Factors Affecting Engagement in Web-Based Health Care Patient Information: Narrative Review of the Literature

Introduction

Background

In the United Kingdom, up to two-third use the internet to obtain health-related information at some point in their journey [1,2]. The internet has become an important source of education for patients, who are increasingly expected to, and are motivated to, play an active role in making decisions related to their health [3]. Patient education is defined as “the process by which health professionals and others impart information to patients that will alter their health behaviours or improve their health status” [4]. This may include information that is factual or related to patient experience, depending on the issue being addressed [5,6]. Reports suggest that 70% of the patients would like their physicians to recommend a source of web-based information relating to their medical condition, but only 4% of the patients receive such a recommendation [7].

Web-based patient health care information has several potential benefits, including convenient 24-hour access potentially wherever you are, ability to enhance knowledge acquisition [8,9], reduce anxiety [9], and improve the quality of conversations during health-related encounters. However, the effect of web-based content on patient empowerment, self-efficacy, and health attitudes has been found to be variable [9]. This may be due to the absence of an evidence-based framework outlining the factors that should be considered for improving engagement with web-based health care information.

Objectives

In this narrative review, we aim to identify and evaluate the factors that should be considered when producing engaging and high-quality web-based health care patient information. We also aim to incorporate these findings into a framework that may be useful as a guide to developing web-based health care information.

Methods

Overview

A comprehensive search of the PubMed and MEDLINE database was performed from January 1, 1946, to January 5, 2020. The Medical Subject Headings database was used to derive keywords and search term combinations, which included “patient information,” “online,” “internet,” “web,” and “content.” All studies pertaining to the factors affecting engagement in web-based health care patient information were included. No restrictions were set on the study type. Only studies in English were included. Analysis of themes arising from the results was performed using inductive content analysis. All retrieved abstracts and titles were reviewed by 2 independent investigators (EA and LAO) for relevance pertaining to engagement with web-based health care content. Disagreements between the reviewers were solved by consensus. Manual cross-checking of the reference lists of the identified papers was carried out to identify any other potentially relevant studies.

Analysis of themes arising from the results was performed using inductive content analysis [10]. This involved the reviewing of titles and abstracts by an author (UJ), with free generation of the categories relating to factors associated with high-quality and engaging web-based content. The categories were named using content characteristic words, and these were expanded into subcategories. The number of times a category was reported in the articles was totaled, and a list of categories was created in order of the frequency of mentions. All authors reviewed this list to decide which categories would be included in the final list.

Eligibility Criteria

As this is the first narrative review of its kind, no limitation was placed on study type or on surrogate measures of the outcome described.

Outcomes

All studies relevant to quality of engagement in web-based health care content were included.

Results

Overview

A total of 814 articles were identified, and of these, 108 (13.3%) were selected for full-text review based on their title and abstract. Full-text screening of the 108 articles resulted in the final selection of 56 (51.9%) articles, from which seven categories were derived. Figure 1 shows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart which depicts the stages of article selection.

‎View this figure

Description of Studies

Development of Categories

The studies were mostly observational and qualitative. A total of 3 randomized controlled trials and 5 systematic reviews were also identified. Categories relating to factors associated with high-quality and engaging web-based content were developed according to themes that were found to arise in the studies identified. The categories settled upon were as follows (with the number of papers reporting included in parentheses): (1) textual information (16 papers); (2) discussion boards or web-based groups (3 papers); (3) video content (11 papers); (4) visuals or pictographs (1 paper); (5) device accessibility (12 papers); (6) stage of patient journey (8 papers); and (7) credibility and completeness of information (4 papers; Table 1).

Textual Information

A study by Ernest et al [12] used the DISCERN Tool to evaluate the quality of written information regarding hormone treatment and surgery, which was found to be poor [66]. DISCERN is a validated tool developed by an expert panel through a process of panel debate and health care information analysis. It is noted to be the “first standardised index of quality of consumer health information” [66]. The DISCERN Tool comprises 15 key questions that investigate publication reliability and details of treatment choices, followed by overall judgment of quality. Each question is answered on a 5-point scale ranging from no to yes.

The study by Vivenkanatham et al [18] evaluated textual information on polymyalgia rheumatica. The study concluded that for web-based health care content to be effective, readability must be accessible to people of all literacy levels. Similar conclusions on readability were also reached in the study by Maciolek et al [20], which reported that patient education materials are most effective when simple language accessible to a wide patient population is used. A quality assessment study of web-based content on rheumatoid arthritis analyzed the readability, applicability, and accessibility of patient education websites [21]. The mean reading level was found to be 12.1 (SD 2.3), according to the Flesch-Kincaid Readability Tool. This tool gives a measure of how difficult a passage written in English is to understand through analyzing factors such as word length, sentence length, and total number of syllables. It provides a grade level according to the US educational system ranging from fifth grade to college graduate [67]. The same study reported that 78% of the websites assessed were easy to navigate [21]. Importantly, only 33% of the websites were assessed to be user friendly for people who are visually or hearing impaired.

An evaluation of web-based information on disorders of sex development noted strengths, including the tendency of webpages to present focused information in chunks and in a logical sequence [12]. A semistructured interview study of general practitioners conducted by Hjelmager et al [13] revealed that health information technology apps require textual information that is targeted to patients and written with the input of health care professionals.

The study by Rofaiel et al [14] assessed the quality of website information about inflammatory bowel disease using the DISCERN score to assess reliability and the relevance of pertinent details. The mean DISCERN score for patient-searched websites was not statistically different from that for physician-recommended websites (3.21 vs 3.62, respectively; P=.16). Numerous studies [12,13,15-17] identified that readability, as analyzed by the Flesch-Kincaid Tool, of web-based information exceeded the recommended sixth to eighth grade reading levels [68].

In all, 3 studies evaluated the understandability and actionability of web-based patient information by using the validated Patient Education Materials Assessment Tool [12,17,20]. The tool comprises inventories (one for print and another for audiovisual content) that list desirable and undesirable characteristics of information and produce a numeric value for understandability as well as actionability (ie, how easy it is to act on given information). The scores range from 0% to 100%, and a higher score indicates that the text is more understandable or actionable. Web-based educational content pertaining to disorders of sexual development and tympanostomy tube placement yielded low actionability scores (mean scores 23% and 44.6%, respectively). Interestingly, a study by Maciolek et al [20] found that the Patient Education Materials Assessment Tool understandability score for academic institution patient educational material exceeded that for content created by private institutions (P=.02) and content from websites unaffiliated with a urologist (P=.01).

Discussion Boards or Web-Based Groups

The study by Cedars et al [27] reported on a study of patients with urethral strictures who participate in web-based discussions and share experiences to gain emotional support and find answers. Patients participating in these web-based groups were more often than not satisfied with their postoperative outcomes. These findings are supported by the study by Teaford et al [28], which explored the experiences of new mothers using a web-based postpartum forum. The study found that web-based forums provided a sense of community and a platform for sharing information. They identified five themes pertaining to the participants’ concept of a web-based community: social support, anonymity, in-groups, drama, and entertainment.

However, the findings of a study by Castaneda et al [29] highlight the importance of exercising caution with web-based forums. The study evaluated the content of peripheral arterial disease eHealth forums and found that 15.1% of the medical advice given on such platforms was inconsistent with guidelines. Furthermore, the study found that 10 of the 18 posts related to negative personal experiences with health care providers.

It is important to note that there may be differences in patient engagement when comparing groups that are run by health care (or affiliated) professional service providers with those set up by patients or lay individuals themselves. A study comparing the difference between the effects of peer-led and moderated groups found that moderated groups were often more active and therefore had higher patient engagement.

Video-Based Content

The study by Pedersen et al [31] found that 4 participants who were interviewed after watching a 14-minute educational video on whiplash injuries felt reassured, particularly because the video aligned with information that they had received at the hospital. The study by Finnegan et al [32] reported that people who visited an information hub where videos were embedded in the webpage spent longer than 2 minutes on that webpage, indicating that visitors with average reading capability watch and listen to the video while browsing the text. The study also reported that this information hub’s YouTube channel, which featured videos explaining the concept of herd immunity, was particularly successful, with visitors spending more than 6 minutes on that page [32].

In the context of managing health conditions, the study by Button et al [33] found that video-based content was particularly helpful for patients having physiotherapy for knee injuries because they were able to visualize the correct technique. Patient understanding and confidence were found to be improved by this intervention. In a separate study of 80 anesthetists using a Video-Assisted Patient Education intervention, 96% reported that patients felt that they had a better understanding of the information provided through the Video-Assisted Patient Education intervention, and 97.5% of the anesthetists felt that it was a useful form of patient education [34]. However, 46% noted increased anxiety caused by the intervention [34].

Interestingly, the study by Ferhatoglu et al [36] found in an evaluation of the quality of YouTube content on sleeve gastrectomy that patient experience and advertisement videos were significantly more popular than academic videos created by medical professionals, according to the Video Power Index (P<.001 and P<.003, respectively). The Video Power Index assesses video performance by comparing the video with leaders in its respective industry, measures its impact and effectiveness across all platforms, and facilitates strategies to cater to target audiences [69]. The study by Erdem and Sisik [37] reported similar findings and found no significant association between video traction (likes, dislikes, or views) and usefulness of the content (Spearman rank correlation ρ=−0.038, P=.61; ρ=−0.003, P=.97; and ρ=−0.118, P=.12, respectively). Interestingly, the study found no significant correlation in usefulness to video length (ρ=−0.106; P=.16) in contrast to the findings of the study by Biggs et al [38], which concluded that medical videos categorized as useful had a mean length of 6 minutes and 51 seconds, with the videos rated in the top 10 having a mean length of 14 minutes and 47 seconds.

Similar to previous findings, a study conducted to assess the quality of YouTube videos on cataract surgery concluded that videos created by medical organizations such as the National Health Service were significantly more useful in terms of educating patients about the procedure than videos sourced by independent medical professionals and other sources (P<.001) [30]. The study by Bademci et al [40] similarly concluded that medical topic videos on deep vein thrombosis sourced from medical professionals and hospitals were significantly more useful than videos from other sources (P<.001).

In a study evaluating YouTube content on varicose veins, it was found that most of the videos were dominated by the private health care sector and that the video content presented a distorted view of treatment options, consequently leading to a skewed patient perception of the therapeutic options available to them [39]. Not only were 32% of these videos found to be of poor quality, but videos from private medical companies were also significantly more likely to favor minimally invasive surgery over ligation and stripping than videos from other sources. Once again, there was no significant association found between quality of content and viewing frequency.

Visuals or Pictographs

Pictographs are pictorial representations of words or phrases. The study by Christensen et al [41] described the value of pictographs in health communication. The study presents the results of building and testing the game Doodle Health, which is designed to produce pictographs through crowdsourcing. It found that this method of pictograph creation yielded positive feedback from focus groups with regard to usability and comprehension. Analysis of the feedback found that 62.2% of the participants praised the game, with a respondent describing it as “engaging and easy to use.” However, the study also found that people from diverse cultural backgrounds have different styles of communication, which may render visually presented information ineffective for minority groups. Crowdsourcing pictographs was suggested as a potential solution to this concern.

Device Accessibility

A study by Gogovor et al [42] concluded that the next generation of web-based educational health care programs should integrate apps for reasons of (1) accessibility, (2) flexibility, and (3) security and trustworthiness. Almost all the studies in our review used web-based platforms and required devices that accessed the web, with some studies requiring devices that accessed the video-sharing platform, YouTube, specifically.

A study by Lüchtenberg et al [43] found that only 18% (25/139) of the health information websites evaluated achieved a high standard of accessibility for users who are visually impaired as assessed by the Web Accessibility Initiative guidelines. The study concluded that web-based educational health care content should at least meet the requirements of priority 1 (level A) and preferably priority 2 (level AA) of the Web Accessibility Initiative guidelines. Developers can ensure accessibility by, for instance, having text alternatives for all nontext content and using high contrast ratios of text and images [36].

Few studies used smartphones exclusively to deliver information. Describing the benefits of using a smartphone, the study by Bashi et al [44] noted that smartphone adoption is becoming widespread, meaning more people can be reached; however, studies on how apps should deliver information in terms of interface and content is particularly lacking. The study by Noel et al [45] assessed the impact of a mobile medical app on plastic surgery patient care. A total of 60 patients answered questionnaires pertaining to the app, and the results supported the app’s utility from the patient’s perspective. The mean score for the app’s ability to answer patient questions was 4.1/5, and 20% of the patients were prevented from unnecessarily calling the emergency department. The use of a smartphone also enabled patients to access information from different locations and allowed for real-time disease management. A systematic review by Nicholas et al [46] revealed that of the 11 studies of apps providing patient information, only a third covered core psychoeducational principles. Furthermore, the average user ratings were not correlated with either comprehensiveness or adherence to best practice guidelines (ρ=−0.11, P=.80 and ρ=−0.02, P=.96, respectively).

Stage of Patient Journey

A qualitative study by Gogovor et al [42] using health care professionals and patient focus groups found that health care professionals suggested stratifying access to information according to stage of advancement in terms of their chronic pain management. A qualitative study of Oneself, a website designed for use by patients with chronic lower back pain, found it to be an effective adjunct to doctor-patient consultations because patients found that they were unable to have all their queries answered in the limited time of a consultation [49].

At different stages of a patient journey, decision aids may be used as adjunctive decision-making tools to support patients. In the study by Hagerman et al [50], the authors conducted semistructured interviews of 33 physicians to identify the desirable characteristics of decision aids. Of the 33 physicians, 20 (61%) stated that patients should be educated on the lack of urgency with regard to making a treatment decision. Of the 33 physicians, 28 (85%) agreed that decision aids should be provided to the patient after the consultation when the patient is at home. Furthermore, 36% (12/33) of the physicians deemed it very or extremely important that decision aids are designed to be used during and after consultations.

Tailoring information delivery to the stage of the patient journey is further supported by the results of the study by Kim et al [51] who developed a system to organize web-based disease-specific information according to a situational knowledge base model. The approach categorizes information about a specific disease (eg, thyroid cancer) into sections corresponding to discrete clinical events (eg, presentation, fine-needle aspiration biopsy, and diagnosis). In all, 75 patients completed a questionnaire evaluating the website, which found mean usability to be 4.6/5, personal relevance of received information 4.7/5, and comprehension of received information 4.8/5.

Credibility and Completeness of Information

A study by Jamison et al [47] found that 86% of the apps for pain conditions reported were created with no involvement by health care professionals. A comprehensive study by Bae et al [30], assessing the quality of the content of YouTube videos for cataract surgery patient education, found that there was an abundance of videos simply showing patients undergoing a live procedure. More than 20% of the educational videos were commercial and hence potentially misleading. This may make it challenging to find high-quality, comprehensive educational videos on the web.

A study by Pithadia et al [35] used the American Academy of Dermatology guidelines as a benchmark to evaluate the accuracy of patient information YouTube videos on psoriasis treatments. It concluded that 12% of the videos contained high-quality patient education content, and most of them were not patient-centric. Similarly, the study by Ferhatoglu et al [36] used the Journal of American Medical Association benchmark criteria to assess the educational quality of sleeve gastrectomy YouTube videos and found that this score was significantly higher in university-affiliated physician videos than in other videos (P<.001).

The Health on the Net Foundation Code of Conduct (HONcode) presents a set of eight principles designed to set the quality standard for web-based patient information [62] (Textbox 1). A study by Laversin et al [70] compared 165 HONcode-certified websites with 165 noncertified websites. Only 0.6% of the noncertified websites conformed to the principles of the HONcode compared with 89% of the certified sites (P<.10). As the study followed certified websites 6 months after certification, the effect of the HONcode certification shows short-term sustainability.

Discussion

Principal Findings

In today’s world of access to knowledge often being initially web-based, it is of importance for health care professionals to be able to create effective content. This is further emphasized in the current environment where minimum contact between patients and health care providers is required. We performed a wide scoping review of the literature to identify the features of web-based content and other telemedicine requirements that may improve quality of engagement with web-based health care content in this growing field. Using these results, we have developed a framework (Figure 2) to facilitate the development of web-based patient health care content. All the categories except for one (visual or pictograph) were reported on by 3 or more papers. We included the visual or pictograph category after discussion because it was felt to be a relevant and important means of communicating information. Although the features identified are, on reflection, intuitive, the framework arms the content creator with the best available strategies in making the content engaging and hence effective.

‎View this figure

Other similar frameworks such as the International Patient Decision Aid Standards (IPDAS) and the Standards for Universal Reporting of Patient Decision Aid Evaluation Studies (SUNDAE) checklists [65,71] have been developed for the evaluation of decision aids. However, to the best of our knowledge, this is the first framework to specifically focus on engagement with web-based content rather than a checklist approach to ensuring that decisions are made with appropriate consideration given to all relevant issues and options. A study design published by Knerr et al [72] aimed to evaluate patient behavior in response to a patient decision aid based on the IPDAS standards but has not reported results. Although patient decision aids can be a form of web-based content, efforts up to now have been directed toward ensuring transparency and trust in imparting information rather than ensuring the ability of web-based material to engage the user [73]. The need to engage people in health care content has been overlooked. We propose that this framework be used to improve engagement, which in turn will improve adherence with medical treatment and hence improve outcomes. Nevertheless, the rigorous process used by the IPDAS and SUNDAE developers is one that we would be interested in following in refining and adapting the framework arrived at through this literature review.

Further work will involve a co-design process with all stakeholders (including patients) to refine the insights we have gained from the studies regarding engagement with web-based content. The personalization of health care content may have bearing on the issue of engagement. Studies have shown that it is possible to segment the population according to the likelihood of responding to health care messages [74,75]. Although our framework provides the best evidence available relating to engagement with web-based content, the holy grail may lie in developing further the field of psychographics for health care. Although researchers have investigated the way that segmentation affects a defined intervention, the effect of segmentation itself is yet to be assessed in a meaningful manner [75]. The internet is able to bring together varied but related content using the concept of the semantic web and the application of folksonomies [76]. The confluence of Web 3.0 (to crowdsource content relevant to a desired health care behavior), psychographic segmentation (including segmentation based on the proposed framework), and machine learning may provide a way forward. We have developed a Web 3.0 health care content platform (Health Shared) and intend to use it for this purpose.

Limitations

The findings of this review should be considered in the context of several limitations. The principal limitation is that most of the components of this suggested framework are not supported by strong evidence. The studies were heterogeneous in their aims, interventions, and outcomes, and some were of poor methodological quality. Few studies discussed the effect of the platform used on patient engagement—for example, smartphone app versus website versus commercial health care information—which may play a role in patient engagement. However, given that most of the studies discussed in the review were general scans of available websites, the framework developed is largely applicable to the website development platform. In addition, few studies describe the differences between informational sites and other modes-of-engagement systems that provide the ability for patient input and enable patients to contact their provider or providers; therefore, we were not able to compare these patient information platforms.

Furthermore, because only the PubMed database was searched, studies are likely to have been omitted from this review. Despite this limitation, the components of the framework are intuitive, and we believe that its application may be beneficial to health care providers and content creators. Evaluation and subsequent validation of the proposed framework by key stakeholders, including patients, clinicians, and content creators, would increase the robustness.

Conclusions

There is a paucity of high-quality data relating to the factors that improve users’ quality of engagement with web-based health care content. Our framework summarizes the reported studies, which may be useful to health care content creators. Evaluation of the utility of web-based content to engage users is of significant importance and may be accessible through tools such as the Net Promoter score. Web 3.0 technology and development of the field of psychographics for health care offer further potential for development [75]. Future work may also involve improvement of the framework through a co-design process.

Although there are often specific health care issues needing to be addressed in response to crisis situations, we believe that this work is more generally important in facilitating patient activation and patient-supported self-management, which are two major pillars in how health care systems need to realign to keep up with increasing demand.