016*** (0.002)
Gender	/	−0.001 (0.000)	−0.001 (0.000)	/	0.000 (0.000)	0.000 (0.000)
Education Level	/	0.006** (0.003)	−0.006* (0.003)	/	−0.003** (0.001)	−0.003** (0.001)
Marital Status	/	−0.024*** (0.006)	−0.024*** (0.006)	/	0.008*** (0.002)	0.008*** (0.002)
Residence Type	/	−0.012** (0.005)	−0.013** (0.005)	/	0.003* (0.002)	0.003* (0.002)
Number of Children	/	−0.007*** (0.002)	−0.006*** (0.002)	/	0.002*** (0.001)	0.002*** (0.000)
Constant	1.546*** (0.003)	1.724*** (0.037)	1.746*** (0.037)	0.090*** (0.001)	0.040*** (0.013)	0.031** (0.013)
Regional Dummy Variable	Without Controls	Without Controls	Controls	Without Controls	Without Controls	Controls
Adj R-squared	0.012	0.032	0.036	0.010	0.029	0.034
Number of Observations	7904	7904	7904	7904	7904	7904

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Robustness tests

Instrumental Variable approach

To verify the robustness of our core findings, we conducted an instrumental variable analysis to address potential endogeneity. Using the 2014 provincial telephone penetration rate in the province of residence as the instrument, we performed a two-stage regression. Results in Table 5 show a strong first-stage association between the instrument and digital endowment (p < 0.01). The minimum eigenvalue statistic of 29.81 exceeds the Stock-Yogo critical value of 16.38, rejecting the weak instrument hypothesis. The Durbin-Wu-Hausman test (p = 0.143) fails to reject exogeneity, suggesting negligible endogeneity bias in the baseline OLS model. The second-stage estimate shows a negative and marginally significant coefficient for digital endowment, aligning with the baseline result in direction while being larger in magnitude, consistent with typical IV estimation patterns. These results provide supporting evidence for a negative association between digital endowment and health inequality.

Table 5.

Instrumental Variable estimates of the relationship between digital endowment and health inequality among older adults

Variables	First Stage	Second Stage
Digital Endowment	/	−0.037* (0.022)
Internet Penetration Rate	0.002*** (0.000)	/
Constant	2.022*** (0.122)	0.101* (0.051)
Control Variables	Controls	Controls
Regional Dummy Variable	Controls	Controls
Endogeneity Test	Robust score chi2 = 2.142 (p = 0.143) Robust regression F = 2.144 (p = 0.143)
Weak Instrument Test	F-value = 29.812
Number of Observations	7904	7904

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Propensity score matching (PSM)

This section redefines explanatory variables and substitutes econometric methods to construct a counterfactual framework through matching estimation, further evaluating digital endowment’s impact on elderly health inequality. Treatment and control groups were defined using internet usage (“Whether you use the Internet or not?”): affirmative responses coded as 1, negative as 0. The approach involves identifying non-internet-using elderly with near-identical characteristics to internet users, then calculating the average health deprivation difference between matched groups. Caliper matching with k-nearest neighbors and kernel matching were implemented to ensure robustness. Post-matching propensity score distributions between groups demonstrated substantial convergence, indicating effective matching. Both methods in Table 6 show statistically significant negative associations (p < 0.01) between digital endowment and health inequality among older adults. These results align with the baseline regression estimates, supporting the robustness of our findings against potential selection bias.

Table 6.

PSM estimates of the relationship between digital endowment and health inequality among older adults

Variables	Kernel Matching	k-Nearest Neighbor Matching with Caliper (k = 4)
Digital Endowment	−0.006*** (0.001)	−0.006*** (0.001)
Constant	0.030 (0.018)	0.026 (0.018)
Control Variables	Controls	Controls
Regional Dummy Variable	Controls	Controls
Adj R-squared	0.034	0.034
Number of Observations	5007	5243

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Alternative Dependent Variable specification

This study further tests model robustness by employing an alternative measurement approach for health inequality. Regression results using the Yitzhaki index (Table 7) show a statistically significant negative coefficient for digital endowment (β = −0.011, p < 0.01), consistent with prior findings and confirming result robustness.

Table 7.

Robustness checks with alternative Dependent Variable specifications

Variables	(1)	(2)	(3)
Digital Endowment	−0.019*** (0.001)	0.011*** (0.002)	−0.011*** (0.002)
Constant	0.142*** (0.002)	0.058*** (0.021)	0.044*** (0.021)
Control Variables	Controls	Controls	Controls
Regional Dummy Variable	Controls	Controls	Controls
Adj R-squared	0.012	0.031	0.036
Number of Observations	7904	7904	7904

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Heterogeneity analysis

Given potential differential relationships of digital endowment on health inequality across population subgroups, this study conducts stratified analyses by gender, urban-rural residency, education level, and region. Results are presented in Table 8.

Table 8.

Heterogeneity analysis on the relationship between digital endowment and health inequality among older adult

Variables	Gender Heterogeneity		Urban-Rural Heterogeneity		Education Level Heterogeneity					Regional Heterogeneity
	Male	Female	Urban	Rural	Illiterate	Primary or Less	Junior High	Senior High	College+	Eastern	Central	Wester
Digital Endowment	−0.004** (0.001)	−0.009*** (0.002)	−0.006*** (0.002)	−0.006*** (0.001)	−0.000 (0.003)	−0.009*** (0.002)	−0.005* (0.002)	−0.008** (0.008)	−0.006 (0.008)	−0.004 (0.002)	−0.009*** (0.002)	−0.006** (0.002)
Constant	0.032* (0.017)	0.074*** (0.019)	0.037* (0.019)	0.032* (0.017)	0.045 (0.024)	0.006 (0.021)	0.030 (0.028)	0.079** (0.036)	0.181** (0.083)	0.043* (0.021)	0.053* (0.025)	0.016 (0.025)
Control Variables	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls
Regional Dummy Variable	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	Controls	/	/	/
Adj R-squared	0.019	0.022	0.034	0.030	0.015	0.035	0.019	0.003	−0.023	0.023	0.030	0.029
Number of Observations	4005	3899	2818	5086	1942	3630	1430	762	140	2729	2623	2552

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Gender Heterogeneity

We stratified the sample by gender to examine potential differences in the digital endowment-health inequality relationship. Table 8 results indicate a consistently negative association across both genders, though with varying strength. Among male older adults, digital endowment shows a coefficient of −0.004 (p < 0.05), suggesting that a one-unit increase in digital endowment corresponds to a 0.4% point reduction in health inequality. This represents a relatively modest decrease. Female older adults demonstrate a stronger association (β = −0.009, p < 0.01), where each unit increase in digital endowment predicts approximately 0.9% point reduction in health inequality,which is approximately twice the magnitude observed in males. These findings indicate a more pronounced negative association between digital endowment and health inequality among female older adults.

Urban-rural Heterogeneity

China’s long-standing urban-rural dual structure has created disparities in healthcare resources and digital economic development, potentially leading to different patterns in the digital endowment-health inequality relationship across residential areas [76]. Table 8 presents stratified analysis results. Both urban and rural older adults show significant negative associations between digital endowment and health inequality (β = −0.006, p < 0.01 for both groups). This indicates that improved digital endowment associates with reduced health inequality among older adults regardless of residence type, with no substantial difference in association strength between urban and rural settings.

Education level Heterogeneity

Educational capital typically correlates with health capital [77]. We stratified the sample into five educational attainment groups: illiterate, primary school or less, junior high school, high school/technical secondary, and college or above. Table 8 reveals distinct patterns across educational groups. The association is statistically insignificant for illiterate and college-educated older adults. However, significant negative associations emerge for those with primary education (β = −0.009, p < 0.01), junior high education (β = −0.005, p < 0.10), and high school education (β = −0.008, p < 0.05). These findings indicate that the digital endowment-health inequality relationship varies substantially by educational attainment, showing strongest effects among those with low to medium education levels.

Regional Heterogeneity

Significant regional socioeconomic disparities across China may create geographical variations in the digital endowment-health inequality relationship. We stratified the sample by eastern, central, and western regions. Table 8 shows no significant association in eastern China (β = −0.004, p > 0.10), while significant negative associations appear in both central (β = −0.009, p < 0.01) and western regions (β = −0.006, p < 0.01). The association strength exceeds the full-sample benchmark in these regions, with the strongest effect in central China. This pattern may reflect the more substantial marginal compensatory effect of digital technologies in regions with relatively scarce healthcare resources.

Mechanism analysis

Based on our theoretical framework, informal social engagement and self-support may serve as mediating pathways in the relationship between digital endowment and health inequality among older adults. We empirically tested these pathways using mediation models, with results presented in Table 9.

Table 9.

Mediation effect analysis of self-support and informal social interactions (family intergenerational support, peer support)

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Health Inequality	Self-Support	Health Inequality	Intergenerational Support	Health Inequality	Peer Support	Health Inequality
Digital Endowment	−0.006*** (0.001)	0.051*** (0.012)	−0.005*** (0.001)	−0.025*** (0.003)	−0.006*** (0.001)	0.023*** (0.015)	−0.006*** (0.001)
Self-Support	/	/	−0.017*** (0.001)	/	/	/	/
Intergenerational Support	/	/	/	/	0.022*** (0.004)	/	/
Peer Support	/	/	/	/	/	/	0.000 (0.001)
Constant	0.031*** (0.013)	2.963*** (0.128)	0.083*** (0.014)	0.008 (0.032)	0.030** (0.013)	0.214 (0.153)	0.085*** (0.012)
Control Variables	Controls	Controls	Controls	Controls	Controls	Controls	Controls
Regional Dummy Variable	Controls	Controls	Controls	Controls	Controls	Controls	Controls
Adj R-squared	0.034	0.006	0.061	0.131	0.037	0.072	0.034
N	7904	7904	7904	7904	7904	7904	7904

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Column (1) confirms a significant negative association between digital endowment and health inequality (p < 0.01). Columns (2), (4), and (6) demonstrate that digital endowment significantly correlates with self-support, family intergenerational support, and peer support (p < 0.01), showing positive associations with self-support and peer support but a negative association with family intergenerational support. When including both mediators and digital endowment in the regression models (columns 3, 5, and 7), self-support shows a significant mediating effect (β = −0.017, p < 0.001), suggesting digital endowment may associate with reduced health inequality through enhanced self-support capabilities, thus supporting H5. Family intergenerational support also demonstrates a significant mediating effect (β = 0.022, p < 0.001), indicating digital endowment correlates with reduced family support, which in turn associates with increased health inequality, supporting H3. Peer support shows no significant mediating role, as it lacks independent explanatory power for health inequality when controlling for digital endowment (p > 0.1), leading to rejection of H4.

To enhance robustness, we conducted bootstrap analysis (Table 10), confirming significant mediation effects for self-support (13.43% of total effect) and family intergenerational support (7.46%), while peer support remains non-significant. These results align with the stepwise method findings, supporting the mediating roles of self-support and family intergenerational support in the digital endowment-health inequality relationship.

Table 10.

Bootstrap significance tests for mediation effects

Mediation Pathway	Indirect Effect	Bootstrap 95%CI	Direct Effect	Bootstrap 95%CI	Total Effect	Bootstrap 95%CI	Mediation Proportion	Conclusion
Digital Endowment→Self-Support→Health Inequality	−0.0009*** (0.000)	−0.001, −0.000	−0.0057*** (0.001)	−0.008, −0.003	−0.0067*** (0.001)	−0.009, −0.004	13.43%	Partial Mediation
Digital Endowment→Intergenerational Support→Health Inequality	−0.0005*** (0.001)	−0.001, −0.000	−0.0061*** (0.001)	−0.008, −0.003	−0.0067*** (0.001)	−0.009, −0.004	7.46%	Partial Mediation
Digital Endowment→Peer Support→Health Inequality	0.0000 (0.000)	−0.000 0.000	−0.0067*** (0.001)	−0.009, −0.003	−0.0067*** (0.001)	−0.009, −0.003	0%	No Mediation

Note: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are reported in parentheses

Discussion

The rapid growth of internet and digital economies presents both opportunities and challenges for older adults. While existing research recognizes the internet’s positive impacts on elderly health, it overlooks differential benefits across individuals. Such differentials may exacerbate existing health disparities. Consequently, attention must extend beyond the digital divide affecting older adults to address resultant intra-group inequalities. This study examines the relationship between digital endowment and health inequality among older adults, exploring potential pathways with important theoretical and practical implications.

This study reveals a significant positive correlation between digital endowment and health levels in older adults, alongside a significant negative association with health inequality within this population. These findings suggest that enhancing digital endowment may help narrow disparities in health benefits resulting from differential internet access and usage, offering a new pathway to reduce health inequality among older adults. The social determinants of health model indicates that beyond direct disease factors, daily living conditions and social structures shaped by individual status and resources significantly impact health [78]. Digital endowment provides new pathways for accessing health information, medical services, social connections, and psychological support, substantially promoting physical and mental wellbeing [10, 18, 36, 38]. Critically, digital technologies overcome geographic, economic, and social capital constraints, empowering information-disadvantaged and medically underserved older adults. This equalizes access to health resources across socioeconomic strata and living environments [23, 43, 44], effectively narrowing health disparities. Robustness tests confirm these findings, enhancing result reliability and policy relevance. This underscores that bridging the digital divide is essential not only for individual welfare but also for achieving health equity and building inclusive societies within active aging strategies.

Second, associations of digital endowment with health inequality differed markedly across population subgroups defined by gender, education, residence, and region. This indicates differential technological adaptation and health benefit acquisition among older adult subgroups. Our analysis identifies key drivers:

Gender Heterogeneity: The negative association between digital endowment and health inequality is more pronounced among older women, a finding consistent with existing literature [58] and supportive of resource substitution theory. Influenced by traditional social roles, older women often face relatively limited social capital and restricted access to healthcare resources, which may motivate them to more actively seek health-related support through digital channels [53]. The low-cost and high accessibility of digital technologies provide this group with more convenient pathways for health management and social participation, which could help narrow gender-based health disparities [79]. Overall, enhanced digital endowment may strengthen health autonomy among older women, thereby exerting a more substantial positive influence on health equity.

Urban-Rural Heterogeneity: The negative association between digital endowment and health inequality remains significant among both urban and rural older adults, with comparable strength across residential settings. This finding aligns with Ran et al. while revealing more complex underlying mechanisms [76]. Despite disparities in digital infrastructure and healthcare resources between urban and rural areas, the association strength shows no significant difference. This pattern may reflect two complementary mechanisms: first, although urban older adults benefit from digital advantages, the inclusive and low-threshold nature of digital technologies enables rural older adults with basic digital competence to effectively access health information and services, generating comparable equity effects [80]; second, the relative scarcity of offline medical resources in rural areas may amplify the marginal utility of digital health services, partially offsetting disadvantages in digital access. These results challenge the traditional resource reinforcement theory, suggesting that digital technologies may promote health equity across diverse residential environments.

Educational Heterogeneity: The negative association between digital endowment and health inequality demonstrates a clear gradient pattern across educational attainment levels. Significant associations emerge among older adults with primary education or below, junior secondary education, and high school/technical secondary education, with the strongest effect observed in the primary education group. In contrast, no statistically significant association is found among illiterate individuals or those with college education and above. For those possessing basic literacy skills (primary to secondary education levels), digital technologies may produce notable resource substitution effects. This group maintains sufficient digital operational capacity while generally lacking alternative social resources, making them more dependent on internet access for health information and social connectivity, thereby yielding substantial marginal health benefits [53]. Enhanced digital endowment through age-friendly adaptations and simplified training precisely meets their needs, improving health information access and social connectivity to alleviate health disadvantages. Conversely, illiterate individuals face higher access barriers, while highly-educated groups experience stronger negative health impacts from excessive internet use in professional and personal contexts, including reduced rest and unhealthy behaviors (sedentary lifestyles, sleep deprivation) [81]. Thus, despite early internet adoption among high-SES groups, their digital and socioeconomic resources demonstrate no synergistic enhancement as digital access universalizes across Chinese society.

Regional Heterogeneity: The negative association between digital endowment and health inequality is primarily observed in central and western China, with no statistically significant effect detected in the eastern region. Central and western areas face relative developmental delays in economic growth, infrastructure, and public service coverage. Digital solutions partially compensate for uneven spatial distribution and shortages of quality medical resources in these regions. Digital technologies eliminate geographical information barriers, thereby providing older adults in central and western with access to authoritative health knowledge and disease prevention information comparable to that available in eastern regions [82]. National inclusive policies promoting internet infrastructure development in central/western China further drive health benefit convergence among older adults [83]. Results show a stronger association in central regions compared to western areas. Central China benefits from superior foundational infrastructure (electricity, telecommunications networks) and potentially higher average educational levels and technology adoption capabilities among residents, including older adults. These factors enhance digital intervention readiness and utilization efficiency. Central China’s developmental gap with the eastern region is associated with greater potential for digital endowment enhancement and a correspondingly stronger association with health inequality reduction. The insignificant effect in eastern China likely reflects a ceiling effect. Eastern regions exhibit advanced economic development, mature digital infrastructure, comprehensive public services, and higher average educational/digital literacy levels. Consequently, older adults in the east already experience high baseline digital endowment with balanced resource accessibility. Additional digital improvements yield diminishing marginal returns in reducing health disparities [84], suggesting saturation in health gains.

Third, regarding mechanisms, this study finds that informal support and self-support may play significant yet distinct roles in the association between digital endowment and health inequality. Empirical results indicate significant mediating effects for self-support and family intergenerational support, but not for peer support. The specific analysis is as follows:

Older adults’ self-support demonstrates a potential mediating pathway between digital endowment and health inequality. Digital endowment shows a significant correlation with enhanced self-support levels among older adults, which in turn associates with reduced health inequality. Digital technologies empower older adults to better understand and manage their health, fostering stronger self-efficacy. Online health knowledge acquisition and management enable resource-constrained individuals to independently maintain health, decreasing reliance on external support and narrowing health disparities [85]. Simultaneously, the internet serves as an efficient social interaction tool, providing channels for emotional support and social connectivity. Through digital communication methods such as social media and video calls, older adults can sustain and even expand their social networks. This not only offers practical informational support but may also provide emotional comfort. This enhanced sense of social embeddedness correlates with improved psychological well-being and health behaviors, potentially linking it to the reduction of health disparities at the population level [86]. This pathway highlights the potential of digital endowment in mitigating health inequality, particularly through reducing external dependency, enhancing health literacy, and strengthening social support networks. These mechanisms collectively enable older adults to better manage their health, thereby alleviating inequalities stemming from socioeconomic status, gender, and other factors.

Family intergenerational support demonstrates a significant mediating effect in the relationship between digital endowment and health inequality. Notably, contrasting with some existing research findings [52], this study reveals a negative association between digital endowment and family intergenerational support, while the latter shows a positive correlation with higher levels of health inequality. This likely demonstrates the Matthew Effect in digital technology use. Older adults with higher digital endowment receive more efficient and abundant remote support from children through digital tools. Conversely, those with limited digital access or children’s support capacity cannot fully benefit from such digitalized intergenerational support. Some may even experience reduced in-person visits due to over-reliance on online interactions [50]. This counterproductive dynamic stems from a mismatch between technological applications and social structures. When digital technologies serve primarily as responsibility-transfer tools rather than emotional connection channels, vulnerable older adults face heightened risks of digital isolation. Consequently, disparities in digital access to intergenerational support and digital endowment variations paradoxically widen health inequality across different family backgrounds.

Peer support does not show a statistically significant mediating effect between digital endowment and health inequality. Although improved digital endowment correlates with expanded peer support, peer support itself demonstrates no significant association with health inequality levels. This suggests digitally facilitated peer networks may entrap older adults in inefficient social engagements that fail to translate into substantive health resource redistribution. Digital technologies expand older adults’ social networks, yet such connections often exhibit homogeneity. Within these closed circles, peers sharing similar health literacy limitations repeatedly disseminate health misinformation, inadvertently reinforcing harmful behaviors [87]. Particularly for socioeconomically disadvantaged elders, peer networks typically lack quality medical resources, and digital connections amplify communication volume without enhancing resource quality. Moreover, symbolic online interactions (likes/greetings) increasingly substitute tangible mutual aid. During health emergencies, virtual expressions cannot transform into actionable care like hospital transport or physical assistance. This observation does not negate digital connections’ value but calls for integrating online peer support with offline practical assistance systems.

The findings offer multi-faceted policy implications:

1. Accelerating Digital China initiatives through elderly-inclusive strategies such as age-friendly digital adaptation, digital literacy training, and reduced internet access costs constitutes an effective pathway to enhance population-wide health outcomes and promote health equity.

2. Digital inclusion policies must address heterogeneous impacts. Targeted interventions should prioritize rural residents, less-educated groups, elderly women, and regions with weak digital infrastructure. Customized services and training are essential to prevent digital exclusion and ensure equitable access to digital health dividends. Concurrently, regions in the digital dividend acceleration phase (e.g., Central China) require intensified investment to capitalize on emerging opportunities.

3. Leveraging digital endowment to boost self-support requires vigilance against potential inequality exacerbation from declining intergenerational support. Policy interventions should encourage offspring to provide integrated online-offline care for digitally disadvantaged elderly parents. Simultaneously explore complementary support systems, including community organizations and volunteers, that deliver digital assistance to elders lacking sufficient familial support. This counters fragmentation effects arising from digitized intergenerational assistance. Although peer support shows no mediating effect, its value as a crucial social support source remains undeniable. Future initiatives should develop strategies to effectively activate and strengthen elderly peer-assistance networks using digital tools.

This study has several limitations. This study’s cross-sectional design (CHARLS 2020) limits causal inference. While robustness checks and comprehensive controls suggest limited bias from endogeneity, unobserved time-invariant confounders may persist. Results represent robust associations rather than causal effects, warranting verification with longitudinal data. Additionally, while we attribute differences in internet’s impact on health inequality between 2018 and 2020 to varying access levels, the specific mechanisms and moderating effects require further investigation. The COVID-19 pandemic context during data collection represents another limitation. Social distancing policies likely increased digital technology necessity while unevenly exacerbating the digital divide among older adults. The pandemic directly affected health outcomes and transformed informal social patterns, potentially amplifying family support roles. Consequently, the observed associations may reflect this unique period, possibly affecting generalizability to non-pandemic conditions. Nevertheless, this extreme context offers valuable insights into digital endowment’s potential role as a crucial health determinant during future societal crises.

Conclusion

The digital era presents new opportunities and pathways for addressing aging challenges and promoting health equity among older adults. By integrating theories of the digital divide, social support networks, and social stratification, this study develops an analytical framework to understand emerging mechanisms of health inequality, with particular focus on the synergistic or counteracting effects between digital and social resources. Our findings reveal robust associations between digital endowment and both improved health status and reduced health inequality among older adults. These associations demonstrate structural variations, appearing particularly pronounced among women, older adults with primary education, and residents of central China, highlighting the need for precisely targeted interventions. Regarding underlying pathways, self-support appears to serve as a core mediator in the negative association between digital endowment and health inequality, while family intergenerational support demonstrates an opposite associative pattern, suggesting technology adoption may yield differential consequences across population groups. The absence of significant mediation effects for peer support indicates that merely expanding digital social networks may be insufficient to overcome structural constraints in health resource distribution. These findings suggest that advancing digital technologies in elderly healthcare requires commitment to equitable and inclusive principles. By accurately identifying different groups’ needs and resource endowments, and optimizing policy design and technology application models, we can potentially steer digital technologies toward becoming effective tools for promoting health equity, thereby contributing to a more equal, healthy, and dignified aging society.

Author contributions

All authors have contributed to the development of the research ideas. WH obtained the dataset. CY and BH cleaned the dataset, performed the data analysis and wrote the original draft. WK analyzed and interpreted the data. WH refined the study design, oversaw the study implementation and analysis, and revised the manuscript. All authors read and approved the final version of the manuscript.

Funding

This work was financially supported by the Jilin Province Administration of Traditional Chinese Medicine Policy Research Project (NO. Zyzc-zy-2021-06).

Data availability

Publicly available datasets were analyzed in this study. This data can be found at: https://charls.pku.edu.cn/.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki. The data were obtained by applying to National School of Development, Peking University, and the ethic approval was granted by the Ethical Review Committee of Peking University (IRB00001052-11015). The potential participants in this study were fully informed of the content and aim of the research. Only those who were willing to voluntarily participate and signed the informed consent form were considered as final respondents in the survey. According to the national legislation guidelines in China, the secondary analysis of public data from CHARLS did not require additional ethics approval.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing of interest.