A systematic review was conducted to identify studies examining inequalities in the use of eHealth technologies to access health care, provided by public health care systems, in relation to social determinants of health in Europe. This review was carried out in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [27] guidelines to ensure a rigorous and transparent evaluation of the review process. For details of items included in the PRISMA checklist, please see Checklist 1. The protocol was registered in the PROSPERO database (registration number: CRD420251015756).

A systematic search was performed in PubMed, Scopus, PsycINFO, and Web of Science up to April 2025. The search strategy combined controlled vocabulary and free-text terms related to digital health (eg, “telemedicine,” “eHealth,” “digital health,” and “patient portal”), social inequalities (eg, “health disparities,” “digital divide,” “inequality,” and “equity”), social determinants and user characteristics (eg, “age,” “gender,” “ethnic minorities,” “disability,” and “socioeconomic”), health care access and use (eg, “access,” “use,” “user,” and “nonuser”) and geographical terms referring to Europe and individual European countries (eg, “Europe,” “United Kingdom,” “Germany,” and “Spain”). Truncation and Boolean operators (AND, OR, NOT) were used to combine these concepts and optimize sensitivity and specificity. The search strategy was developed collaboratively by the research team, without consultation with an information scientist. Complete search strategies and term lists are available in the Multimedia Appendix 1.

The search process and its reporting followed the PRISMA-S (Preferred Reporting Items for Systematic Reviews and Meta-Analyses literature search extension) guidelines to ensure transparency, completeness, and reproducibility of each search component [28]. In addition, a backward citation search (snowballing) was conducted from the reference lists of the included studies. The literature search was rerun on October 20, 2025, to capture any recent publications since the initial search.

Studies were eligible for inclusion if they met the following criteria: (1) study population focused on adults aged 18 years or older. Studies that included younger populations but allowed for interpretation of results in adults more than 18 years were also included; (2) assessed any type of eHealth technology designed to connect individuals to public health systems; (3) reported quantitative data related to use of digital health services provided by public health systems in association with social determinants of health; (4) were conducted in European countries, defined as member states of the European Union, the European Economic Area (Iceland, Liechtenstein, and Norway), the United Kingdom, and Switzerland. This definition was selected to ensure conceptual and structural comparability of health care systems with public health care systems coverage and established digital health infrastructure; and (5) were peer-reviewed studies published in English or Spanish between January 2015 and April 2025.

The following exclusion criteria were applied: (1) studies focusing exclusively on populations using private health care services; (2) focusing on the effects of the COVID-19 pandemic on eHealth usage; (3) focusing on health apps designed for preventive purposes, such as interventions, general health information provision, self-care, or supporting the management of specific conditions; and (4) qualitative studies, study protocols, conference abstracts, theses, editorials, opinion pieces, and systematic reviews.

Study screening was conducted using the Rayyan web-based tool [29]. No automation tools were used in this process. Two researchers independently screened titles and abstracts (GM and AA), followed by full-text review of potentially eligible studies. Disagreements were resolved through a third reviewer (MF). Study authors were contacted when additional information or clarification was required.

Data extraction was carried out collaboratively by 3 researchers (GM, AA, and MF) and reviewed by the other team members to ensure accuracy and consistency. The following information was extracted: authorship, year of publication, country, study design, population characteristics, type of eHealth technology, outcome, social determinants of health included, and key quantitative findings and their narrative description. In addition, eHealth tools identified in the studies were regrouped into 4 categories according to their functionality in (1) eHealth portals—which typically integrate multiple functionalities into a single tool. These include services such as prescription renewal, appointment reminders, asynchronous messaging with health care professionals, access to test results, and in some cases, teleconsultation features or visualization of EHRs; (2) EHRs—this category includes those tools that allow users to consult their medical results, diagnoses, clinical notes, and other information recorded in their health record. These functionalities may be, and often are, integrated into digital portals managed by public health services or specific hospital apps. Notable examples include platforms such as MyCare, myHealth@QEHB, or the Care Information Exchange in the United Kingdom; (3) telemedicine and remote primary care—understood as the provision of direct health care through digital channels. It includes video medical consultations, direct-to-consumer digital care models, as well as the remote provision of primary care; (4) online appointment booking. This classification enabled a more nuanced analysis of inequalities based on the specific functionality assessed in each study. To ensure clarity in the tables, this categorization is represented numerically (1-4).

Given the variability in terminology and definitions across included studies, we compiled a comparative table showing how each study defined key determinants and eHealth usage, which is provided in Multimedia Appendix 2.

Risk of bias was assessed using the appropriate critical appraisal tools developed by the Joanna Briggs Institute (JBI), selected based on each study’s methodological design (eg, cross-sectional and cohort) [30]. The assessment was conducted at the study level rather than at the outcome level. This approach was chosen because the included studies were primarily observational and reported heterogeneous outcomes, making study-level assessment the most consistent and feasible strategy. The appraisal focused on key domains affecting internal validity, including participant selection, measurement of exposures and outcomes, and control of confounding factors. Studies were classified into 3 quality categories based on the proportion of affirmative responses to the JBI appraisal items: high quality (≥70%), moderate quality (40%‐69%), and low quality (≤39%) [31]. A detailed summary of the risk of bias assessment is provided in Multimedia Appendix 3 [32-49], using the corresponding JBI Critical Appraisal Tool.

At least 2 reviewers assessed the risk of bias independently. Discrepancies were resolved through discussion and consensus, involving 3 researchers (GM, AA, and MF) during this phase.

The effect measures reported in the studies varied. Some studies presented descriptive comparisons between population subgroups (such as percentages or proportions), while others reported adjusted measures of association derived from regression models, such as odds ratios (OR) with corresponding CIs. When available, information was classified according to the type of eHealth service analyzed.

In addition, some studies applied more advanced equity metrics, such as the concentration index and the horizontal inequity index. The concentration index summarizes whether use of services is disproportionately concentrated among socioeconomically advantaged or disadvantaged groups. The horizontal inequity index adjusts for differences in health care needs (eg, morbidity), isolating inequities that persist beyond expected variations in medical necessity. A few studies also used decomposition analysis to examine the contribution of individual determinants (eg, education, income, and employment) to overall inequality, and indirect standardization to account for differences in health needs when comparing groups.

Due to the heterogeneity in study designs, eHealth technologies evaluated, and inequality dimensions analyzed, a narrative synthesis was therefore conducted to summarize the findings. Results were presented in evidence tables and described narratively in the text. Studies with comparable outcomes were grouped, as explained in the “Data Extraction” section.

The initial search identified 2366 records from 4 databases. Of these, ultimately, 18 studies [32-49] were included in the systematic review. A PRISMA 2020 flow diagram detailing the selection process is presented in Figure 1.

The included studies were published between 2017 and 2025. A general upward trend in the number of publications was observed over time, with a noticeable increase in high-quality studies in the most recent years. The distribution of included studies per year according to their methodological quality category is illustrated in Figure 2, where the trend line represents the total number of studies published per year, and the green and yellow “x”’s represent the quality of the individual studies per year.

The most frequent study design was cross-sectional observational, accounting for 72% (13/18) of the studies [32-37,39-42,45,46,49], followed by prevalence studies at 17% (3/18) [38,43,47], with one retrospective cohort [44] and one ecological cohort study each representing 6% (1/18) [48]. The 18 studies [32-49] were conducted across 10 European countries, with the majority from Northern and Western Europe. Sweden had the highest number of studies, representing 28% (5/18) [33,36,37,44,45], followed by England at 22% (4/18) [35,46,48,49] and Finland at 11% (2/18) [32,41]. Other countries contributed one study each: Spain [42], Norway [47], Germany [40], Iceland [38], Denmark [43], the United Kingdom (distinct from England) [34], and one pan-European study [39]. Sample sizes ranged from under 438 to 1,991,995 participants. They varied from small studies such as Hörhammer et al [32] in Finland (n=438) to very large population-based datasets such as Dahlgren et al [33] in Sweden (n=1,991,995). Specifically, 17% (3/18) of studies had sample sizes below 1000 [32,34,35], 44% (8/18) ranged between 1000 and 10,000 [36-43], 28% (5/18) exceeded 10,000 participants [33,44-47], and 11% (2/18) did not report sample size [48,49]. Most studies focused on the general adult population aged 18 years and older.

According to the risk of bias assessment, 22% (4/18) of the studies [40,42,43,48] were classified as of moderate quality, while the remaining 78% (14/18) were rated as high quality [32-39,41,44-47,49]. No studies were considered to have low methodological quality, suggesting that the overall quality of the evidence included was generally high.

The eHealth tools analyzed were regrouped into 4 functional categories as described in the “Methods” section: (1) eHealth portals, (2) EHRs, (3) telemedicine and remote primary care, and (4) online appointment booking. Of the 18 included studies [32-49], 56% (10/18) analyzed eHealth portals [32,34,35,38-41,43,48,49], 33% (6/18) examined EHRs [37,41,42,46-48], 22% (4/18) addressed telemedicine and remote primary care [33,36,44,45], and only 11.1% (2/18) focused on online appointment booking [41,48]. Figure 3 presents the distribution of eHealth tool categories by publication year, highlighting a growing diversification in digital health functionalities over time. Studies appear more than once when assessing more than one tool usage.

All 18 studies reviewed examined social inequalities in the use of eHealth tools, with age and gender being the most analyzed determinants [32-49]. A second major block of variables related to socioeconomic position, especially social class [39,43,44], education [33,34,36,37,42,45], employment [32,36,40,45], income [33,45], and deprivation [35,46,48,49], often using both individual-level and area-based indicators. Fewer studies addressed cultural-related factors such as ethnicity [34,46,48,49], country of birth [33,45], nationality [42], and language [35,40,46], though these were sometimes used as proxies for integration and accessibility. Determinants related to residential context (eg, urban vs rural [40,45,49] and distance to care [33]) and family structure (eg, marital status [37,39,40], household size [39], and parity [35]) were included less frequently. Health status and clinical complexity were operationalized in diverse ways, including self-rated health [34,36,37], chronic conditions [32,33,48,49], diagnoses or number of specialties involved [45,46] or resource use band [44]. Finally, several studies assessed digital competence [41,42], literacy [32,34], and internet habits [36], highlighting the role of skills and confidence in shaping access to digital health care. Figure 4 presents a heatmap summarizing the extent to which each social determinant was not only examined but also found to be associated with inequalities in the use of specific eHealth tools. Each cell displays the proportion of studies that reported significant disparities out of those that analyzed a given determinant for each tool. For example, a value of “5/5” under “gender” and “EHRs” indicates that all 5 studies assessing gender differences in the use of EHRs reported significant inequalities by this determinant. This visual summary highlights both the breadth of research on each determinant and the consistency of evidence across different tools. eHealth portals were the most frequently studied, covering a wide range of social factors; however, the findings were somewhat mixed, with inequalities observed in some areas (eg, age and education) but not others (eg, gender and ethnicity). In contrast, studies on EHRs and telemedicine and remote primary care, though less numerous, tended to yield more consistent evidence of social inequalities, particularly along the lines of age, gender, and socioeconomic status. Finally, online appointment booking was the least explored modality, with very limited evidence available.

A summary of the main characteristics and key findings of the included studies is presented in Table 1. A more detailed version is available in Multimedia Appendix 4 [32-49].

The included studies revealed consistent social inequalities in access to telemedicine, structured along key social determinants of health and disproportionately affecting vulnerable groups.

Among the 10 studies that analyzed eHealth portals [32,34,35,38-41,43,48,49], 9 included age as a determinant [32,34,35,38-41,48,49]. Of these, 78% (6/9) found age to be a consistent factor associated with portal use [32,35,38,39,41,49]. Younger adults (15-34 years) showed higher use, and use declined after the age of 60 years [45]. Two other studies found no differences [34,46].

Regarding gender, 70% (7/10) of the studies analyzing eHealth portals included this variable [32,34,38-41,48]. Findings were mixed. Some studies found no significant differences [32,34,39,40], while others reported higher usage among women [48], and greater overall use by women [38].

Regarding socioeconomic factors, 70% (7/10) of the studies [34,35,39,40,43,48,49] examined associations between portal use and indicators such as education, income, social class, employment, and area-based deprivation. Higher education [39,40,42,45] and less deprivation [38,41,49] were consistently associated with greater use (eg, postgraduate adjusted odds ratio [aOR] 2.38, 95% CI 1.42‐4.02) [40].

Employment status was analyzed in 30% (3/10) of studies, showing inconsistent results [32,39,40]: only one study found that being used was related to lower use (OR 0.57, 95% CI 0.38‐0.84) [45]. The other two reported no associations [32,39].

Household composition, including size of household and marital status, was found as not associated in 2 of the 3 studies that analyzed it (2/10) [39,40]. However, Pierce et al [35] tied higher parity to lower use.

Cultural background—analyzed in 5 [34,35,40,48,49] of the 10 portal-focused studies (50%) [32,34,35,38-41,43,48,49]—showed no strong effects, with no significant differences in 2 studies [34,40]. Pierce et al [35] reported that 48.7% of nonusers were non-native compared to 24.4 % of those with high engagement; Kc et al [48] observed higher use in White patients [48]; Zhang [49] noted higher Asian patient use postadjustment.

In terms of area of residence, 2 [39,49] of the 3 (3/10) studies [39,40,49] that analyzed it found higher use among urban areas (OR 1.23, 95% CI 1.02‐1.48) [39] and activation [49]. Knöchelmann et al [40] found no differences.

Health status was examined in 40% (4/10) of the studies [32,34,48,49], with mixed associations: some studies reported lower use among individuals with long-term conditions [48,49], while others found higher use among those with poor health [34] or no significant differences [32].

Finally, digital competence was analyzed in 30% (3/10) of the studies, where it was described as a strong predictor [32,34,41]: (OR 8.82, 95% CI 6.15‐12.64) [41] and (aOR 2.96, 95% CI 2.02‐4.35) for high literacy [34]. Lower health literacy was associated with lower use [32].

Among the 33% (6/18) of studies [37,41,42,46-48] that analyzed EHR use, age was examined in 5, and all of them found that younger adults were more likely to use these tools [37,41,42,46,48]. Engagement declined in older adults, especially ≥75 years, with reduced registration (aOR 0.40, 95% CI 0.36‐0.44) and activation (aOR 0.39, 95% CI 0.32‐0.47) [46].

Gender was assessed in all 6 studies [37,41,42,46-48]. Four (67%) found higher EHR use among women [37,41,46-48]. In contrast, England reported higher male registration into the tool (aOR 1.10, 95% CI 1.04‐1.16) but lower activation of it (aOR 0.85, 95% CI 0.78‐0.94) [46], which was necessary for usage. González-Cacheda et al [42] found no differences.

Socioeconomic status was examined in 67% (4/6) of the studies [37,42,46,48]. All confirmed that higher levels of education increased use [37,42], and deprivation showed a clear gradient: individuals in more affluent areas were more likely to register and activate the tool [46,48].

Cultural background, assessed in 50% (3/6) of the studies [42,46,48], predicted lower activation in Asian (aOR 0.61, 95% CI 0.53‐0.71), Black (aOR 0.45, 95% CI 0.36‐0.56), and mixed (aOR 0.77, 95% CI 0.60‐0.97) groups compared to White [46]. Evidence comes primarily from a large cross-sectional study conducted in England (n=28,637) [46], complemented by national datasets from England [48] and Spain [42]. Language barriers reduced use as no interpreter need increased registration (aOR 1.63, 95% CI 1.33‐1.99) and activation (aOR 3.16, 95% CI 1.96‐5.09) [46]. National [42] and White patients [48] had higher use.

Only one study (17%, 1/6) [37], a Swedish survey of over 3000 adults, analyzed marital status, reporting that being in a partnership was linked to higher engagement [37].

Health status findings showed mixed results among the 50% of studies (3/6): one study found no differences [37], another observed lower use in long-term conditions [48], while Chapman reported higher registration among patients with complex conditions (aOR 2.54, 95% CI 2.30‐2.82) [46].

Finally, both studies assessing digital competence were conducted in Finland (n=4495) [41] and Spain (n=7454) [42], using large national datasets, and both found that lower digital skills were associated with reduced use of EHRs [41,42].

Among the 22% (4/18) of studies [33,36,44,45] that analyzed telemedicine and remote primary care, 100% (4/4) examined age and consistently identified younger adults as the main users, while older adults showed lower use [33,36,44,45]. In Sweden, those aged 60‐79 years had substantially lower use (IRR [incidence rate ratio [0.45], 95% CI 0.40‐0.50) and those ≥80 years even lower (IRR 0.19, 95% CI 0.16‐0.25) [44].

Gender was assessed in 100% (4/4) of the studies analyzing telemedicine and remote primary care [33,36,44,45], with 75% (3/4) reporting higher usage among women [33,44,45]. In Sweden, women accounted for 65.4% of users (IRR 1.39, 95% CI 1.28‐1.51) [44], while one study found no gender difference [36].

Socioeconomic factors were assessed in 100% (4/4) of the studies [33,36,44,45]. Higher education was associated with greater use (OR 1.41, 95% CI 1.19‐1.67) [36], with postsecondary users showing the highest engagement (7.2%; OR 1.51, 95% CI 1.74‐1.56) [33]. More users were observed among individuals with higher incomes (7.7% vs 4.0%; OR 2.46, 95% CI 2.38‐2.54) [33]. Employment was also linked to usage: 85.7% of users were employed versus 14.3% unemployed [45], while one study reported lower odds among retirees (OR 0.68, 95% CI 0.46‐1.02) [36]. One study found no differences across social class [44].

Cultural background and origin were assessed in 50% (2/4) of the studies, both of which were based on large population-based datasets [33,45]. Both studies reported lower uptake among foreign-born users compared to natives (3.2% vs 7.2%; OR 0.53, 95% CI 0.52‐0.55) [33].

Urban residency was analyzed in 2 [33,45] of 4 studies [33,36,44,45]. One study associated it with higher use [45], while residing ≥10 km from health care services was significantly associated with higher telemedicine use (7.1%; OR≤1.11) [33].

Health status was analyzed in 100% (4/4) of the studies [33,36,44,45], showing mixed results. Better health predicted higher use (OR 1.31, 95% CI 1.01‐1.69) [36]. In contrast, heart failure was associated with lower use (OR 0.49, 95% CI 0.40‐0.59) [33]. Some conditions, such as depression, asthma, and chronic obstructive pulmonary disease, were linked to higher use, with greater morbidity also associated with increased engagement (IRR 2.67, 95% CI 2.02‐3.54) [44,45].

Inequalities in online appointment booking were examined in 11% (2/18) of the studies [41,48], both showing that younger adults were more likely to use these systems. In Finland, use declined sharply with age [41], while in England, practices with more patients aged 15‐34 years had slightly higher booking rates (+1.35%) [48]. Gender differences were assessed in 100% (2/2) of the studies. Both studies reported lower online booking among men [41,48].

Socioeconomic and cultural factors were analyzed in 50% (1/2) of the studies, revealing lower booking rates in practices located in more deprived areas, as well as reduced uptake among non-White populations [48].

Health status was assessed in 50% (1/2) of the studies, showing that patients with long-term conditions booked less frequently (–0.77%) [48].

Finally, digital competence was evaluated in 50% (1/2) of the studies and was strongly associated with use; patients with good digital skills were over 10 times more likely to book online (OR 10.91, 95% CI 7.24‐16.44) [41].

Although European public health care systems are characterized by universal health care coverage, this systematic review aimed to examine inequalities in access to and use of eHealth tools across social determinants of health. The findings reveal persistent and significant social inequalities, with lower usage consistently reported among older adults, individuals with lower socioeconomic or educational status, rural populations, ethnic minorities, people with poor health, and those with limited digital competence. Findings regarding gender were more mixed, with some studies reporting slightly lower usage among men. The majority of included studies were methodologically robust, with most rated high quality and a few moderate quality according to JBI appraisal and conducted with widely varying sample sizes, which should be considered when interpreting these patterns and their potential translation into real-world health care practice. eHealth portals were the most investigated tool, but EHRs and telemedicine showed more consistent evidence of social gradients. Online appointment booking was the least studied category, limiting the strength of conclusions in this area.

Although digitalization in health care has been presented as improving access and system efficiency [1,2,50], our findings reveal a generational gap across all tool types [32,33,35-39,41,42,44-46,48,49]. Lower usage among older adults reflects not only differences in digital competence—such as less experience and lower technological confidence [51,52]—but also a lack of service adaptation to their needs and potential age-related cognitive barriers that limit effectiveness [53]. These observations, derived mostly from high-quality studies, strengthen the reliability of the age-related patterns observed. They were consistently found both in medium-sized samples [37] and in large administrative datasets [33,44], reinforcing the robustness of these patterns across study designs. At the same time, heterogeneity in measurement and study scale reduces the precision of effect estimates, particularly in smaller studies. In addition, unintuitive interfaces, low perceived usefulness, and a preference for in-person care further contribute to lower uptake [54]. While family support may partially mitigate this gap [54,55], findings are inconsistent [40,56]. This raises concerns about reduced autonomy, informal caregiver burden—often falling on women—and the exclusion of those without social support.

Gender-related patterns were less consistent. Some studies showed higher usage among women for EHRs and telemedicine, while differences in other services were weaker, albeit with a general trend toward greater women’s engagement. These mixed results may reflect contextual factors such as national health care organization, cultural norms, and sample composition. Some null findings came from smaller or moderate samples [32,40], whereas large studies such as Dahlgren et al [33] consistently showed gender differences. Greater women’s involvement may be interpreted in light of traditional gender roles, which assign women the responsibility for caregiving, reflected in a higher willingness to seek care, familiarity with health services, and confidence in sustained, active communication [57] as well as in more continuous use, as suggested by one of the included studies [46]. However, previous literature shows that in certain contexts, the gender gap is reversed, possibly due to broader disparities in digital engagement. Women tend to use a more limited range of digital services and engage with them less frequently and intensively than men, reflecting differences in digital familiarity and confidence [58,59].

It is well established that the socioeconomic gradient not only determines access to material resources but also shapes individuals’ ability to navigate digital environments. In our review, educational attainment, income level, and residence in disadvantaged areas were systematically associated with lower use of digital health tools, in line with studies from other digitalizing systems [60]. Evidence regarding employment status was less consistent, with some studies suggesting lower use among unemployed individuals. Beyond material barriers (access to internet or digital devices), differences in digital literacy, confidence, and prior experience—the so-called third-level digital divide—further perpetuated exclusion [61,62]. Although education strongly predicted competences, structural conditions such as local digital infrastructure and social position also shaped digital engagement. In disadvantaged regions, limited digital services may affect both patients and health care professionals, potentially reinforcing existing health care disparities [63]. Given the overall rigor of the included studies, these findings appear reliable; however, their transferability to different health care contexts should be approached with caution. The WHO stresses that closing this gap requires not only investment in infrastructure, but also training and support strategies specifically targeted at the most vulnerable groups [64]. The consistency of socioeconomic gradients across both small samples [34] and very large datasets [33] strengthens confidence in these associations and suggests that they are not driven by sample-specific variability.

Beyond socioeconomic status, ethnic minorities have shown lower engagement with all the tools analyzed, even after adjusting for socioeconomic variables. Language barriers, lack of familiarity with the health care system, distrust of institutional platforms, and poor cultural adaptation of services emerge as recurring obstacles [46,65,66].

Although urban environments are generally associated with higher use of eHealth portals [39,49], one study showed greater use of video consultations in rural areas [33], consistent with previous literature [67], which suggests that digitalization compensates for limited in-person services. This pattern, particularly in studies with careful design and with large samples such as Dahlgren et al [33], suggests that geographical effects are not artifacts of small-sample instability. However, rural areas often have lower educational attainment, which can reduce digital competence [42,45]. These findings indicate that geography can both mitigate and amplify inequalities, depending on the system’s capacity to adapt.

Regarding health status, some studies showed lower use among people with multimorbidity, in line with previous literature [68,69], reflecting the Inverse Care Law, where populations with greater health care needs often face reduced access [70,71]. However, other studies found higher use among people with poorer health, likely due to more frequent contact with the health care system [40,48]. These mixed results are partly attributable to methodological differences and varying study quality, with patterns more clearly observed in studies with comprehensive data and stronger design [33,37]. Variation in sample sizes also contributed: larger datasets such as Eriksson et al [44] and Dahlgren et al (N≈2 million) [33] produced more stable health-related estimates than smaller clinical samples, helping explain discrepancies in the direction and magnitude of associations. Overall, individuals with greater clinical needs may face additional barriers when poor health intersects with low education or socioeconomic status.

Although the overall methodological quality of the included studies was acceptable—with most rated high quality and a few moderate quality according to the JBI appraisal—the predominance of cross-sectional designs limits causal inference. Substantial heterogeneity in study populations, digital tools, and outcome indicators also affects comparability across studies. Furthermore, the wide variability in sample sizes introduces important differences in estimate precision, which should be taken into account when interpreting inconsistent findings across determinants. These methodological and contextual differences influence how findings can be translated into real-world practice: variations in health care organization, digital infrastructure, and population characteristics across countries may alter the magnitude and direction of observed inequalities. Therefore, interpretation and policy translation should be undertaken with caution, recognizing that local context strongly shapes the impact of the digital divide.

This systematic review has several limitations. One limitation is that the search strategy was not developed in collaboration with an information scientist, which might have influenced the sensitivity and comprehensiveness of the retrieved records. The conclusions may also be affected by differences in study design, sample representativeness, and reporting. Although most studies were rated as high quality according to the JBI appraisal, some moderate-quality designs introduce a potential risk of bias that should be considered when interpreting the results. The exclusion of qualitative studies—although methodologically justified to ensure comparability—may have constrained the review’s ability to capture rich contextual insights, such as user experiences, perceived barriers, and cultural or behavioral factors influencing digital health use, which are crucial for understanding the mechanisms behind observed inequalities. Heterogeneity in populations, definitions of eHealth use, and measurement of social determinants limited comparability and prevented meta-analysis, requiring a narrative synthesis. Restricting the search to peer-reviewed publications in English or Spanish may have introduced language and publication bias, excluding relevant evidence published in other languages or in gray literature. Nevertheless, this criterion enhances replicability and scientific robustness. The focus on public health care systems in Europe restricts generalizability to other regions. Additionally, the studies’ highly variable sample sizes, while not undermining the findings, do influence precision and should be considered when weighing the strength of evidence across determinants.

Despite these limitations, the review has several strengths. It followed PRISMA guidelines, used a comprehensive search across 4 major databases, supplemented by backward citation searching, and included independent screening and data extraction by multiple reviewers to improve validity. By synthesizing evidence across diverse European settings, this review offers an up-to-date and policy-relevant overview of eHealth-related inequalities, and its classification by type of digital service enables a more granular understanding of patterns of exclusion. No amendments were made to the registered protocol after PROSPERO registration.

These findings carry important implications for real-world health care policy and practice. By highlighting persistent inequalities in access and use of digital health tools across multiple social determinants, the review underscores the need for targeted interventions, including digital literacy programs, inclusive eHealth platform design, and support strategies for older adults, ethnic minorities, and socioeconomically disadvantaged populations. The innovative contribution of this review lies in its focus on quantitative evidence from European public health care systems, integrating results across diverse social determinants and types of eHealth tools. This provides a policy-relevant, evidence-based perspective to guide equitable digital transformation and foster inclusive health systems.

Future research should explore the long-term effects of digital exclusion on health outcomes, assess the equity impact of digital health policies, and promote participatory research with underserved communities to co-develop inclusive digital solutions. There is a clear need for intersectional frameworks to understand how intersecting inequalities shape digital health access, especially in highly vulnerable populations. Additionally, research is scarce in low-income countries, where sociodemographic and economic contexts may lead to different and sometimes contrasting patterns, for example regarding gender. Further studies are also needed to evaluate the actual impact of digital health tools on access to health care services more broadly—not only their uptake or use but also how they affect entry into the health system and care continuity. Moreover, more attention should be paid to the underlying mechanisms driving these inequalities, to move beyond description and identify structural, institutional, or behavioral factors that generate unequal access. Finally, while the number of studies exploring social disparities in digital health access has increased, there remains a gap in longitudinal and in-depth research that can capture the complexity and dynamics of these inequities over time.

This systematic review shows that digital transformation in European public health systems is not equally benefiting all population groups. Across the studies analyzed, persistent social inequalities in the use of digital health tools—particularly EHRs, eHealth portals, and telemedicine and remote primary care—were observed. The most excluded from digital care services are older adults, individuals with lower socioeconomic or educational status, ethnic minorities, rural populations, people in poor health, and those with limited digital competence.

Rather than closing gaps, digitalization may replicate or deepen existing health inequalities if equity is not explicitly addressed. Ensuring inclusive design, supporting digital literacy, and engaging underserved communities are essential steps to make digital health a tool for reducing—not reinforcing—structural exclusion. Although a few studies explicitly explored interactions between determinants, the detected patterns suggest that the combination of factors such as advanced age, low socioeconomic status, migrant background, and limited digital competence tends to produce cumulative effects that significantly limit the adoption of eHealth tools. In interpreting these findings, both the heterogeneity of studies and their risk of bias should be taken into account. While these findings offer a policy-relevant perspective, the heterogeneity of study designs, populations, and outcomes, as well as the risk of bias, limits direct translation into policy and practice. By synthesizing quantitative evidence across diverse social determinants and types of digital health tools, this review not only highlights persistent inequalities but also offers a policy-relevant perspective. These findings can inform equity-oriented interventions, guide inclusive eHealth design, and support digital literacy programs, helping policymakers and health systems ensure that digital transformation benefits all population groups.