This systematic review was designed following the JBI Manual for Evidence Synthesis [19,20] and the Cochrane Handbook for Systematic Reviews of Interventions [21].

This systematic review captures peer-reviewed research papers addressing digital preventive services uptake targeting older adults. The digital tools considered included telephone (automated or synchronous), text messaging, web-based educational materials or materials delivered via video, or patient portal. The search strategy was developed in consultation with a health sciences librarian (RJ; see Multimedia Appendix 1 for the search strategy from Embase). Search terms were selected based on the review question’s main content areas of older adults and clinical preventive services and included a combination of keywords and subject headings. Databases used in the search included MEDLINE, CINAHL, Embase, and PsycINFO. The search strategy was modified only slightly between the databases to incorporate their specific subject headings. Database searches were conducted on December 11, 2023, and updated on July 13, 2025.

Eligible literature included experimental (eg, randomized controlled trials) and quasi-experimental study designs. Qualitative and observational studies that did not test an intervention were excluded. Studies were included if they addressed preventive services uptake among older adults in response to single (eg, breast cancer screening, colorectal cancer, and blood pressure) or multiple services (eg, blood pressure, cancer screening, and seasonal vaccination). Interventions were limited to high-priority preventive care services based on high clinically preventable burden and established immunization schedules for older adults [8,22]. Preventive care services included 5 related to screening (cardiovascular disease, type 2 diabetes mellitus, hypertension, colorectal cancer, and breast cancer) and 6 related to vaccines (seasonal influenza, pneumococcal polysaccharide, shingles, COVID-19, and tetanus and or diphtheria).

Studies were eligible if they included samples that were primarily community-dwelling older adults. Because some studies did not report disaggregated age and used broad age categories (eg, “older adults” defined as 55+ years), we applied the following rules: studies in which all participants were aged 60 years or older were considered eligible; and if participants younger than 60 years were included, at least 50% of the sample had to be aged 65 years or older. This approach ensured consistency with our focus on older adults while allowing inclusion of studies with varied reporting practices. Studies were excluded if they focused on subpopulations (eg, familial risk and rheumatoid arthritis) or noncommunity-dwelling older adults (eg, long-term care residents) because they do not reflect the asymptomatic population for which preventive services guidelines are designed. Publications were limited to the last 10 years from when the review began to capture current strategies [9,22]. Studies were excluded if they did not report the results of primary research (ie, editorials, letters, search and research protocols, or systematic reviews).

Two reviewers (LB and MAS) completed abstract screening and full-text screening, blinded to each other’s decisions. Search results were added to the abstract management software Covidence for deduplication. A small sample of 10‐15 records was tested against the stated inclusion and exclusion criteria [23]. The initial test of the screening process involved calibrating agreement, clarifying criteria, and making any necessary modifications before full screening. Following screening calibration, the remaining screening was conducted independently by the 2 reviewers (LB and MAS) [23]. Conflicts were assessed and resolved using consensus by 2 reviewers (LB and MAS). Interrater reliability was calculated using Cohen κ within Covidence software. The Cohen κ for abstract screening was 0.68, indicating substantial agreement [24].

Quality of evidence was assessed using the Cochrane revised tool to assess risk of bias in randomized trials (Risk of Bias 2 [RoB 2]) [25] and the Risk of Bias in Non-randomized Studies—of Interventions (ROBINS-I) tool [26]. Both tools concentrate on the study’s internal validity [25,26]. In both tools, bias refers to a tendency for the results to deviate systematically from what would be expected from a randomized trial, conducted on the same participant group with no flaws in its execution [25,26]. RoB 2 assesses 5 domains of bias: (1) randomization process, (2) deviations from intended interventions, (3) missing data, (4) measurement of outcomes, and (5) selection of reported results [25]. For the RoB 2 assessment, each domain is assigned a risk of bias assessment of low, some concerns, or high risk of bias [25]. ROBINS-I assesses six domains of bias: (1) confounding, (2) selection of participants, (3) classification of interventions, (4) missing data, (5) measurement of outcomes, and (6) selection of reported results [26]. For the ROBINS-I assessment, each domain is assigned a risk of bias assessment of low, moderate, serious, or critical [26]. An overall risk of bias assessment for each paper is then conducted for both RoB 2 and ROBINS-I tools based on the assessment of the domains [25,26]. Risk of bias assessments were conducted by 1 team member and verified by a second reviewer.

Data extraction was conducted using a standardized form. Elements for extraction included study characteristics, study population, intervention characteristics, type of preventive service targeted, single or multiple services, and change in preventive care uptake. Preventive services uptake was defined as the completion of a screening test or immunization, either by self-report, review of medical chart, or insurance claims. Data extraction was conducted by 1 research assistant and confirmed by 2 additional reviewers (MAS and LB).

Data were synthesized using a narrative approach to explore the effectiveness of digital preventive services interventions for older adults. This method was suitable, given the variability in designs and interventions among the studies analyzed [27]. Technology complexity was determined using a researcher-developed scale of low to high based on the degree of interactivity required of older adults to receive the preventive-services intervention [28]. Scale and rating criteria are detailed in Multimedia Appendix 2. The uptake of clinical preventive services was compared across different intervention strategies to identify potential patterns and insights. Data heterogeneity prevented meta-analysis.

The studies included in this review (n=23) represented 1,205,652 participants from 11 countries: the United States (n=11), Spain (n=3), Hong Kong (n=2), and 1 each from Australia, China, Denmark, France, Israel, Lebanon, and the United Kingdom. Studies were published between 2015 and 2025. A flowchart summarizing the search, screening, and study selection process is included in Figure 1 [29]. Designs included randomized controlled trials (n=20) and quasi-experimental studies (n=3). The majority (n=14) of included studies had a low risk of bias; the remaining had some or unknown risk of bias, but this was not a concern for the conclusions drawn by each study. The preventive services targeted in the interventions included colorectal cancer screening (n=10), influenza vaccination (n=6), breast cancer screening (n=3), pneumococcal vaccination (n=2), combined vaccinations (n=1), and COVID-19 vaccination (n=1; Table 1).

Digital tools used included the single or combined use of telephone calls, text messages, patient portal messages, email, and videoconferencing. No studies included a description of digital tool support or training for participants. Under half of the studies (n=9) included a description of the digital intervention’s reach (ie, messages opened and calls answered; Table 1).

The analysis revealed a wide range of approaches aimed at improving patient engagement and adherence to preventive health screenings and vaccinations. Three main categories of interventions were classified: digital and automated communication, personalized and tailored communication, and in-person interventions supplemented with digital supports.

Of the 23 studies, 9 incorporated theoretical frameworks into their intervention development or design. These studies applied a variety of behavioral theories that focused on intentions, beliefs, motivations, readiness, economics, stages of change, and osteopathic principles. The primary purpose of using theory in the design of interventions was to inform the content and framing of messages (Table 5). Among these studies, 6 of 9 reported significant increases in uptake in the intervention groups, with 3 demonstrating clinically important differences. In contrast, only 4 of 13 atheoretical studies reported a statistically significant increase in preventive care uptake.

Overall, the findings from this review demonstrate that digital interventions for promoting preventive screenings and vaccinations in older adults yield mixed results, with effectiveness varying widely depending on the intervention. While certain strategies showed substantial improvements in uptake, the majority of digital or automated reminders (eg, generic text messages and portal alerts) resulted in only marginal or no increases in screening and vaccination rates.Among the interventions reviewed, those incorporating personalized elements, such as tailored telephone counseling or supplementary in-person education, were consistently more effective than impersonal, automated communications. This suggests that while automation can facilitate uptake, human interaction remains critical for promoting preventive health behaviors in older adults.

Digital and automated communication tools, used in nearly half of the studies, can yield statistically significant improvement in uptake of preventive services among older adults. Older adults appear to respond well to generic approaches that notify and send reminders regarding their preventive activities and screenings. However, the effectiveness of simple, generic reminders is generally limited. These outcomes align with earlier systematic reviews and behavior change literature, which also report modest behavior change impacts from brief digital interventions, especially when applied generically to the older adult population [53].

Evidence from this work suggests that interventions incorporating personalized elements, such as tailored telephone counseling or supplementary in-person education, are generally more effective than impersonal, automated communications. This echoes findings from studies on preventive services and digital health adoption, which suggest that human interaction, particularly when delivered by a trusted source, such as a primary care provider, is a critical facilitator of successful behavioral change in older adults [54,55]. The preference for messages from known, trusted sources and for personalized content reflects broader trends in health communication and aligns with user preference studies, which show higher acceptability for content that has a connection to both their care and provider [56,57]. Further, studies have identified the personalization of messages to be an important condition for patient engagement [56,57].

Despite the importance of tailoring interventions to the user’s context and to engage them effectively [58], there are challenges related to personnel cost, time, and effort with this approach. However, emerging evidence suggests that artificial intelligence (AI) may offer scalable solutions by automating personalization and tailoring at a scale for low cost [59]. One opportunity, such as AI chatbots, is able to automate interactions while also simulating interactive and empathic dialogue [60,61]. Chatbots trained with motivational interviewing techniques have shown promise in supporting smoking cessation [60,62], addressing vaccine hesitancy [61], and promoting behavior change [63]. However, AI-driven personalization carries important limitations and risks: models can amplify existing biases and worsen inequities if trained on unrepresentative data, produce clinically inaccurate or outdated recommendations without continuous validation, and generate “black box” outputs that are difficult to interpret or explain [64]. Additional research is needed to evaluate the comparative effectiveness, acceptability, and ethical implications of AI-based personalization in diverse populations.

The appropriateness of digital health interventions for older adults is strongly influenced by technology complexity, digital literacy, and the availability of support [65]. While most studies included had low complexity, a consistent limitation across moderately complex digital interventions was the lack of attention to training needs, technical support, or onboarding required for older adults to effectively use digital health platforms such as patient portals, SMS text message reminders, or email notifications. The assumption of a baseline digital fluency may inadvertently perpetuate ageism and disadvantage those with weaker digital skills, compounding barriers to access and potentially widening health disparities [15]. Studies that evaluated engagement metrics frequently reported low rates of message opening or system use, with only about half of older adult recipients successfully engaging with portal messages or receiving SMS text message reminders, highlighting a significant constraint on the intervention’s reach. One recent survey has shown that while many older adults prefer patient portals and email, substantial proportions still rely on traditional methods (hospital websites and telephone calls), with significant numbers excluded by digital-only approaches, especially those aged 75+ years [55]. However, there is an increasing demand for patient portal communication in the population aged 65+ years [66]. This suggests that multimodal, accessible communication is critical for maximizing engagement in older populations.

Notable in the digital health intervention studies included in this review was the use of theory in intervention design. Approximately half of the included studies explicitly referenced theory, most commonly behavior change, primarily to inform the content and framing of message delivery. While theory use was described in 9 studies, the depth of integration in intervention design is not explicit in all study descriptions. This is similar to critiques raised in a related cardiovascular preventive care research study, where the rigor and depth of theory application were often unclear or inadequately reported, making it difficult to determine whether theory guided the overall intervention architecture or was limited to message wording [67]. This risk of “surface-level” use of theory, where theoretical constructs inform only minor aspects, such as phrasing or framing, rather than the full intervention pathways, has been documented as an ongoing issue in behavioral health research [68,69]. Fewer than half of published behavior change interventions explicitly cite a guiding theory, making it difficult to determine how theory actually contributed to intervention outcomes. The use of theory to guide the studies would have afforded greater explanatory value to the findings, in particular, offering insights into the largely mixed findings that were observed.

This review has several limitations that warrant consideration. The diversity of study designs, populations, and intervention types prevented meta-analysis, and findings were instead synthesized narratively. While this approach allowed for exploration of patterns across varied contexts, it limits the ability to quantify pooled effects. Data extraction was conducted by a single reviewer and verified by 2 others, which may introduce bias, as dual independent extraction is considered best practice [21]. Many studies lacked consistent control or comparison groups, and several used complex, multicomponent interventions, making it difficult to isolate the contribution of individual digital modalities. Although all included studies had a mean age of at least 60 years, many were not specifically designed for older adults and also targeted younger age groups. Finally, the review was limited to peer-reviewed, English-language publications, which may have excluded relevant findings from non-English–speaking regions.

While digital communication tools can support modest increases in preventive care uptake among older adults, their effectiveness appears to depend on the degree of personalization. This systematic review’s findings highlight that automation alone may be insufficient and that interventions facilitating patient engagement or combining digital outreach with practical supports may be more likely to achieve meaningful improvements in screening and vaccination rates.