Digital health literacy among university students: a systematic review of global trends, determinants, measurement, and the literacy-to-intervention gap

Abstract

Introduction

University students interact with the complexities of public health knowledge in digital environments. Digital health literacy (DHL), in this regard, is crucial as an upstream skill for ensuring safe, inclusive, and knowledgeable health care engagement among university students where mis-information, algorithm amplification, and AI-based content exist.

Objective

This systematic review synthesizes global evidence on DHL among university students. It examines measurement approaches, socio-demographic determinants, online health information-seeking behaviours, mental health and well-being correlates, and the gap between literacy assessment and tested university-based interventions.

Methods

This systematic review followed PRISMA 2020 guidance. Searches were conducted in PubMed, Scopus, and Web of Science for peer-reviewed studies indexed up to March 31, 2024; backward and forward citation tracking was updated to January 2025 to capture directly related empirical papers. Two reviewers independently performed screening, data extraction, and quality appraisal using design-appropriate tools: JBI for analytical cross-sectional studies, CASP for qualitative evidence, and AMSTAR 2-informed criteria for review-derived or pilot evidence. Because of heterogeneity in design, instruments, and outcomes, findings were synthesized narratively and thematically.

Results

Twenty-one studies met the eligibility criteria. Most were cross-sectional studies conducted during or after the COVID-19 pandemic. Five themes were identified: DHL as a competence for appraising online health information; socioeconomic, gender, health-status, and contextual determinants of DHL; associations with mental health, well-being, fear, and sense of coherence; measurement heterogeneity and psychometric maturation; and a persistent literacy-to-intervention gap.

Conclusion

DHL is a relevant competency for Health Promoting Universities, but the current evidence base remains dominated by descriptive self-report and cross-sectional designs. Future research should prioritize longitudinal studies, objective task-based competence assessment, and theory-driven intervention trials embedded within campus health systems, library infrastructures, and digital mental health pathways.

The future research must focus on conducting longitudinal research, assessing task-based competency objectively, and testing theory-based interventions that operate through campus health systems, library infrastructure, and pathways for digital mental health.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12889-026-28523-6.

Introduction

Digital health literacy is abbreviated by DHL. DHL refers to the process through which one seeks information about health issues on digital media, understands it, evaluates it, develops it, and finally uses the information in order to be able to accomplish a certain goal [1, 2]. DHL is not just a term used to refer to general health literacy; it incorporates more sophisticated skills in relation to digital media usage, digital privacy, interactive skills, and navigation platforms [2, 3]. It is crucial to make this conceptual distinction; users in algorithmically mediated environments are not only passively exposed to static information but instead actively search, compare, share, synthesize, and use multiple layers of health-related data [4–6].

The university population is an area of strategic importance as they are a substantial consumer of digital services. They are also in a transitional phase of their lives, moving from a reliance on their families for their health care to independently making health care decisions, and looking online for information before seeking formal clinical health services [7–9]. The COVID-19 pandemic has reinforced this reliance by placing all aspects of university life – such as academic education, socialising, messages around public health, mental health counselling, and helping-seeking behaviour - onto virtual platforms [10, 11]. The acute phase of the pandemic has passed but there are still structural issues, including: enduring vulnerabilities in digital trust, susceptibility to online misinformation, health related cyberchondria, and institutional failings around promoting health for students through virtual means [12, 13, 14].

Infodemic literature shows that increasing amounts of information combined with compounding false and accurate claims, as well as algorithmic loops, distort public perceptions of risk and their behaviour towards prevention of risk, and institutional trust [5, 15, 14]. Emergent empirical information on COVID-19 filter bubbles, the contrasting ways in which misinformation about COVID-19 spreads, and artificial intelligence based data collection have shown that the Internet and social media/marketing are not passive platforms for the distribution of information, rather, they are influenced by current technologies that alter how we think and act about our health [4, 6, 16]. The application of big data, blockchain technologies for health, and linked data architectures have overhauled the way health information is generated, validated, exchanged and secured/owned [17, 18].

In the realm of higher education groups, DHL has an intimate relationship with mental health and wellness, although there are many areas where these things need to be accurately defined by empirical evidence. Mental health represents certain mental health symptoms or negative feelings (e.g., anxiety, depression, etc.), and well-being is about a broader range of subjective, mental, or functional measures of how good one’s life is (e.g., the WHO–5 well-being index or the sense of coherence). DHL represents a key upstream influence for both regions by supporting students with the means to locate quality resources, reduce uncertainty when making decisions, limit all types of negative media exposure, and successfully use online DHM services [19–22]. There is evidence from different countries, with a unique postsecondary education experience, including Germany, Slovenia, Vietnam, East/South-East Asia, Hispanic communities, Italy, Portugal, Ghana, Taiwan, Ecuador, Greece, Romania, and China. These researchers have all demonstrated the correlation of an increase in DHL with increased satisfaction of information resources, a better ability to assess the quality of information received, greater psychological resiliency, and a reduced risk of developing psychological vulnerability; however, the current literature has limitations based on the use of cross-sectional design and questionnaire data [7–9, 12, 13, 23].

While there is increasing interest in digital health, including digital mental health, relatively few systematic reviews have specifically addressed student digital health literacy and its measurement. The majority of systematic reviews on the topic have focused on either the broad level of health literacy or on clinical effectiveness and acceptability to users of particular digital health interventions [24–26]. In addition, little attention has been given to the baseline capacities required for students to engage with these digital interventions safely, critically, and equitably. This large gap means that digital health technologies may worsen health inequity; individuals in the target population will not have the requisite ability to assess credibility of sources, evaluate data privacy risks associated with the technology they use or transfer digital text to appropriate offline help-seeking behavior [3, 8, 27]. Hence, this systematic review synthesizes evidence from around the world concerning student digital health literacy and assesses how far the field has moved from describing and measuring (correlationally) to tested and scalable, embedded digital health interventions at universities.

DHL is framed within the principles of the Health Promoting University (HPU) model and Okanagan Charter as a fundamental facet of integration of health in all areas of higher education institutions, including their culture, policies, operations, academic environments and partnership with the community [28]. Therefore, from a systemic view DHL is not viewed as an isolated individual skill or ability but rather as an institutional responsibility requiring a verifiable information ecosystem, digital care pathways that are made accessible, privacy-oriented communication channels and health literate operations [14, 28].

Objectives

1. To systematically identify and synthesize empirical literature evaluating digital health literacy (DHL), eHealth literacy, and crisis-specific (COVID-19) DHL among global university populations.

2. To map and critically evaluate the measurement instruments, socio-demographic determinants, and health outcomes associated with DHL in higher education settings.

3. To analyze the mechanistic relationships between DHL and online information-seeking behaviors, mental health outcomes, subjective well-being, sense of coherence, and digital health tool adoption.

4. To critically assess the evolutionary trajectory of the literature from descriptive DHL assessment toward the robust design, execution, and evaluation of evidence-based campus interventions.

Methods

Research and reporting design

The systematic review and narrative synthesis were conducted in accordance with the Preferred Reporting.

Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 [29]. To assure methodological.

granularity and prevent over-interpretation, the bibliometric mapping methodology, which was used in some previous versions, was omitted in the core findings due to the fact that less than 21 eligible studies.

provided insufficient evidence for the validation of keyword co-occurrence and creation of a network graph. A well-structured conceptual framework has been built to visualize and analyze the data of all eligible studies without bibliometric analysis methods since they were based on the low number of studies.

Protocol registration and justifiable departures from the protocol

This systematic review was prospectively registered at the International Prospective Register of Systematic.

Reviews (PROSPERO; registration number: CRD420251064452). While the initial protocol was more generally focused on digital health interventions, it was determined that during the screening process the corpus of evidence was centered on measuring, determining, and information seeking mechanisms for DHL, not efficacy trials. The topic was then limited to DHL pathway and this considered a protocol departure.

In order to mitigate potential selection bias following the above scoping, the identified records were separately rescreened using the updated DHL-based inclusion criteria. The search keywords incorporated all DHL-specific terms right from the start, such as digital health literacy, eHealth literacy, eHEALS, DHLI, and COVID-DHLI. The ultimate body of 21 articles was harmonized throughout the PRISMA flow chart, Table 2, references, and Supplemental File 2. No article was selected without fulfilling all the PCCD criteria.

Table 2.

Comprehensive Synthesis and Baseline Profile of the Included Empirical Corpus (N = 21)

Author and Year	Country/Region	Study Design	Sample Size (N)	Primary Instrument / Target Construct	Core Substantive Finding
[7]	Germany	Cross-sectional web survey	14,916	COVID-adapted DHLI	High baseline DHL paired with difficulties in critical reliability appraisal; high social media reliance was associated with lower evaluation competencies.
[27]	United States	Cross-sectional survey	256	Health literacy and COVID-DHL items	Traditional health literacy and DHL were independent predictors of preventive compliance; DHL also predicted vaccine willingness.
[11]	Slovenia	Cross-sectional survey	3,621	COVID-adapted DHLI	Nearly half of students reported difficulty assessing reliability; sufficient DHL was associated with use of official channels.
[10]	Vietnam	Cross-sectional mediation analysis	1,003	COVID-adapted DHLI; WHO-5	DHL mediated the association between online information search importance and subjective well-being.
[34]	China, Malaysia, Philippines	Cross-sectional multi-country survey	5,302	COVID-adapted DHLI	Search engines and social media dominated information-seeking; higher DHL was associated with use of trustworthy resources.
[23]	Spain, Puerto Rico, Ecuador	Cross-sectional validation study	2,318	COVID-DHLI-Spanish	The Spanish COVID-DHLI showed acceptable validity; gender and socioeconomic gradients in DHL were observed.
[3]	Italy	Psychometric validation study	3,985	Italian COVID-DHLI	A four-factor structure was confirmed; the privacy subscale showed psychometric instability and was excluded.
[35]	Portugal	Psychometric validation study	1,815	Portuguese DHLI	A four-factor structure was supported, indicating suitability of the instrument for Portuguese university students.
[36]	Portugal	Cross-sectional survey	3,084	COVID-adapted DHLI	Searching for public health and social measures was associated with faster achievement of sufficient DHL subscale scores.
[37]	Denmark	Cross-sectional survey	1,518	COVID-HL questionnaire	Information-seeking trajectories highlighted the role of social networks and institutional sources in processing health information.
[22]	Ghana	Cross-sectional mediation analysis	1,160	COVID-DHLI; WHO-5	Pocket-money sufficiency was associated with subjective well-being, with DHL operating as a protective mediator.
[19]	Ghana	Cross-sectional moderation analysis	1,160	COVID-DHLI; Sense of Coherence	DHL was positively associated with sense of coherence; subjective social class and age moderated this relationship.
[38]	Taiwan	Cross-sectional path analysis	1,631	DHL; WHO-5; Fear of COVID-19	DHL had direct and indirect associations with well-being through fear of COVID-19 and information satisfaction.
[23]	Ecuador	Cross-sectional regression analysis	917	Spanish COVID-DHLI; WHO-5	Higher DHL was independently associated with subjective well-being after sociodemographic adjustment.
[21]	Italy	Cross-sectional survey	3,001	DHL; Sense of Coherence; Future Anxiety	Sense of coherence and socioeconomic determinants were linked to well-being, future anxiety, and recurrent health complaints.
[26]	Bangladesh	TAM-guided cross-sectional survey	311	Technology Acceptance Model constructs	Perceived usefulness, ease of use, and social influence were associated with intention to use digital health for mental health.
[9]	Greece	Cross-sectional survey	604	COVID-HL questionnaire	A high proportion of students reported difficulty evaluating online credibility, supporting campus-level critical appraisal training.
[12]	Romania	Cross-sectional survey	1,381	COVID-adapted DHL scale	Age, subjective social status, gender, and chronic disease history were predictors of DHL capacity.
[13]	China	Latent profile analysis	1,060	DHL profiles; fear/depression scales	Three DHL profiles were identified; the low-DHL profile was associated with higher fear and depression.
[25]	Australia	Qualitative sub-study within a trial	115	Digital mental health engagement	Low-intensity video-based intervention was acceptable; students preferred implementation through trusted university channels.
[24]	Germany	Rapid review and methodological pilot study	4	Digital educational escape room	Prototype testing suggested feasibility for stress-management education; larger studies are required.

DHL Digital Health Literacy, DHLI  Digital Health Literacy Instrument, HL  Health Literacy, WHO-5  World Health Organization Five Well-Being Index, TAM  Technology Acceptance Model

Eligibility criteria

Study eligibility was structured using a customized Population, Concept, Context, and Study Design (PCCS) framework, detailed operational definitions for which are outlined in Table 1.

Table 1.

Inclusion criteria and operational definitions

Criterion	Operational Definition	Exclusion Criteria
Population	Students currently enrolled in recognized universities, colleges, or higher education institutions (including undergraduate, postgraduate, professional, and health sciences cohorts).	Studies restricted exclusively to secondary school students, primary school children, staff-only samples, or industry professionals, unless student-specific data were reported separately.
Concept	Digital health literacy (DHL), eHealth literacy, crisis-specific or COVID-19-related DHL, online health information-seeking behavior, digital mental health resource engagement, or student-centric digital health promotion frameworks.	Generic media or computer literacy assessments devoid of an explicit health information appraisal, digital health engagement, or health literacy focus.
Context	Global higher education settings across any geographical region, including campus health promotion systems, university health services, or general student populations.	Studies evaluating clinical patient populations outside the administrative boundaries of higher education ecosystems.
Study Design	Peer-reviewed empirical research utilizing quantitative (cross-sectional or longitudinal), qualitative, mixed-methods, instrument validation, implementation, or pilot methodologies.	Editorials, prospective protocols omitting empirical results, narrative commentaries, purely conceptual treatises, non-peer-reviewed white papers, and grey literature lacking extractable data.
Temporal & Language	Peer-reviewed studies published in the English language between January 2000 and March 2024.	Non-English publications and grey literature documents; these boundaries are explicitly addressed within the limitations section.

Information sources and search strategy

Electronic systematic searches were undertaken in PubMed, Scopus, and Web of Science from their earliest years to March 31, 2024. The backward and forward snowball search of citations for all eligible articles was done up until January 2025 to identify empirical studies directly relevant to those included through such searches. One pilot study from 2025 was identified via citation searching as being directly relevant to student-focused digital health promotion intervention design. No grey literature search was performed to keep the scope comparable to peer-reviewed articles, which is a limitation.

The search structure incorporated sophisticated Boolean operators, Medical Subject Headings (MeSH), and truncation keywords related to the higher education population, digital literacy ideas, and health outcomes. The full search string performed on PubMed is presented as follows: (“digital health literacy“[Title/Abstract] OR “eHealth literacy“[Title/Abstract] OR “electronic health literacy“[Title/Abstract] OR “digital literacy“[Title/Abstract] OR “health information seeking“[Title/Abstract] OR “online health information“[Title/Abstract] OR eHEALS[Title/Abstract] OR DHLI[Title/Abstract] OR “COVID-DHLI“[Title/Abstract] OR “media health literacy“[Title/Abstract] OR “digital competence“[Title/Abstract]) AND (student*[Title/Abstract] OR university*[Title/Abstract] OR college*[Title/Abstract] OR undergraduate*[Title/Abstract] OR postgraduate*[Title/Abstract] OR “higher education“[Title/Abstract]) AND (health[Title/Abstract] OR wellbeing[Title/Abstract] OR “well-being“[Title/Abstract] OR “mental health“[Title/Abstract] OR COVID-19[Title/Abstract] OR infodemic[Title/Abstract] OR telehealth[Title/Abstract] OR “digital mental health“[Title/Abstract] OR intervention*[Title/Abstract])

Study selection and data extraction

All retrieved records were exported into reference management software for automated and manual deduplication. Deduplication was repeated using exact DOI matching, normalized-title matching, and manual verification; this audit identified 21 duplicate records, comprising 20 exact DOI duplicates and one manually verified title/abstract duplicate. Two reviewers then independently screened titles and abstracts, followed by full-text assessment. Studies were excluded if they did not involve higher education students or did not address empirical DHL, eHealth literacy, or digital health information engagement. Disagreements were resolved through consensus, with a third senior reviewer available for arbitration. Inter-rater agreement at the full-text stage was substantial (Cohen’s kappa = 0.82).

Data extraction was completed using a standardized, pre-piloted matrix. Extracted items included author, year, country, study design, sample size, participant characteristics, DHL instrument, DHL domains, socio-demographic determinants, information behaviours, mental health or well-being outcomes, intervention or implementation elements, and key findings. Extracted data were cross-checked against the master reference list to ensure alignment across the PRISMA flow diagram, Table 2, and the quality appraisal matrices.

Every single data point was cross checked against the master reference list, to ensure absolute data alignment across the PRISMA flow diagram, Table 1, and the quality appraisal matrices, without any mismatch.

Methodological quality and risk of bias appraisal

In order to reflect the architectural diversity across the studies, and to deal with risk of bias plus overall methodological quality, we used design specific appraisal instruments instead of leaning on one generic tool only. For analytical cross sectional designs, the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Analytical Cross-Sectional Studies [30] was applied, and used to look closely at each paper. For empirical qualitative work, the Critical Appraisal Skills Programme (CASP) Qualitative Checklist [31] guided the assessment.

Evidence coming from pilots, or review derived findings, was treated with extra care, following the guiding principles of AMSTAR 2 [32], so reporting would stay transparent and traceable. Since the last screening made it clear that the current higher education literature includes no full scale randomized controlled trials, (RCTs) that test the efficacy of the DHL intervention, we did not use the CASP RCT tool. The quality appraisal itself was done separately by two reviewers, and then a third person confirmed the final scoring decisions, to keep everything aligned.

Synthesis approach

1. A meta-analysis could not really be performed due to the substantial heterogeneity across the psychometric instruments, the target clinical outcomes, and also the statistical analytical models, plus the reporting metrics used throughout the included literature. As a result, a more formal narrative synthesis was set in motion, using a structured iterative thematic approach, aligned with the Guidance on the Conduct of Narrative Synthesis in Systematic Reviews [33]. This synthesis then went forward in five separate phases, sort of like a stepwise run not one single sweep, really.

2. First there was systematic tabulation and profiling of baseline study characteristics;

3. Second, inductive coding of specific DHL measurement frameworks, personal determinants, and mental health outcomes;

4. Third, conceptual clustering of those codes into recurrent structural themes;

5. Fourth, cross-national and cross-instrument comparisons to find contextual patterns that repeated across settings; and.

6. The fifth step involves the construction of an integrated conceptual model, outlining how the structural determinants relate to each other through digital information environments to individuals’ DHL competencies, then to proximal information behaviors, and ultimately to distal health outcomes.

Importantly, the scores on the quality of methodology did not act as an arbitrary cutoff point for excluding studies; rather, the assessment of the risks of bias was used to inform the level of epistemological confidence in each synthesized theme.

Results

Study selection

The database searches identified 2,155 records. After removal of 21 duplicate records, 2,134 unique records underwent independent title and abstract screening. Of these, 1,976 records were excluded because they were not empirical studies, did not sample higher education populations, lacked an explicit DHL or digital health information-seeking focus, or addressed generic non-health educational technology.

A total of 158 full-text reports were assessed for eligibility. Of these, 142 were excluded because of irrelevant or unextractable outcomes (n = 75), lack of explicit DHL or university student focus (n = 30), or ineligible publication type, including protocols, editorials, and narrative commentaries (n = 37). Sixteen studies were included from database screening, and five additional studies were identified through backward and forward citation tracking, giving a final corpus of 21 studies for narrative synthesis.

In the end, the whole process resulted in a finalized, stabilized corpus of 21 eligible studies that were used for rigorous narrative synthesis (Fig. 1).

Fig. 1.

PRISMA 2020 flow diagram for study selection

As shown in Fig. 1, the PRISMA 2020 process moved from 2,155 database records to 2,134 unique records after deduplication. The high exclusion rate at title and abstract screening was mainly due to non-higher education populations, absence of empirical DHL measurement, and generic educational technology studies without health-literacy content. The final corpus of 21 empirical papers was reconciled across the manuscript, the PRISMA diagram, the extraction matrix, and Supplementary File 2.

Characteristics of included studies

The included studies were published between 2021 and 2025. Database-identified studies were published within the March 2024 search window, while [24] was identified through updated citation tracking and treated as pilot implementation evidence. The corpus spans Europe, Asia, Africa, the Americas, and Australia. Most studies used cross-sectional survey designs; others included psychometric validation, latent profile analysis, qualitative implementation work, or methodological pilot testing. Table 2 lists all 21 studies individually by author, year, setting, design, sample size, instrument, and core finding.

Summary of quality assessment section

Methodological quality ranged from moderate to high. Typical strengths were clear research questions and objectives, higher education participant samples, consistent survey techniques, and appropriate statistical analysis techniques. A summary of the quality assessment is provided in Supplementary File 2.

The recurrent limitations were the use of convenience sampling, self-reporting, lack of information about response rate for some studies, and the prevalence of cross-sectional research design. Such limitations limit causal conclusions and necessitate a cautious interpretation of the relationship between digital health literacy and mental health/well-being. Very few studies have employed objective/performance-based measures of digital health literacy; instead, they mostly utilized perceived competence scales.

Validation papers provided excellent psychometric evidence, but they also showed structural variance in baseline factor structures across cultures, especially around data privacy protection and user generated digital content. Meanwhile the qualitative and pilot data gave deep insights into implementation dynamics, but methodologically they were underpowered, so they really couldn’t support directional causal claims.

Thematic synthesis

Theme 1: DHL as a competence for appraising online health information

A steady cross-national result showed that university students have really serious issues when it comes to critically evaluating, sorting out, and combining the huge amounts of digital health data that are everywhere online. For example, in Germany, 42.3% of the surveyed students had trouble to judge whether information is reliable, and 38.9% could not really spot the commercial motives behind it [7]. In the same direction, 49.3% of students in Slovenia [11] and 45.1% in Greece [9] said they faced strong challenges in checking if the information is actually valid. Taken together, these numbers suggest that even if someone is quite comfortable using digital tools, that alone does not mean they have strong critical digital health literacy.

Theme 2: Socioeconomic, gender, health status, and contextual determinants of DHL

It is relatively evident from research findings that DHL, just like other social factors, is strongly influenced by social gradients. For instance, the level of perceived social status, financial stability, gender, age, education, and presence of a chronic illness have been noted to be influential upstream factors affecting one’s ability to read and comprehend information. It was found that, in the case of Ghana, subjective social status and adequacy of cash in hand were key determinant factors of DHL and subsequent health outcomes [19, 22]. The same pattern has been observed among the Romanian and Hispanic population in universities [12, 23]. In general terms, this places the role of DHL within the wider context of equity, which tends to vary depending on external structures.

Theme 3: Associations with mental health, well-being, fear, and sense of coherence

DHL was repeatedly connected with psychological outcomes that are both close-by and farther out, if that makes sense. In Taiwan and Vietnam, empirical path analyses showed that DHL influences subjective well-being directly and indirectly, mainly by reducing health related worries, lowering how easily people fall for misinformation, and also improving satisfaction with information [10, 38].

In Ecuador, higher DHL turned out to be an independent predictor of life satisfaction, after taking socio-demographic controls into account [23]. This same direction of evidence was also backed by latent profile mapping in China, where students that clustered in the low DHL phenotype showed notably higher symptom levels for clinical fear and depression [13].

Lastly, in European and African cohorts, DHL was folded into broader psychosocial coping models, connecting it with a stronger internal sense of coherence and less anxiety about the future [19, 21].

Theme 4: Measurement heterogeneity and psychometric maturation

An important aspect within this field has been the adaptation of DHL tracking instruments through translation, adaptation, and psychometric validation in completely dissimilar languages. Despite the fact that different versions of the Digital Health Literacy Instrument (DHLI) were found to have been utilized more frequently, the phenomenon of measurement heterogeneity was somewhat persistent.

In validation studies of the instrument conducted among Italians, Portuguese people, and Spanish-speaking participants, high structural validity was established. However, cross-cultural discrepancies persisted at the level of domains as well. The observed inconsistencies pertained mainly to the subscales dealing with data privacy protection and information production [3, 23, 35]. This structural discrepancy hampers direct cross-country comparison of results. It is also indicative of a definite necessity to conduct testing for measurement invariance and assess competencies objectively in the task-based format.

Theme 5: The literacy-to-intervention gap

The strongest corpus-level pattern was the imbalance between descriptive measurement and intervention testing. The literature has measured DHL deficits extensively, but almost no randomized or controlled university-based trials have tested DHL curricula, digital care pathways, or institutional policies.

Emerging implementation evidence suggests that students may accept low-intensity digital mental health or health-promotion interventions when they are flexible, secure, and delivered through trusted university platforms university portals, system-level engagement barriers remain persistent [24, 25]. This literacy-to-intervention gap represents the primary translational roadblock in the field today.

Conceptual model

To synthesize the findings without relying on underpowered bibliometric modelling, we developed a conceptual framework from the thematic synthesis (Fig. 2). The model positions DHL as a downstream outcome of structural determinants and the digital information environment. DHL then shapes proximal information behaviours, which are associated with distal mental health and well-being outcomes. The literacy-to-intervention gap is framed as the key translational challenge for universities.

Fig. 2.

Conceptual model derived from the narrative synthesis

Figure 2 presents an integrative conceptual model linking structural determinants, digital information environments, DHL capacity, proximal information behaviours, and distal outcomes among university students. The model emphasizes that DHL does not exist in isolation; it is shaped by socioeconomic position, subjective social class, financial constraints, institutional infrastructure, infodemic pressure, and algorithmic information exposure.

After its adoption, DHL impacts individuals’ information behaviors, including sourcing reliable information, assessing reliability, assessing relevance, maintaining privacy, and seeking help confidently. These behaviours are associated with downstream outcomes such as subjective well-being, sense of coherence, lower fear or anxiety, and engagement with digital mental health support.

The model also identifies the literacy-to-intervention gap as a translational friction point, indicating that future university action should move from repeated baseline measurement toward scalable, evidence-based, and service-embedded educational interventions.

Discussion

Principal findings

This systematic review shows that the evidence on digital health among higher education students is best understood as an emerging DHL literature rather than a mature intervention-effectiveness literature. Most included studies were cross-sectional, survey-based, and conducted during or shortly after the COVID-19 pandemic. The final corpus comprised 21 empirical studies: 16 identified through database screening and five through citation tracking, as shown in the PRISMA 2020 flow diagram (Fig. 1).

The main finding is that students’ ability to locate, understand, evaluate, protect, and apply online health information is variable and socially patterned. As shown in Fig. 2, DHL is shaped by upstream structural determinants, including socioeconomic position, gender, chronic disease history, and subjective social class, as well as by digital information conditions such as infodemic intensity, commercial media environments, privacy risk, and algorithmic filtering.

Once set, personal DHL is rather consistently tethered to nearer information-appraisal skills and farther psychological supports, such as more noticeable information satisfaction, sharper choosing of vetted institutional sources, an increased internal sense of coherence, lower public health fears and a more optimized sense of subjective well-being. This kind of pathway, was most clearly corroborated in Vietnamese and Taiwanese structural models, where DHL worked like a kind of essential channel to life satisfaction, by steadyly buffering pandemic panic and cyberchondria [10, 38]. In turn, West African studies have found that although baseline economic background and age are indeed the determining factors influencing an individual’s vulnerabilities, the individual’s DHL functions as a precious psychological resource, which allows the student to maintain a sense of coherence and well-being during periods of crisis [19, 22].

Also, similar cross-sectional results obtained for wider samples of students from Europe, Asia, and Latin America are in line with those obtained above. In particular, high DHL serves as a kind of protective mechanism for dealing with excessive information flow; thus, it makes it possible for the student to independently find their way around digital health literacy materials and increases their trust in these materials when they are under significant media pressure [8, 9, 12, 13, 21, 23].

Interpretation in relation to the results

Four themes emerged from the empirical evidence reviewed through narrative synthesis. First, DHL research in university students is geographically diverse but remains historically anchored in COVID-19 crisis contexts. Second, most instruments focus on information searching and reliability appraisal, while privacy protection and health content creation are less consistently measured. Third, DHL is associated with mental health and well-being outcomes, including fear, depression, distress, future anxiety, subjective well-being, and sense of coherence. Fourth, tested campus interventions remain rare and are not yet sufficient to support firm conclusions about effectiveness.

Fourth, the empirical evidence that actually addresses tested campus interventions is surprisingly rare and not mature enough to back any firm conclusions about operational effectiveness. These structural tendencies are very consistent with the broader macrosocial evidence produced by the international COVID-HL network, which basically mobilized fast cross-national mapping to document student vulnerabilities during peak pandemic institutional closures [7, 34–36]. These results get additional support from psychometric validation work, because it indicates that variants of the Digital Health Literacy Instrument (DHLI) tend to keep high structural fitness across different linguistic settings, including Portuguese, Italian, and Hispanic groups, but the field has generally moved forward along a single lane of measurement validation and correlational mapping [3, 23, 35].

Underlying drivers of the measurement-heavy evidence base

The overwhelming dominance of descriptive cross-sectional designs in the current DHL literature is methodologically understandable, at least given the historical window of data gathering.The unexpected arrival of the pandemic’s infodemic led to an immediate high stakes call for a quick analysis of the ability of students to decipher the volatile institutional safety protocols amid the global surge of misleading information online. It is the “machine” of the COVID-HL consortium that, in a way, provided the researchers with the opportunity to deploy web-based survey methods quickly, providing a relatively efficient and also fairly comparable baseline of data points [7, 8, 36].

However, this diagnostic phase did not translate quickly into intervention science. Barriers include psychometric heterogeneity, inconsistent cut-off criteria, limited longitudinal follow-up, lack of objective task-based testing, and institutional fragmentation across student health services, counselling units, information technology departments, and university libraries. Experimental interventions in this field also raise ethical issues related to data tracking, privacy protection, and exposure to misinformation.

Addressing such a problem as a measurement-intervention gap is essential because contemporary student wellbeing promotion, as well as digital mental health interventions, hinge solely upon the ability of students to subjectively determine the quality of the application and address all possible risks concerning data privacy. It is apparent from qualitative data gained during clinical trials that some consistency emerges among different student populations: specifically, there is a high level of conceptual acceptability of low-intensity applications for digital psychiatry. However, engagement with such technologies does not seem to correlate primarily with users’ attitude towards it but rather depends upon institutional approval, clearness of the platform, and its seamless integration into the process of education [25]. Moreover, technology acceptance mapping suggests that digital health interventions in LMI countries can prove to be a highly efficient strategy when it comes to overcoming service stigma and structural accessibility problems. At the same time, actual usage of the tool remains determined primarily by users’ perception regarding their usefulness and ease of operation, along with the discussion of these instruments in the local context [26].

Theoretical interpretation

These findings can be interpreted through three complementary theoretical lenses: health literacy theory, salutogenesis, and the Health Promoting University framework.

First, classical health literacy theory conceptualizes literacy as an evolutionary matrix of functional interactive and critical competencies that you need to access, understand, evaluate, and use health information. DHL expands that paradigm too, basically by layering on its own technological pressures, like digital network navigation, asynchronous communication, awareness of data encryption, and platform specific credibility checks [1, 2]. This extension is fairly strongly supported by the validation studies in our corpus, which show that DHL is a separate multi-dimensional psychometric construct that can’t really be swapped out with general media or computer literacy scores [3, 23, 35].

Second, Aaron Antonovsky’s sociological theory of salutogenesis gives a kind of deep mechanism for why individual DHL directly maps onto better psychological well-being. Salutogenesis argues that keeping health is tied to a strong internal sense of coherence, that is the ongoing cognitive belief that your internal and outer realities are, in a broad way, predictable, controllable, and worthwhile. Students who have high DHL are set up to read messy conflicting online health feeds as coherent and manageable evidence, so decisional uncertainty gets softened and maladaptive anxiety spirals don’t take hold. This mechanism is directly corroborated by the African and European path analyses in the corpus, where sense of coherence shows up as a key mediator between earlier literacy and safeguarded mental health outcomes, also associated with less future anxiety across different socioeconomic segments [19, 21].

Third, the Health Promoting University framework, codified through the Okanagan Charter, shifts the interpretive lens away from individual skill deficits and it kind of reallocates responsibility to systemic institutional operations [28]. Under that ecological lens, DHL is not simply a personal asset, but a direct reflection of organizational design. A health-promoting university is structurally expected to embed health metrics straight into campus governance, policy, the physical settings, and digital communications [28]. Our conceptual framework (Fig. 2) respects this entire system style by placing ambient infodemic intensity and institutional digital design as the main structural drivers that shape individual student literacy outcomes, kinda in a cause-and-effect way.

Implications for university administrators

The results have implications that can be used in practice by university administration, student services officers,

librarians, and student health service personnel. DHL is not just an additional skill but a necessary student competency. The process of training needs to take place during first year orientation, library research instruction, student health service induction, and digital mental health onboarding.

Furthermore, efforts should be made to create credit-eligible micro-certifications or certification modules that target DHL in particular, with an emphasis on skills such as evaluating sources, identifying algorithmic bias, detecting misinformation, data security, and appropriate use of telehealth technologies on campus.

Further, empirical proof shows that numerous learners are still struggling with recognizing business-related content online. Besides, the assessment of the correctness of the received information is not an easy task for students, and hence passive dissemination of information is totally insufficient (“just provide information”). Given the extensive use of search engines by students along with excessive uncontrolled use of commercial social networking sites, structured assessment models are essential [9, 12].

Further, there is a need for higher education institutions to develop their institutional digital platforms, learning management systems, and student guidance portal in an integrated and quality manner. In case more trust is desired on the platform and user retention in the long run, these portals should be developed through a student co-design process. This will also almost fit into place with the qualitative findings that the subjects have high levels of compliance and confidence in clinical settings when digital interventions are tied to the infrastructure of counseling [25].

Implications for policymakers and health systems

Regarding national policy level, educational institutions and public health bodies must incorporate student-based DHL indicators in their quality assurance framework and digital health strategy. Relevant indicators include reliability verification, privacy competence, digital mental health help-seeking, and equitable access to university-endorsed health information.

It can be seen from the above analysis that it would be imperative for equity measures to remain a key consideration in the DHL policies of any nation. With regard to literacy acquisition that depends heavily on socio-economic gradients, such as socio-economic position, social class perception, age, gender, and chronic disease status, it would be essential for governments to dismiss the simplistic notion that all members of generation Z have similar abilities as far as digital tools are concerned [12, 19, 21, 22].

In countries with low to middle economic status, students’ DHL approach must consider efforts geared towards making data affordable, mobile-friendly architecture, and multilingual communication without making assumptions about how people learn. Groups of higher education students within developing health care systems have proven highly receptive to online psychiatry tools due to the ability to bypass social and institutional barriers associated with mental help in the local setting [26].

Emerging frontiers: AI-enabled mental health and digital trust

Rapid changes in digital environments require DHL frameworks to move beyond static website appraisal toward AI-health literacy. Students increasingly encounter AI chatbots, symptom checkers, algorithmic triage tools, and personalized health feeds. They therefore need skills to interpret probabilistic outputs, detect hallucinations and bias, protect sensitive information, and distinguish supportive digital tools from formal clinical care.

The included studies were not designed to assess AI-specific DHL, so this area should be treated as an emerging research agenda rather than a confirmed finding of the present review. Nevertheless, responsible AI integration, AI-assisted psychotherapy, conversational agents, and digital trust are becoming directly relevant to university mental health and student health information systems [39, 40, 41].

Moreover, the optimization of such new digital frameworks also requires constant re-evaluation of digital health literacy in light of this flexible skill that continues to change over time. This is especially critical where the usage of chatbots powered by large language models, alongside other AI enhancements, has started gaining popularity within the realm of mental healthcare [39]. Moreover, in order to ensure the reliability and availability of such digital health frameworks in the future, there is an increased need to give due importance to implementation evaluation studies that rely on well-established frameworks that measure the effectiveness of these digital health interventions [42].

Strengths and limitations

This review has several strengths. First, the manuscript has been realigned around DHL pathways, matching the actual evidence base. Second, the study-selection process has been reconciled using a PRISMA 2020 flow diagram with a final corpus of 21 studies. Third, the extraction matrix, thematic categories, reference list, and quality appraisal file were harmonized to remove numerical and roster inconsistencies.

Fourth, risk of bias was assessed using design-appropriate tools: JBI for cross-sectional studies, CASP for qualitative evidence, and AMSTAR 2-informed principles for pilot or review-derived evidence. Fifth, causal language was removed and the synthesis is framed as observational and associational.

Several limitations should also be acknowledged. First, restricting inclusion to peer-reviewed English-language literature may have introduced language and geographic bias. Second, exclusion of grey literature may have omitted institutional reports or unpublished campus pilot programmes.

Third, heterogeneity in DHL instruments, subscales, scoring, and outcomes prevented meta-analysis. Fourth, reliance on cross-sectional survey data and self-reported measures increases vulnerability to common-method variance and limits causal inference. Fifth, the narrowing of the protocol from broad intervention mapping to DHL-focused synthesis is a protocol deviation, although it has been transparently reported and addressed through independent re-screening and full corpus reconciliation.

Future research agenda

To move beyond descriptive, measurement-heavy evidence, future digital health research in higher education should prioritize five translational directions:

Experimental Intervention Design: shift from cross-sectional mapping toward experimental, quasi-experimental, stepped-wedge, and longitudinal designs that can assess the long-term effectiveness of organized DHL curricula on clinical health outcomes.

Objective Competence Assessment: Build and validate performance based, task centered testing tools that directly watch real time privacy set ups, misinformation detection accuracy, and source validation behaviors—rather than leaning on self perceived efficacy scales, because those kinda blur what people can actually do.

Testing for Measurement Invariance: Conduct stringent tests of psychometric measurement invariance with respect to the DHLI and its various forms, to ensure that the construct is reliable and can maintain the same meaning and stability across a number of cultures and disciplines, including socioeconomic background, gender identity, and disability status.

Focus in Implementation Science Research: Use existing validated models of implementation science like RE-AIM and CFIR to measure outcomes of interest, including acceptability and fidelity in organizations, system uptake, equity, reach, and cost-effectiveness of DHL modules within university settings, to ensure that “adoption” does not remain an empty word.

Modeling of AI-Health Literacy: Theoretical and empirical differentiation of health literacy specifically associated with AI, followed by the study of its practical effects on the behavior of college students in regards to seeking help for their psychological well-being, the security of algorithmic data, and digital campus management.

In conclusion, this synthesis provides insight into the idea that DHL is more than just a characteristic of individual students but also requires institutional management and public health considerations. Further research into higher education should consider whether incorporating DHL in educational curriculums could lead to credible information engagement, privacy protection, and enhanced well-being among students.

Conclusion

DHL is a valuable competency among the population of universities, which involves dealing with advanced digital environments for health, shifting public health recommendations, and digital mental health systems on campuses. From the synthesized research evidence above, one can clearly deduce that individuals with high levels of DHL are more likely to access authentic information channels, have greater coherence, lower health concerns, and improved subjective well-being.

Nonetheless, the limitations inherent to the research body are tied to its post-COVID-19 era relevance, observational cross-sectional study design, and utilization of self-report measures. At this point, the need is for a shift from studying the literacy gap among students to creating and evaluating DHL-based interventions using the Health Promoting University model.

Authors’ contributions

A.C. Conceptualization, Methodology, Writing – Original Draft, Supervision, Writing – Review & EditingA.P Methodology, Supervision, Writing – Review & EditingP.C. Data Curation, Writing – Original Draft, Supervision, Writing – Review & EditingR.V Writing – Original Draft, Supervision, Writing – Review & Editing.

Funding

Open access funding provided by Symbiosis International (Deemed University). This study was not supported by any specific grant or funds from external sources for financing the research.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The study was conducted as a systematic review of published literature that was not conducted using primary data collection on human subjects or animals or personal records.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.