A literature search strategy was first developed by connecting the 2 main keywords “MedlinePlus” and “Medline Plus” with the Boolean operator “OR” and then customizing the query based on the field codes of the chosen citation database. This strategy was executed across 9 citation databases, including PubMed, Embase, CINAHL, Global Health, PsycINFO, Cochrane Database of Systematic Reviews, Scopus, Web of Science, and Dimensions. The first 5 databases are well-known research article indexing resources in the health sciences and are primary sources for clinical studies, including Health IT and digital health interventions. We also searched the Cochrane reviews, the gold standard for systematic reviews in biomedical and health sciences, to identify relevant reviews and examine their cited empirical studies, ensuring comprehensive coverage of the literature. Furthermore, Scopus, Web of Science, and Dimensions were included because they index a broad range of peer-reviewed research articles across medicine and health sciences and are commonly used in systematic reviews and evidence-mapping studies. Our initial search was conducted in October 2021 and was updated in October 2022 and in September 2024 (Multimedia Appendix 1). In addition, we checked all references of the included articles (backward searching) and newer articles that cited the included studies (forward searching) in Scopus for an exhaustive literature search.

The search results from each database were exported, aggregated, deduplicated, and imported into Covidence [11] for screening. Independent reviewers from our research team conducted the screening process, initially at the title-abstract level (FY, Erin Harvey, and Pengxuan Wang), followed by a full-text assessment (FY and XW). Discrepancies between reviewers were resolved through discussion to achieve agreement.

The studies were included if they (1) used MedlinePlus as the primary or one of the primary information or education resources for patients and/or their caregivers; (2) targeted patients or caregivers, as indicated by the recruited participants or by the research aims; (3) were English literature; and (4) were original research articles. The studies were excluded if they were (1) non-English literature, (2) did not use MedlinePlus as one of the information sources, (3) did not include patients or caregivers as direct information recipients (eg, only physicians or other health care professionals are information or training recipients), (4) were not original research articles (eg, perspectives, comments, editorials, news, bulletins, erratum, or abstract only), (5) were information readability or coverage assessment studies, or (6) were information retrieval, search behavior, or query analysis studies.

To achieve the research aim, we adopted 2 frameworks to guide our data extraction and synthesis process. The first is the PICO (problem/population, intervention, comparison, and outcome) framework [12], widely recognized and used in evidence-based practice and research in health sciences. The PICO framework systematically addresses the targeted population, specific intervention or exposure, intervention comparison to an alternative (if applicable), and desired outcomes or end points. The second framework is the Collaborative Chronic Care Model (CCCM) [13], a refined version of the widely adopted original chronic care model. The CCCM emphasizes collaborative, continuous, evidence-based, and proactive care delivery for individuals with chronic, medical, or mental health conditions. The CCCM encompasses 6 major elements: health system organization, work role redesign, self-management support, decision support, information management, and community linkages. We created a customized data extraction framework (Table 1) and organized the data accordingly. Guided by the PICO and CCCM frameworks, our data extraction focused on the following 4 aspects: bibliometrics, participants, interventions, and outcomes.

Bibliometrics capture the document details of the included studies, including reporting year, source title, funding sources, author names and affiliations, and citation impact. We assessed the citation impact of included studies using both raw citation counts (ie, how many times an article has been cited) and the field-weighted citation impact (FWCI), a field- and time-normalized citation impact metric [14] provided by the Elsevier Scopus database. An FWCI of 1 indicates that, compared with papers published in the same field and time, this article is cited at the global average level; a value greater than 1 suggests that it is cited more than the worldwide average and vice versa. Being an improved citation impact measure, FWCIs have been commonly adopted by research impact evaluators [15,16]. In addition, we used Altmetrics to capture the publication’s digital attention [17], with scores obtained from Altmetric Explorer [18] through an institutional subscription. The composite Altmetric Attention Score (AAS) is a weighted and quantified measure of the online attention and engagement that a research output (eg, journal article) has received by tracking various sources such as public policy documents, mainstream news, social networks, blogs, and other platforms. Generally, an AAS of 20 or higher is considered good [19] and is often used as a benchmark in research evaluations [20].

The participant category captures participants’ roles, patient diagnoses, CCCM-aligned demographics (including social determinants of health [SDOH]), sample size, and age range of patients and caregivers who used MedlinePlus. Participant roles include physicians, nurses, research associates, patients, and caregivers involved in using MedlinePlus as an information intervention. Demographics include SDOH, such as self-management skills, education and literacy levels, and economic and financial factors (eg, income, employment, and health insurance). Reporting SDOH and demographic characteristics (eg, race, gender, and primary language) provides contextual information and aligns with CCCM's emphasis on whole-person care and linkage to community resources.

Interventions describe how MedlinePlus was used as a health information intervention, including the type of content assessed, platform features used, and modes of delivery. Using the CCCM-aligned construct of intervention delivery, we systematically documented the key intervention attributes, including intervention features, health information content, any additional interventions beyond the MedlinePlus resource, and the settings where the interventions were implemented.

Outcomes cover clinical and nonclinical outcomes as well as outcome comparison and assessment, if applicable. Clinical outcomes are associated with “a measurable change in symptoms, overall health, ability to function, quality of life, or survival outcomes that result from giving care to patients” [21]. Nonclinical outcomes encompass a wide range of measures, including participants’ attitudes toward health information resources, health information–seeking behavior, experience with health information interventions, information usage, and perceptions. We also examined the study durations and follow-up methods used to assess outcomes of MedlinePlus as an information intervention.

In total, 3 reviewers (FY, Pengxuan Wang, and XW) independently extracted data from the Scopus database, Altmetric Explorer, and the full-text articles of the included studies. We recorded data in a Microsoft Excel template for each included study based on our data extraction framework and discussed and reconciled data inconsistencies to reach consensus.

Two reviewers (FY and XW) independently assessed the quality of the included studies using the Mixed Methods Appraisal Tool (MMAT) [22] by following the MMAT user guide. MMAT has been widely used in systematic reviews to appraise qualitative studies, quantitative studies, randomized controlled trials (RCTs), nonrandomized studies, and mixed methods studies [23,24]. The MMAT tool offers a checklist. First, each study undergoes 2 screening questions to ensure that it is an empirical study (ie, “S1. Are there clear research questions?” and “S2. Do the collected data allow addressing the research questions?”). Second, reviewers select an appropriate study category for each included study. Third, based on the 5 criteria in the selected category, reviewers rate each criterion as “Yes,” “No,” or “Can’t tell,” documenting any observations in a “Comments” column. We rated “Yes” when a study provided sufficient information to fully comply with the criteria, “No” when the study failed to meet the criteria, and “Can’t tell” when the information in the full-text article was insufficient, unclear, or partially met the criteria.

After discussing and resolving discrepancies between the 2 reviewers’ independent ratings, the research team achieved consensus and presented the final rating results at the criteria level for all included studies, following the MMAT user guide (Multimedia Appendix 2).

In total, 28 studies reported in 27 sources met our inclusion criteria (Figure 1). Cohen κ measured the interreviewer agreement, which was 0.63 at the title-abstract level and 0.9 at the full-text level, indicating substantial agreement between reviewers. The results of all included studies are delineated by 4 categories of measures in Tables 2-5.

The included studies were reported between 2004 and 2024 (Table 2). The year 2010 had the highest number of studies (n=4), followed by 2013 (n=3). The years 2006, 2007, 2012, and 2016 each had 2 studies, while all other years had either 1 or no studies (Figure 2).

Of the 28 studies, 25 were conducted in the United States, with 2 conducted in Iran [34,44] and 1 conducted in Argentina [4]. The document types varied and included 1 book chapter [34], 4 health informatics conference proceedings (ie, AMIA Annual Symposium [5], Studies in Health Technology & Informatics [4,37], and IEEE International Conference on Healthcare Informatics [46]), and 1 doctoral dissertation [43]. The remaining studies were published in 12 academic journals. The Journal of Consumer Health on the Internet was the most frequent publication venue, with 5 publications, followed by the Journal of the Medical Library Association (4 publications), the Journal of the American Medical Informatics Association (2 publications), and Information Services & Use (2 publications).

In total, 20 publications reported funding support from 9 agencies. The NLM was the primary funder, supporting 15 studies. Other funders included several US federal agencies (eg, Agency for Healthcare Research and Quality, National Center for Advancing Translational Sciences, National Institute of Nursing Research, and Department of Defense), 1 foundation program (ie, Robert Wood Johnson Generalist Faculty Scholars Program), and Tabriz University of Medical Sciences in Iran.

A total of 125 individuals from 65 organizations contributed to the included studies. Notable contributors included authors from the NLM (3 publications), George Washington University (2 publications), University of Missouri (2 publications), and Columbia University Medical Center (2 publications).

Of the 27 sources, 22 were cited in scholarly works. Eight publications had FWCI scores above 1.0, with the study by Ramesh et al [37] having the highest FWCI of 5.79 (Table 2), indicating that the publication has received more citations than expected for its field. Altmetric scores were available for 11 publications, with the highest score of 63 by McCarthy et al [35], indicating that it is widely shared, discussed, and referenced across multiple platforms. This paper is also the only one with an AAS above 20, indicating high engagement (frequent mentions across multiple sources, including news, policy reports, and mainstream discussions).

Of 28 studies, 18 involved outpatients, while 5 focused on inpatients and their family members. Notably, 2 studies designed digital information interventions incorporating MedlinePlus but recruited their participants for testing either through crowdsourcing [37] or from a local community [45]. Patients represented a range of clinical conditions, such as diabetes, cardiovascular or heart valve diseases, pediatric cancer, heart failure, and depression. More than half of the studies recruited patients with multiple or varied diagnoses (Table 3).

In addition to patient participants, several studies also involved health care professionals: 12 involved clinicians with various backgrounds, 3 targeted nurses, 2 included research assistants, and 2 involved other hospital staff. Community-based participants comprised librarians from medical, health sciences, or public libraries (9 studies) and pharmacists or pharmacy students (3 studies).

SDOH factors were captured in the patients’ demographic data. The studies explicitly addressed underserved and low socioeconomic status groups [29,38,41,45] and participants with low [31] or varied health literacy levels [37,38]. Several studies either focused on or included racial and ethnic populations, such as African American populations [26,38], Spanish-speaking underserved populations [29,32,45], indigent patients [31], Asian populations [37], Persian populations [34], or diverse race populations [5].

Health insurance status was detailed in 6 studies [5,6,25,31,35,43], with 3 indicating that partial participants were uninsured, while 2 reported that the vast majority were insured. While 1 study reported that 148 (75%) of 197 study participants had an annual income exceeding $50,000 [25], other studies reported low income or unemployment among their participants [29,31,38,41,45]. In addition, internet access was addressed in multiple studies. While 3 studies reported secure internet access [6,25,45], 2 studies indicated that participants had limited internet or computer access [28,31], highlighting persistent access barriers relevant to digital health implementation.

The sample sizes in these studies ranged from 10 to 12,877 participants. Most studies had sample sizes under 300 participants, while several studies reported larger samples exceeding 500 participants [5,7,28,31,35,36]. Among studies reporting age, participants ranged from 17 to 88 years, with mean ages typically ranging between 35 and 65 years.

Table 4 presents that 13 studies explicitly identified their main intervention feature as either “information prescription” or “information Rx.” Two of them [6,7] investigated the effectiveness of IPs delivered via paper or email. Additionally, 3 studies [31,33,41] investigated the use of kiosks to deliver multimodal MedlinePlus information. While most studies used MedlinePlus resources on desktops or laptops, 2 studies explored iPad or tablet use [41,42], and 6 studies embedded MedlinePlus information as hyperlinks in patient portals of EHRs [4,5,32,36,37,39].

Primarily using MedlinePlus web resources, the interventions encompassed a broad range of health information topics, tutorials, and medical term explanations. Additional intervention practices included (1) hands-on computer and health information search training for patients [25,39,41,45]; (2) training for physicians, nurses, staff, and students [6,26,41,42]; (3) the use of other web-based health resources provided by health professionals or organizations [30,34,42,45]; (4) the distribution of supplementary MedlinePlus materials such as bookmarks, pamphlets, brochures, magnets, or waiting room TV loops featuring MedlinePlus videos [9,27,31,33,35,36,40-42]; (5) patient health education or counseling offered by health care professionals as part of the intervention [32,33,35]; and (6) access to librarian assistance, instruction, or training [7,31,35,40,44]. Two recent studies either customized MedlinePlus content with additional local community components (eg, recipes, infographics, and local resource listings) [45] or integrated MedlinePlus into a digital platform with the research team’s self-curated health information [46].

Most studies implemented MedlinePlus interventions in a clinical setting, such as clinics or hospitals, while others used MedlinePlus in community pharmacies [26,33], a community health resource center [45], libraries [30,36], an online platform [46], or patient portals [4,5,37].

Overall, studies reported more nonclinical outcomes than clinical outcomes (Table 5). Of 28 studies, 4 measured clinical outcomes [29,35,43,44]. Two of these studies reported significant differences between the intervention and control groups. For instance, Fawcett [43] reported significant changes in diabetic foot ulcer measurements (F_2,26=3.91; P=.03; and χ²₂=18.5; P<.001). Kazemi Majd et al [44] found fewer patients in the intervention group died or were readmitted compared to the control group during 1 year (relative risk 0.47, 95% CI 0.20‐1.06; P=.005), but not in the first 6-month follow-up (relative risk 0.6, 95% CI 0.15‐2.40; P=.05). However, the other 2 studies [29,35] reported no significant differences in clinical outcomes between the intervention and control groups, although a difference in patient behavior was noted, with the intervention group being more likely to use the internet or fill their prescriptions.

Most studies (24/28) reported positive nonclinical outcomes, focusing on (1) participants’ attitudes, satisfaction, and behaviors related to online health information resource usage and information-seeking behaviors; (2) frequency and user experience of MedlinePlus usage; and (3) perceived usefulness, trust, ease of use, and competency in using MedlinePlus information for health decision-making, physician communication, self-management, and self-efficacy on finding professional help and disease knowledge.

The study durations varied, typically around 12 months, ranging from 1 to 34 months. Most studies used surveys or interviews for participant follow-ups.

All studies were empirical with clear research goals, objectives, or research questions, and they collected data to address their research questions, achieving MMAT S1 and S2. Among them, 2 qualitative studies were analyzed by MMAT criteria from 1.1 to 1.5, and only one criterion (ie, MMAT 1.1) was met in each study [34,36]. For 7 RCTs evaluated by MMAT 2.1 to 2.5, the number of criteria met ranged from 1 to 5, with many studies lacking sufficient details on either the outcome assessors blinded to the intervention (MMAT 2.4) or follow-up response rates of adherence to the intervention (MMAT 2.5). Each of the 4 nonrandomized studies met at least two criteria (eg, MMAT 3.2 and 3.5), and only one study [5] met MMAT 3.4 by addressing potential confounding variables adequately. Seven quantitative descriptive studies met 2 or 3 criteria (MMAT 4.3 and 4.5), and the majority rarely provided sufficient information about their sampling strategies or reduced nonresponse bias risk (MMAT 4.1, 4.2, and 4.4). Finally, among 8 mixed methods studies, 1 did not meet any criteria, while 6 met only 1 criterion. The study by Wilcox et al [39] met all 5 criteria. Most mixed methods studies lacked sufficient details on whether and how the qualitative and quantitative findings were integrated and interpreted (MMAT 5.3). Overall, when considering the average number of MMAT criteria met per study type, qualitative studies had the fewest met criteria (one per study), while nonrandomized studies had the most (ie, an average of 3.25 criteria per study). Detailed quality assessment results were documented in Multimedia Appendix 2.

As the primary funding agency for half of the included studies, the US NLM has sustained MedlinePlus as a reliable and trusted health information resource. While most included studies took place in the United States, 3 studies were conducted in developing countries, demonstrating MedlinePlus’s applicability and value across cultural, medical, and organizational contexts [4,34,44].

Despite nearly 25 years of availability, MedlinePlus has generated a relatively small body of empirical research examining its role in clinical care and chronic disease management. The limited number of identified studies (n=28), coupled with generally below-average citation impact (ie, FWCI scores <1.0) and below-benchmark Altmetric scores (<20), suggests that MedlinePlus has been underused and underexamined in health care practice and research. These gaps highlighted missed opportunities to rigorously document its implementation effectiveness and potential contribution to patient- and caregiver-centered care.

Guided by the CCCM framework, this review highlights the engagement of diverse participant groups with MedlinePlus across coordinated and team-based care settings, including health care professionals, inpatient and outpatient contexts, patients with varied health conditions, family caregivers, pharmacists, and librarians. In particular, medical or public libraries often played a critical role in “physician-directed” information delivery [6], improving community health literacy [47], and serving as community partners to make MedlinePlus accessible to broader audiences. These findings underscore its potential role in integrated care delivery.

Although most studies reported patients’ ages, races, and socioeconomic backgrounds, many lacked SDOH-related specifics such as health insurance, annual income, employment, or health literacy. This gap limits our understanding of how sociodemographics impacted MedlinePlus use, especially given the strong association between health literacy, internet access, and digital engagement [48].

As an information intervention in clinical care, early studies on MedlinePlus [25,26,28] were directly influenced by the paper-based IP or Information Rx program launched by the NLM in 2002. However, later work reflected a growing interest in integrating MedlinePlus into EHR-enabled environments. For example, Ramesh et al [37] highlighted earlier efforts to embed MedlinePlus within Health IT systems. This trend continues, with more recent studies incorporating MedlinePlus into patient portals [4,5,39,41] and stand-alone digital health platforms for patients and caregivers [45,46], supporting their self-management and health information access beyond traditional clinical settings.

Several studies explored the use of kiosks in community pharmacies to deliver multimodal MedlinePlus information [31,33,41], aligning with the community linkage in the CCCM framework. Other studies also reported disseminating MedlinePlus information via iPad or tablet, or combining MedlinePlus content with additional resources, such as the American Heart Association patient materials, RxList, Mayo Clinic, and even religious texts (ie, the Holy Quran). These interventions were implemented across settings, from clinics to virtual environments, highlighting MedlinePlus’ adaptability and versatility in supporting varied models of health information delivery.

Clinical outcomes were reported in only 4 studies. Their results varied from no to significant differences in assessed measures, sometimes with mixed results even within the same study. These inconclusive results indicate the complex nature of directly linking MedlinePlus information interventions to patient health improvements. Researchers identified barriers, including inconsistent physician participation [27], small sample sizes [6], low computer literacy [29] or digital divide [5], limited evaluation periods [29], and information intervention design challenges [35]. Nevertheless, most studies reported nonclinical outcomes, including notable improvements in participants’ attitudes, satisfaction, and behaviors related to online health information seeking. These nonclinical outcomes highlight the benefits of using MedlinePlus in enhancing patients’ health engagement, communication, education, and self-efficacy, providing strong evidence to inform physician-directed health information interventions that complement routine clinical care.

Overall, in light of the identified research gaps, we recommend the following priorities for future research. First, additional studies are needed to rigorously examine the impact of MedlinePlus on patient- and caregiver-centered health outcomes and ensure transparent reporting and effective dissemination of findings within and beyond scholarly communities. Second, more rigorously designed RCTs are needed to evaluate its clinical efficacy. The existing RCTs missed or underreported several key methodological components. Thus, future studies should provide detailed information on allocation concealment, randomization implementation, and follow-up response rates regarding participants’ adherence to the intervention. Third, digital solutions should seamlessly integrate MedlinePlus content into patients’ health care journeys, such as through personalized apps or culturally adapted digital health platforms. Although the most recent 2 studies [45,46] suggest these burgeoning trends, future studies shall ensure that patients receive relevant information tailored to their specific needs, preferences, and literacy levels. Finally, future research shall consistently collect and report SDOH-related demographics, such as income, insurance, education, and health literacy, and support MedlinePlus as an equitable intervention responsive to the care needs of patients and caregivers of different backgrounds.

Despite its global importance, MedlinePlus faces challenges from rapid technological advances (eg, keeping pace with evolving web design and digital infrastructure), competition from commercial products (eg, Healthwise within EPIC EHRs), and uncertainties about the long-term sustainability of federal funding amid federal priority shifts. However, the synthesized evidence from this review demonstrated consistent nonclinical benefits and clinical potentials, validating MedlinePlus’ enduring value and utility. The findings provide evidence to inform future Health IT innovations, investments, and policy decision-making.

This study has several limitations. First, although we aimed to identify all relevant studies, we may have missed those published in languages other than English or not indexed in the 9 bibliographic databases we searched. Second, we excluded articles that reported the use of MedlinePlus for nonpatient or noncaregiver populations. While this exclusion helped us focus on MedlinePlus’s role in clinical care, it might have limited the interpretations of MedlinePlus’s real impact beyond clinical contexts. Third, the included studies are heterogeneous in their research designs, and many lack demographic information on socioeconomic status, constraining a comprehensive understanding of how SDOH-related demographics influence MedlinePlus utilization. Future studies shall address this gap by ensuring more complete and consistent demographic data collection and reporting.

This systematic literature review is the first to examine how MedlinePlus has been integrated into clinical care, supporting patients and caregivers through increased health information access. Researchers agree that promoting MedlinePlus resources will help address the digital divide, particularly for individuals who are homebound, reside in rural areas, or are underserved. Although health care providers are the key stakeholders in advocating MedlinePlus, the effective delivery and optimization of MedlinePlus health information for patient care requires collective and coordinated community efforts. Our synthesis through the CCCM-guided lens will serve as a foundation for future research and practice in designing and developing accessible, equitable, and scalable digital health information solutions to improve patient engagement, health literacy, self-management, and health outcomes.