The digital transformation of health care has fundamentally reshaped how individuals access, interpret, and apply health information. The internet has shifted from a supplementary resource to a primary channel for health engagement, with online health information seeking (OHIS) becoming a normative component of health behavior in many regions of the world [1,2]. According to Eurostat [3], approximately 60% of individuals aged 16-74 years in European countries searched for health-related information online in 2025, reflecting a 15% increase over the past decade. Similar or higher rates have been reported in parts of Asia [4]. Reliance on digital health information increased further during the COVID-19 pandemic, when in-person access to care was constrained [5,6].

However, OHIS is no longer confined to static websites accessed through search engines. The digital health ecosystem has undergone rapid structural evolution. Early forms of OHIS centered on retrieving information from institutional websites and web-based forums [7]. The subsequent expansion of social media platforms introduced participatory environments characterized by peer exchange, user-generated content, and the networked dissemination of health narratives [8,9]. More recently, generative artificial intelligence (AI) chatbots have emerged as conversational intermediaries capable of producing tailored responses to health queries in real time [10-12]. These shifts represent not merely technological upgrades but fundamental transformations in the epistemic architecture of health information. Health information environments have evolved from retrieval-based systems to interactive, algorithmically mediated knowledge ecosystems. These environments are also becoming increasingly opaque in how information is generated, prioritized, and presented.

Within this evolving digital infrastructure, OHIS functions as both an opportunity and a risk. On the one hand, digital platforms may enhance patient activation, support shared decision-making, and facilitate chronic disease self-management [13]. On the other hand, they expose individuals to information overload, variable credibility, commercial influence, and algorithmic filtering that shapes information visibility [14,15]. The rise of generative AI further complicates this landscape by introducing systems capable of producing fluent but potentially unverifiable health guidance [16]. As such, OHIS should be understood as a cognitively demanding and socially situated process embedded within digitally structured environments rather than merely as a behavioral act [17].

Diabetes provides a relevant context for exploring OHIS. The International Diabetes Federation estimates that diabetes will affect 853 million adults worldwide by 2050, and global diabetes-related expenditures are projected to approach US $1 trillion [18]. Beyond its scale, diabetes is characterized by sustained and complex self-management demands. Effective management requires continuous engagement in medication use, blood glucose monitoring, dietary regulation, and physical activity planning [19]. These activities are not episodic but ongoing, requiring patients to repeatedly search for, interpret, evaluate, and apply health information in daily life [20]. As diabetes self-management occurs largely outside clinical settings [21], individuals frequently turn to digital environments to bridge informational gaps between consultations [22]. More than 60% of individuals with diabetes in the United States engage in OHIS [23]. OHIS, therefore, represents a central component of contemporary diabetes self-care. Within patient-centered care models that emphasize shared decision-making and patient activation [24], digital health information has become intertwined with routine disease management.

Yet, digital engagement in health contexts remains unevenly distributed across populations. The digital divide involves more than internet access alone. Differences in digital competencies, health literacy, technological resources, and socioeconomic conditions also play important roles [25]. These structural differences influence how individuals access, evaluate, and use online health information. For populations already at elevated risk of diabetes-related complications, including older adults and individuals with lower income or educational attainment, limited digital capacity may intensify existing social and health inequities [26]. Digital health environments, therefore, represent sociotechnical spaces that can expand patient empowerment while also reproducing disparities in knowledge access, participation, and health outcomes [27].

Despite expanding research on health information seeking, existing reviews have not comprehensively addressed OHIS among individuals with diabetes within this rapidly evolving digital context. Prior syntheses have examined general information-seeking behaviors [28], aggregated multiple chronic conditions without isolating diabetes-specific patterns [29], or focused narrowly on online communities [28]. These approaches provide important descriptive insights but do not integrate multilevel determinants, technological evolution, and equity considerations within a unified framework. Moreover, most predate the widespread integration of social media platforms and generative AI systems into routine health information practices.

Given the rapid transformation of digital infrastructure and the central role of self-management in diabetes care, an updated, theoretically informed synthesis is needed. This scoping review addressed this gap by mapping the evidence on OHIS behaviors among individuals with diabetes. It focused on the sources of information used, the types of information sought, and the multilevel factors influencing these behaviors. These patterns were examined across evolving digital health environments. In addition, a secondary descriptive analysis was conducted to explore temporal shifts in OHIS patterns across technological eras. By situating OHIS within the broader digital health ecosystem, this review also sought to advance conceptual understanding of how individuals with diabetes engage with increasingly algorithm-mediated information environments. The findings may inform strategies that promote equitable and evidence-informed digital health engagement.

Given the heterogeneity of study designs, populations, and outcome measures in this field, a scoping review methodology was adopted to systematically map the existing evidence, clarify key concepts, and identify knowledge gaps. The review was conducted following the framework proposed by Arksey and O’Malley [30] and further informed by methodological refinements from the Joanna Briggs Institute guidance for scoping reviews [31]. The research questions were structured using the Sample, Phenomenon of Interest, Design, Evaluation, Research type (SPIDER) framework (Multimedia Appendix 1) [32]. Reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) [33] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses literature search extension (PRISMA-S) [34] guidelines to enhance transparency and reproducibility. The completed PRISMA-ScR and PRISMA-S checklists are provided in Multimedia Appendices 2 and 3, respectively.

The review protocol was not registered before commencement.

Detailed inclusion and exclusion criteria are presented in Textbox 1. We included empirical studies published in English that described OHIS behaviors among individuals with diabetes. OHIS was defined as the active effort to obtain health-related information from digital sources and platforms to address health information needs [35]. Studies were excluded if they did not directly assess OHIS behaviors, focused solely on health literacy or eHealth literacy without a clear information-seeking component, were nonempirical, or lacked full-text availability.

A comprehensive literature search was conducted across 5 electronic databases: PubMed, Scopus, Web of Science, CINAHL, and Embase. Each database was searched independently through its web-based interface. Retrieved records were imported into EndNote (Clarivate Plc) for reference management and duplicate removal before screening. Citation tracking was additionally performed through backward reference searching of included studies using Google Scholar (Google LLC). No additional sources, such as trial registries, conference proceedings, organizational websites, or gray literature databases, were searched. Study authors were not contacted, and no supplementary search methods were used beyond those described.

The initial search was conducted in April 2025 and updated on May 4, 2026, to capture newly published studies. The search strategy was developed using the SPIDER framework and centered on 3 core concepts: diabetes, health information-seeking behavior, and online or digital contexts. Search terms were adapted from previous reviews [13,28] and refined through team discussion. Boolean operators were applied to combine keywords, and searches were performed across database-specific fields (eg, title, abstract, keyword, and topic fields). In PubMed, MeSH (Medical Subject Headings) were additionally used. No limits or filters were imposed. The full reproducible search strategies and the number of records retrieved from each database are provided in Multimedia Appendix 4.

After duplicate removal, titles and abstracts were screened by 1 reviewer (PHT) according to the predefined eligibility criteria (Textbox 1). Screening decisions were independently verified by a second reviewer (YMH). Discrepancies were resolved through discussion, with a third reviewer (HYC) consulted as needed. Full-text articles deemed potentially eligible were subsequently retrieved and assessed by the review team to confirm final inclusion.

Data charting was conducted by 1 reviewer (PHT) and cross-verified by other team members to ensure consistency and accuracy. A predefined extraction form, informed by prior literature [13,28], was used to guide data collection. Extracted variables included publication year, author, country, study design, sample characteristics, type of diabetes, disease duration, reported online platforms, types of information sought, and determinants of OHIS. Data were organized using Microsoft Excel.

Descriptive mapping was performed for online platforms and types of information sought; frequencies and percentages were calculated to summarize study characteristics. To identify determinants of OHIS, a mixed deductive-inductive analytic approach was adopted [36]. A deductive content analysis was conducted using the Social Ecological Model as a guiding framework to categorize determinants at the individual, interpersonal, organizational, and digital-environment levels [37]. Codes were iteratively refined through team discussion. To enhance interpretive depth, qualitative findings related to patient motivations, emotional responses, perceived dilemmas, and coping strategies were further synthesized. Themes were generated inductively and organized into higher-order domains that captured recurring psychological and behavioral processes underlying OHIS.

To explore the impact of technological evolution, studies were stratified into 3 periods reflecting major phases of digital health development: early web-based environments (2002-2010), the expansion of social media and mobile technologies (2011-2018), and the integrated digital and AI-enabled ecosystem (2019-2025). Study characteristics, platforms, and determinants were compared descriptively across periods to identify temporal shifts in OHIS patterns.

Consistent with the purpose of a scoping review, this study aimed to map the breadth and characteristics of the available evidence rather than evaluate intervention effectiveness [30]. Therefore, no formal quality appraisal or risk of bias assessment was conducted.

The initial search yielded 4589 records, comprising 1524 (33.21%) from Scopus, 924 (20.14%) from Embase, 871 (18.98%) from CINAHL, 691 (15.06%) from Web of Science, and 579 (12.62%) from PubMed. After the removal of 1933 (42.12%) duplicates, 2656 (57.88%) records underwent title and abstract screening. Of these, 161 (3.51%) articles were assessed for full-text eligibility, and 76 (1.66%) studies met the inclusion criteria [23,38-112]. Citation tracking identified an additional 5 eligible studies [113-117], resulting in a final sample of 81 included studies for synthesis [23,38-117] (Figure 1).

The included studies were conducted across 32 countries spanning 5 continents. The largest number originated from the United States (n=21, 26%), followed by the United Kingdom (n=7, 9%) and Australia (n=5, 6%; Figure 2). Studies were published between 2002 and 2025, with a marked increase in publication frequency over time. A total of 12 (15%) studies [38-49] were published before 2010, 29 (36%) [50-76,113,114] between 2011 and 2020, and 40 (49%) [23,77-112,115-117] between 2021 and 2025, indicating growing scholarly attention to digital health engagement in diabetes. Quantitative cross-sectional surveys were the most common study design (n=42, 52%), followed by qualitative studies (n=30, 37%) and mixed methods studies (n=9, 11%). Qualitative approaches became more prevalent after 2011, whereas mixed methods designs appeared with increasing frequency but remained relatively uncommon across publication periods (Figure 3).

Sample sizes ranged from 8 to 7999 participants. Among the 62 (77%) studies that reported diabetes type, 31 (50%) focused on type 2 diabetes, 8 (13%) on type 1 diabetes, 17 (27%) included both type 1 and type 2 diabetes, and 6 (10%) targeted gestational diabetes. Disease duration was reported in 38 (61%) studies. Detailed study characteristics are presented in Table 1.

Across the 81 included studies [23,38-117], individuals with diabetes engaged with diverse digital platforms for health information seeking (Table 2). The most frequently reported category was general internet use (n=46, 57%), often described as “using the internet” or “looking up diabetes online” without specifying particular platforms. Social media platforms and online forums (eg, Facebook, Twitter, WeChat) were also commonly reported (n=38, 47%). Health-related websites, such as WebMD and Mayo Clinic, were identified in 16 (20%) studies, followed by search engines, primarily Google, in 15 (19%) studies. Less frequently reported sources were diabetes-specific websites (n=9, 11%), government or public health websites (n=7, 9%), and mobile health apps (n=6, 7%). Generative AI chatbots were reported in 1 (1%) recent study, suggesting the early but emerging integration of AI-based tools into OHIS practices.

Temporal stratification revealed shifts in digital engagement patterns. Earlier studies (2002-2010) predominantly described the use of search engines and static health websites. From 2011 onward, social media platforms and online communities became increasingly prominent. In the most recent period (2019-2025), interactive and mobile platforms were more frequently reported, and AI-based tools appeared in the literature, albeit to a limited extent. The proportion of studies reporting social media and online community use increased from 17% (2/12) during the early web era (2002-2010) to 68% (27/40) during the integrated digital and AI era (2019-2025), whereas reports of health-related website use declined from 42% (5/12) to 10% (4/40) over the same period (Table 2). Mobile health apps and generative AI tools were reported exclusively in the most recent era. These findings suggest a transition from static information retrieval to more participatory, interactive, and technology-enabled digital environments.

Of the included studies, 49 (60%) [42,43,47-53,56-60,62-68,70-73,75,76,78,80-83,87, 89,90,92,93,95,97-101,104,105,107,108,111,112] reported on the types of online health information sought (Table 3). Information about self-management and lifestyle was the most frequently reported category (n=28, 57%), including diet, physical activity, glucose monitoring, and stress management. General diabetes knowledge was reported in 21 (43%) studies, and treatment-related information (eg, treatment options and medication side effects) appeared in 18 (37%) studies. Other commonly reported categories included symptoms or complications (n=13, 27%), general health topics (n=12, 24%), and lived experiences and peer support (n=11, 22%). Technology-related information (eg, glucose meters, insulin pumps) was reported in 3 (6%) studies. Less frequently discussed topics included psychosocial concerns (eg, depression), family impact, health insurance, and COVID-19–related information.

Across diabetes populations, self-management and lifestyle information, general diabetes knowledge, and treatment-related information were most commonly reported in studies involving type 2 diabetes or mixed diabetes populations. Technology-related information was identified exclusively in studies that included individuals with type 1 diabetes, whereas studies of gestational diabetes primarily reported information seeking related to self-management and lifestyle, as well as lived experiences and peer support (Figure 4). Overall, the content of information seeking remained relatively stable across technological eras, with self-management information consistently representing the most frequently reported category. However, more recent studies increasingly described experiential knowledge exchange, emotional support seeking, and peer-to-peer information sharing within social media environments. This pattern suggests that contemporary OHIS extends beyond information retrieval and increasingly incorporates social interaction, collective knowledge construction, and emotional support.

To enhance data visualization and summarize the distribution of evidence, an evidence gap map was developed based on diabetes type and levels of influencing factors (Figure 5). Overall, the evidence was concentrated at the individual and environmental levels, with the highest number of studies focusing on type 2 diabetes. By contrast, organizational-level factors were sparsely represented across all diabetes groups.

This scoping review provides a comprehensive synthesis of OHIS behaviors among individuals with diabetes within a rapidly evolving digital ecosystem. Among 81 studies, we identified substantial diversification of digital information sources, a temporal shift toward interactive and socially mediated platforms, and a complex constellation of multilevel determinants shaping OHIS engagement. These findings suggest that OHIS has evolved beyond a supplementary activity and now represents an integral component of contemporary diabetes self-management. Compared with earlier reviews [28,29], this synthesis extends prior work by integrating technological evolution, multilevel determinants, and equity considerations within a unified framework. By positioning OHIS within the broader digital health environment, including search engines, social media platforms, and emerging generative AI systems, this review advances a more contextualized understanding of how individuals with diabetes navigate increasingly algorithm-mediated information environments.

Our findings demonstrate a clear temporal shift in information sources. Earlier studies primarily reported reliance on search engines and institutional websites, whereas more recent publications emphasize social media platforms and online communities. These interactive environments offer experiential knowledge, emotional validation, and peer-generated advice that may complement formal medical guidance [118-120]. Engagement in peer support spaces has been associated with improved glycemic outcomes and greater self-management competence, highlighting their potential clinical relevance [121,122].

At the same time, these platforms introduce epistemic vulnerabilities. Information quality is uneven, commercial influences are pervasive, and algorithmic amplification may privilege engaging content over evidence-based information [123]. Rapidly evolving platforms, such as TikTok, exemplify environments in which users may find it difficult to evaluate credibility [124]. The dual role of social media as both a source of support and a channel for misinformation highlights the need for digital stewardship within clinical practice. HCPs may extend their role beyond traditional consultations by curating credible digital resources and supporting digitally informed peer leaders within patient communities [125].

The integration of large language model–based chatbots represents a further transformation in OHIS [10,126]. Although only 1 included study directly reported the use of generative AI among individuals with diabetes [109], emerging literature suggests increasing experimentation with conversational AI tools for health queries [127]. Unlike traditional search engines, AI chatbots generate synthesized responses in natural language and may reduce cognitive burden while tailoring information to varying literacy levels [128,129]. These tools may also provide affective support through conversational interaction [130].

However, concerns remain regarding accuracy, transparency, and contextual appropriateness [131,132]. The absence of patient-specific data limits clinical reliability, and algorithmic opacity complicates trust calibration [133]. Moreover, individuals with chronic conditions have expressed skepticism toward AI-mediated health advice [134]. Future research should therefore explore how patients integrate AI-generated responses into decision-making processes, how trust is established or eroded, and how these tools interact with clinician guidance. As conversational AI becomes embedded in routine information practices, its influence on diabetes self-management requires systematic investigation.

Patients’ health information needs appeared remarkably consistent despite rapid changes in digital technologies. Across technological eras, self-management, general diabetes knowledge, and treatment-related information remained the most frequently sought topics [20]. This pattern suggests that OHIS serves as an ongoing complement to diabetes care and self-management beyond clinical encounters. Stratified analyses further indicated that information needs varied across diabetes populations. Technology-related content was more prominent among individuals with type 1 diabetes, whereas individuals with gestational diabetes more frequently sought information related to self-management and peer support (Table 4). These findings suggest that a one-size-fits-all approach to digital health information may be insufficient and highlight the need for resources tailored to the circumstances and priorities of different diabetes populations.

Despite the expanding availability of online health information, barriers related to readability, usability, and information overload remain common [135]. Individuals with limited health literacy may face additional challenges when navigating complex medical terminology and poorly designed websites [136]. Improving digital health information, therefore, requires not only increasing access but also enhancing its relevance and usability. Co-development of online resources involving patients and HCPs, together with the use of structured evaluation tools such as DISCERN (a validated instrument for evaluating the quality of consumer health information on treatment choices) and PEMAT (Patient Education Materials Assessment Tool), may help improve the quality, clarity, and relevance of online content [137]. Regular assessment of the readability and usability of web-based materials by health care organizations may further support equitable access to digital health information and help reduce disparities in engagement and benefit.

Consistent with broader digital health literature, engagement in OHIS followed clear sociodemographic gradients [35,138]. Older age, lower educational attainment, limited income, and a diagnosis of type 2 diabetes were associated with reduced online engagement. These patterns likely reflect structural digital divides encompassing technological access, digital skills, and confidence [25]. As type 2 diabetes disproportionately affects older adults and socially disadvantaged populations [139], sociodigital inequities may amplify existing health and metabolic vulnerabilities [13].

Although many sociodemographic factors are not modifiable, recognition of their influence enables the implementation of targeted and proportionate support strategies [62,140]. Individuals who are younger or recently diagnosed may benefit from curated digital resource pathways and structured orientation to credible online platforms. For patients who experience barriers to digital engagement, alternative communication modalities, including printed educational materials and structured in-person education, remain essential complements to digital approaches [141]. Addressing digital inequities should therefore be viewed not only as a technological issue but also as a core component of equitable and patient-centered diabetes care.

Beyond demographic predictors, psychological dynamics also strongly shape OHIS. Perceived knowledge gaps, abnormal clinical results, and desires for greater autonomy were commonly associated with increased online information seeking [142,143], whereas confusion, fear, and exposure to conflicting information sometimes contributed to information avoidance [144]. These findings are consistent with theories of uncertainty management and information avoidance, suggesting that engagement with online health information involves marked emotional and cognitive processes [144]. Online health information seeking may therefore serve not only as information acquisition but also as a strategy for emotional regulation and sense-making. Patients frequently used online communities to seek reassurance, validate personal experiences, and obtain experiential knowledge, particularly when experiencing uncertainty or diabetes-related distress.

Social networks also appeared to play a prominent mediating role in OHIS. Family members and peers often assist with interpreting and contextualizing online health information, which reflects the concept of distributed health literacy within social systems [145]. While such support may enhance understanding and facilitate diabetes self-management, it may also introduce potential pathways for misinformation if mediators lack adequate literacy [146]. As diabetes management commonly occurs within family and community contexts, the interpretation and application of online information may depend not only on individual skills but also on the knowledge and beliefs of broader social networks. These findings suggest that strategies to improve digital engagement and eHealth literacy may benefit from extending beyond individual patients to include family members, caregivers, and peer communities [70].

Patient-provider communication emerged as an influential determinant of OHIS behaviors. Patients frequently turned to online sources when consultations left informational or emotional needs unmet [2]. Greater trust in HCPs was associated with reduced independent searching, yet explicit clinician endorsement of credible digital resources seemed to encourage more structured and informed engagement [147]. These findings suggest that OHIS should be incorporated into clinical communication rather than treated as a competing information source [148]. Encouraging patients to discuss online findings during consultations may help facilitate collaborative decision-making and support chronic disease management [149]. Such integration also depends on HCPs possessing adequate eHealth literacy and familiarity with digital resources. Variability in clinicians’ digital competencies further highlights the need for workforce development and additional research in this area [150].

These findings also carry broader implications for health care systems and digital health policy. Health care organizations may need to expand access to evidence-based and user-friendly digital resources while supporting HCPs in digital communication and evaluation of online information quality. As AI-generated and algorithmically curated content becomes increasingly embedded within digital health environments, concerns related to transparency, governance, and digital health equity may become more pronounced. Integrating digital health guidance into routine care may therefore help promote safer, more informed, and more equitable patient engagement within evolving digital ecosystems.

Several limitations should be acknowledged. The review was restricted to studies published in English and did not include gray literature. This may have resulted in the omission of relevant evidence from non-English–speaking contexts and introduced potential publication bias. Considerable heterogeneity in study design, populations, and reporting of socioeconomic variables limited detailed subgroup analysis. Although disparities in age, income, and education were noted, inconsistent reporting prevented a standardized synthesis of barriers across low-income or older groups. Most included studies were conducted in Western and high-income settings, which may limit the transferability of findings to low- and middle-income regions where digital infrastructure and health systems differ. In addition, the rapid evolution of digital technologies, including AI-mediated platforms, means that emerging developments may not yet be fully captured. Although this review identified a large and growing body of literature, the rapid expansion of the field may challenge the timeliness of synthesis, given the evolving nature of digital platforms. Future research should include broader geographic representation and more consistent measurement of social determinants to strengthen contextual understanding of OHIS.

This scoping review provides the most comprehensive synthesis to date of OHIS among individuals with diabetes in a rapidly evolving digital health ecosystem. Unlike previous reviews that focused primarily on specific platforms, online communities, or general information-seeking patterns, this review integrates technological evolution, multilevel determinants, and digital health equity within a single conceptual framework. The findings demonstrate that OHIS has become a dynamic, socially embedded, and increasingly algorithm-mediated component of diabetes self-management.

By identifying temporal shifts from static web resources to interactive social media platforms and emerging AI-enabled tools, this review contributes to a broader understanding of how digital ecosystems shape patient engagement, decision-making, and self-management. The evidence also highlights persistent disparities across age, education, income, and digital competencies, alongside notable gaps in organizational-level research. These findings provide actionable insights for HCPs, health care systems, and policy makers seeking to support safe, equitable, and evidence-informed digital engagement. Integrating digital health guidance into routine clinical care, strengthening eHealth literacy, and implementing transparent digital health governance strategies may help ensure that future technological advances improve diabetes outcomes without exacerbating existing health inequities.