Patients’ chronic illness management skills, including medication use, symptom monitoring, and other health behaviors, are a vital step in achieving good health outcomes [1-3]. Effective self-management of chronic illnesses relies on several factors, including possessing knowledge about one’s condition, believing that one’s behaviors affect one’s health, and having the ability to use and access to necessary resources to execute recommended health behaviors [4]. Theories guiding patient education include adult learning theory, which emphasizes self-directedness in learning, the incorporation of prior knowledge, and problem-centered approaches [5]. In addition, self-efficacy theory emphasizes patients’ belief in their power to manage their disease [6]. Patient education can directly affect knowledge and self-efficacy [7,8]; for example, comprehensive in-person patient education has been shown to improve quality of life (QOL), reduce hospitalization, and reduce readmission among patients with heart failure [9]. In addition to improving patient health outcomes, adequate chronic illness management can reduce health care costs; for example, an estimated US $100 billion of health care costs in the United States could have been avoided in 1 year through improved medication adherence [10].

Although patient education is often highly valued by health care professionals, the lack of time and insufficient staffing are cited as barriers to providing optimal in-person patient education, if it is provided at all [11-14]. The average primary care visit involves only 16 minutes of face-to-face time between clinician and patient [15]. Because of these constraints, patient education interventions that can be used outside of the clinic, including video-based resources, SMS text message reminders, and interactive modules have been developed to replace or augment in-person patient counseling. Digital therapeutics (eg, mobile apps) have been shown to improve outcomes in diabetes and smoking cessation [16-18]. Technology-based resources may be less effective for some patients, such as those without access to technology devices or the internet or those with low eHealth literacy but could be tailored to individual factors such as health literacy, eHealth literacy, and self-efficacy [19-22].

In addition to the growing interest in technology-based interventions more broadly, there is increased pressure in the wake of the COVID-19 pandemic to develop and implement virtual care and education resources for patients because clinics halted in-person visits during the pandemic [23,24]. Stay-at-home orders and physical distancing led to virtual patient programming by primary care clinics, geriatricians, and pharmacists [25-27]. This drastic change in health care delivery yielded substantial literature studying its effects on patient satisfaction, behaviors, and health outcomes [28]. Studies of digital patient education tools for patients with cancer and fertility issues found that such tools resulted in increased health knowledge but were not sufficient in terms of addressing patients’ emotional needs and were associated with patient stress and disappointment [29,30]. As the COVID-19 pandemic evolves, provider time constraints persist, and technology-based interventions continue to expand more broadly, finding ways to optimize patient education virtually will continue to be a priority.

Digital interventions are tools that use technology to improve patients’ health and include apps, games, modules, and videos. Video-based educational interventions, which are a form of digital intervention, not only use educational principles such as adult learning theory and self-efficacy theory but also navigate challenges of disparities in technology access and eHealth literacy. The efficacy of digital interventions, including video-based education tools, has been studied for specific chronic illnesses, but, to our knowledge, a systematic review of the efficacy of video interventions across major chronic illnesses has not been performed. In addition, studies of the qualities that make educational content most effective, including setting, frequency, and intervention components, have been inconclusive [31]. A more complete understanding of the efficacy of video-based education tools for chronic illnesses can meaningfully guide the development of new patient education interventions. Therefore, this study aimed to systematically review the literature to determine whether patient-targeted video-based educational self-management tools for chronic illnesses are effective interventions to improve self-management and health outcomes.

We conducted a search of the English literature in consultation with a medical librarian (DW). Our initial search was conducted in MEDLINE using Medical Subject Headings (MeSH) terms related to chronic illness combined with terms to identify intervention studies, which included video technology, self-management, and hospital discharge, ultimately yielding the final search terms (Multimedia Appendix 1). We conducted parallel searches in PubMed and CINAHL. Searches were not restricted to specific publication years and included studies published through October 2021.

The inclusion criteria were as follows: (1) adult patient population (aged ≥18 years); (2) intervention studies of video-based self-management patient education; and (3) one of the following chronic health conditions: asthma, chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), chronic pain syndromes, diabetes, heart failure, HIV infection, hypertension, inflammatory bowel disease (IBD), and rheumatologic disorders. Articles with diverse digital interventions were included so long as at least 1 component was video based. Video-based self-management patient education interventions encompassed videos that aimed to educate patients regarding their disease or its management. Articles were excluded if they focused on chronic conditions other than those outlined in the inclusion criteria (eg, malignancy, other infectious diseases, injury or trauma, surgery, intellectual disabilities, prediabetes, obesity, ostomy care, and chronic psychiatric conditions), had pediatric or clinician study populations, were noninterventional studies, or involved interventions that were not video based or not designed for patients.

The articles identified from all electronic databases were evaluated for duplicates to ensure that each was a unique manuscript. Next, title and abstract reviews were conducted to evaluate for inclusion. Papers not clearly excluded through title and abstract review underwent article extraction review. Of the 112 papers, 10 (9%) were reviewed initially by the study team to ensure consensus and validate the extraction form. Data from the articles were extracted into a standardized REDCap (Research Electronic Data Capture; Vanderbilt University) form by an author (VP, CK, MW, AV, and ND), and 19.6% (22/112) of the articles underwent a second independent extraction by a different author (VP, CK, MW, and AV) for ongoing quality assurance. The standardized form prompted extraction of the following elements: study characteristics (location, enrollment, and size), study design (randomized controlled trial [RCT] vs pre-post study and intervention components), sample demographics, and results.

Articles were qualitatively evaluated to determine heterogeneity across key extraction components, including study characteristics, study design, sample demographics, and results. After the qualitative evaluation, data were quantified and summarized across these same components. Data visualization and descriptive statistics were generated using Stata SE (version 15.1; StataCorp LLC). Meta-analysis was not possible because the program outcomes were too heterogeneous. Data reporting was conducted per the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) checklist (Figure 1 [32]).

Risk of bias was assessed by applying the revised Cochrane risk-of-bias tool for randomized trials (RoB 2) for the included RCTs and the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I; The Cochrane Collaboration) tool for the included pre-post studies [33,34]. The RoB 2 assessed multiple domains, including risk of bias from the randomization process, deviations from the intended intervention, missing data, measurement of outcome, and selection of reported results [33]. The ROBINS-I tool also assessed multiple domains, including risk of bias from confounding, selection of study participants, classification of the intervention, deviations from intended interventions, missing data, outcome measurement, and selection of reported results [34].

Of the 785 studies that underwent title and abstract review, 112 (14.3%) were eligible for full article extraction (Figure 1). Of the 112 articles extracted, 59 (52.7%) were eligible for inclusion in this review (Table 1). Studies were excluded if they did not include an intervention (27/112, 24.1%), were pilot or qualitative studies (15/112, 13.4%), did not include eligible conditions (13/112, 11.6%), or did not use a video-based self-management intervention (3/112, 2.7%).

The included studies’ publication years ranged from 1979 to 2021. Two-thirds (40/59, 68%) of the studies were conducted in North American or European countries, and 31% (18/59) took place in an urban setting. A majority of the studies (39/59, 66%) were superiority RCTs, 22% (13/59) were pre-post studies, and 12% (7/59) were noninferiority RCTs. The most commonly represented chronic diseases included diabetes (11/59, 19%) [36,39,49-52,59,77,79,81,87], asthma (11/59, 19%) [43,44,46,55,64,69,80,84,89,93], COPD (8/59, 14%) [38,61,66,68-70,75,89], and heart failure (9/59, 15%) [35,38,41,45,60,71,76,83,90]. Less commonly represented diseases included CKD (7/59, 12%), rheumatologic diseases (5/59, 8%), hypertension (5/59, 8%), chronic pain syndromes (4/59, 7%), HIV infection (2/59, 3%), and IBD (1/59, 2%).

Of the 59 studies, 22 (37%) explicitly targeted a particular group with the intervention, such as older adults, veterans, and Black patients. Among the studies that reported overall sample demographics (30/59, 51%), the mean age was 53.9 (SD 10.6) years, 52.5% of the participants were male, and 47.5% were female. The majority of the studies (45/59, 76%) did not include race data for the overall sample. Approximately one-third (20/59, 34%) of the studies explicitly stated a conceptual framework for the interventions (Table 2), including adult learning theory [41,67,69,83,87,89,91], goal management theory [51,62,77], and health literacy perspective [64,93]. In the included studies, the median number of participants enrolled was 130 (IQR 65-199), and the median number of participants included in analysis was 119 (IQR 60-167). Only 3 (5%) of the 59 studies reported a sample mainly composed of individuals with low socioeconomic status [54,84,92]. A little more than half (32/59, 54%) of the interventions took place in the outpatient setting, 40% (23/59) took place at home, 19% (11/59) in the inpatient setting, and 3% (2/59) in the emergency department (ED).

All interventions included an educational video, and many of the interventions included printed materials (24/59, 41%), in-person counseling (19/59, 32%), and interactive modules (12/59, 20%; Table 3).

The plurality of video-based interventions (28/59, 47%) occurred once (Table 4). A quarter (16/59, 27%) of the interventions took place at the patients’ discretion (Table 4).

Half (132/262, 50.4%) of the metrics were outcome measures and half (130/262, 49.6%) were process measures. Outcome measures included disease severity, health care use, QOL, symptoms, and cost. Process measures included patient knowledge, health behavior, self-efficacy, and medication use. A meta-analysis was not performed because the interventions were too heterogeneous, varying in setting, frequency, and design. Most of the interventions (53/59, 90%) had at least 1 successful or noninferior component. Studies that explicitly stated conceptual frameworks did not have more substantial improved outcomes.

Of the 59 studies, 7 (12%) studied noninferiority of video-based tools compared with in-person education [38,42,51,65,67,69,70]. Among these papers, 82% (28/34) of the outcomes resulting from video-based tools were noninferior, 6% (2/34) resulted in inferior outcomes, and 12% (4/34) resulted in mixed or unclear effects.

A little more than half (31/59, 53%) of the studies involved repetition of the intervention. Outcome evaluation ranged from immediately after the intervention to 16 months after the intervention. The majority of the papers (28/31, 90%) assessed outcomes that occurred on a short-term basis (within 1 year). Outcomes resulting from repetition of the intervention were more frequently improved than those resulting from the intervention completed once (49% vs 40%, respectively; P=.23).

The RoB 2 (Multimedia Appendix 2 [35-80]) was used to evaluate the RCTs (n=46). Of the 46 RCTs, 24 (57%) had low risk of overall bias, 17 (37%) had some concerns for risk of overall bias, and 5 (14%) had high risk of overall bias. Of the 46 RCTs, 15 (33%) carried risk of bias owing to the lack of blinding of outcomes assessors, 9 (20%) had risk for bias owing to problems with randomization, and 2 (4%) carried risk of bias owing to loss of a substantial number of study participants over the study period. The ROBINS-I tool (Multimedia Appendix 3 [81-93]) was used to evaluate the pre-post trials (n=13). Of these 13 trials, 6 (46%) were judged to have low risk of bias and 7 (54%) had moderate risk of bias. Risk of bias occurred because of risk of confounding, missing data, and outcome assessors’ awareness of participant group assignment.

A systematic review of the efficacy of video-based education across major chronic illnesses has not previously been performed. This study confirms prior published evidence that video-based tools improve patient knowledge in the short term. We extend past findings by demonstrating that there is less consistent evidence for the effectiveness of these tools in improving patients’ QOL and disease severity [96,97]. Furthermore, our findings confirm a prior study that demonstrated a lack of uniform evidence of significant improvement of health behaviors from video-based educational interventions [98]. Prior reviews focused on comparing video-based interventions with either in-person counseling or printed materials. Our study demonstrates how health interventions have become more complex, involving several interventions alongside videos, including printed materials, interactive modules, and telephone or SMS text message communication.

Video-based interventions were most successful in improving knowledge measures; for instance, the majority of patient knowledge outcomes (eg, Dutch Heart Failure Knowledge Scale score, diabetic foot care knowledge, knowledge of asthma triggers, and inhaler technique knowledge) [44,52,60,62,69,71,74,77,78,80,83-85,87,91-93] showed significant improvements after the video-based intervention. This is consistent with prior literature [96]. Furthermore, a little more than half of the self-efficacy outcomes were improved by video-based interventions. Examples of successful self-efficacy outcomes included pain self-efficacy scales, asthma attack prevention self-efficacy surveys, inhaler technique confidence, and heart failure self-care confidence scales [43,49,73,83,85,87-89,91]. Some of the studies explicitly stated self-efficacy theory as a framework in tool development, whereas many of the others included patient empowerment language and disease self-management techniques in their tools [36,43,86,88]. These findings suggest that video-based education tools with goals of increasing patient knowledge are more likely to be successful. This systematic review showed a trend toward increased success in video-based educational tools with repeated exposure to the intervention, but these results were not statistically significant.

Ideally, increases in patients’ knowledge and self-efficacy lead to increased engagement in positive health behaviors and appropriate medication use. A little more than half of the health behavior outcomes, such as exercise, diet, and health self-monitoring, were improved by video-based intervention. Examples of successful health behavior outcomes included diabetes foot care behavior; heart failure management behaviors such as daily weight, fluid, and salt monitoring; physical activity; caloric fat intake reduction; and blood glucose self-monitoring [35,40,43,47,50,59,72,73,76,78,81,83,90]. A third of the included papers measured at least 1 health behavior result, indicating that there is active interest in how video-based interventions modify behaviors. Our review demonstrated that only half of the medication use outcomes were improved by video-based intervention. The majority of successful medication use outcomes were inhaler use and technique [43,51,68,75,89]. Most of the unsuccessful medication use outcomes were for pain medication use measures in IBD and chronic pain syndromes [37,57,74]. Although it is promising that some successful health behavior and medication use outcomes were achieved through video-based education tools, these findings suggest that these tools alone may be insufficient to consistently improve these outcomes in practice.

It is not surprising that educational interventions more consistently improve patient knowledge rather than behaviors because several factors beyond education may influence a patient’s ability to engage in health behaviors or adhere to medications. Examples include community resources, socioeconomic status, social support, and stress. These findings suggest that video-based educational tools have the potential to improve health behaviors and medication use but may need to be augmented with additional support to consistently make positive health behavior changes accessible to patients.

Ultimately, the interventions aimed to increase patient engagement in positive health behaviors, and we hypothesized that this engagement would result in reduced disease severity, fewer symptoms, improved QOL, and decreased inappropriate or cost-ineffective health care use. Increased patient engagement has been shown to improve health-related outcomes, including clinical indicators and avoidance of costly use [20,99]. Our systematic review demonstrated that video-based education tools less consistently improved disease severity and cost, extending the findings of a prior systematic review [98]. Disease severity was the most frequently reported outcome, but only one-third of the disease severity outcomes were improved by video-based interventions, predominantly in COPD, diabetes, and hypertension. Ankylosing spondylitis and other rheumatologic function scores and temporomandibular joint disability scores were not improved with video-based interventions. Health care use improvement was defined as increased adherence to scheduled follow-up visits and decreases in unplanned hospital readmissions and ED use. Among the included papers, a minority of health care use outcomes showed significant improvement. More than three-quarters of the QOL outcomes, such as depression and anxiety scales, self-rated QOL scales, and global well-being scores, were not substantial improved by video-based tools. These findings suggest that video-based tools do not consistently improve patients’ disease severity, health care use, and QOL or, alternatively, that interventions and study follow-up must extend over longer periods of time to measure changes in these outcomes.

Of all the measures of interest, cost was the most underrepresented. Only 2 (3%) of the 59 papers reported the effect of video-based interventions on cost measures at all. This may be due to a requirement of longer follow-ups, larger sample sizes, and more complex data collection required to demonstrate cost effects. Cost measures may be indirectly measured through health care use variables such as adherence to preventive and scheduled health care encounters and rates of unscheduled ED visits and hospitalizations. One opportunity for further study includes continuing data collection through longer follow-up periods to clarify the effects of these interventions, if any, on cost.

There is great diversity in both the chronic health conditions studied in the literature and the specific educational programming offered. Often, interventions combined videos with interactive modules, SMS text message reminders, and telephone counseling. These combinations of interventions may simulate the diverse sources of information that patients are exposed to in practice, but they complicate analyses to determine what programming components specifically contribute to an improvement in patients’ knowledge, confidence, behaviors, and outcomes. Virtual platforms and technologies continue to be leveraged to augment traditional health care delivery and patient education tools [100-104]. Although only a few of the papers studied noninferiority of video-based education tools to in-person education, the majority of these papers found at least 1 noninferior outcome, indicating that video-based tools are promising compared with in-person strategies. Of the 59 included studies, only 1 (2%) directly compared video-based intervention with other virtual education interventions, comparing video-based education with SMS text message communications, and found improved outcomes with texting communications [67]. Further clarification of the relative efficacies of these tools is important as televisits, video visits, and patient portals become more prevalent. Furthermore, assessment of new technology must take into consideration variations in access to, as well as knowledge of, and dexterity with, technology in patient populations [22,105,106].

A minority of the included studies (20/59, 34%) explicitly stated the conceptual frameworks underlying their video-based educational tool. This limited the analysis of which conceptual frameworks, including adult learning theory, goal management theory, health literacy perspective, self-efficacy theory, and community-based participatory research, yielded more successful outcomes. Further study is needed to clarify which educational theories and specific components make video-based patient education tools most successful.

Although this study used rigorous methods to review the existing literature, there are important limitations. First, the heterogeneity of the factors reviewed, including large number of chronic illnesses, specific programming components, timing, and intervention goals, makes drawing specific conclusions about the efficacy of video-based tools difficult and precludes execution of a meta-analysis. In addition, enrollment in studies and engagement with video-based health tools in trials may not reflect the real-world behaviors of the population at large. Patients who enroll in research studies tend to have higher health literacy, and the effect of varying health literacy on video- and technology-based health tools will need to be assessed to ensure equity as the use of these tools continues [107-109]. A minority of the studies targeted specific populations, such as those with lower health literacy, and none targeted patients with low eHealth literacy. Strategies of studies targeting specific populations with low health literacy included nontechnical language, narrative testimonials, and inclusion of target populations in the development of the educational video-based tools [49,54,68]. Studying these populations is vital in understanding the true efficacy of video-based tools for all patients [22].

This study suggests that for many process measures and some outcome measures, video-based chronic illness management tools are effective in improving patient outcomes. In some of the cases, video-based tools were found to be superior to printed materials and noninferior to in-person counseling. From time constraints with regard to clinic visits to changing policies on social distancing as the COVID-19 pandemic evolves, a need for alternatives and adjuncts to in-person counseling is clear. Patient-facing video-based tools for chronic illness management show promise. Additional studies that assess disease severity, symptoms, and cost measures over longer follow-up periods and specifically targeting patients along the spectrum of digital competencies are needed to understand how to optimize patients’ benefit from video-based education tools.