This review has been registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42023430716). Furthermore, the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines have been applied (Multimedia Appendix 1) [29].

Literature searches were conducted in 5 databases: Web of Science, Embase, Science Direct, Ovid, and PubMed. The selection of these databases was based on their provision of comprehensive access to full-text journals and conference proceedings pertaining to prominent conferences and meetings focused on digital technology and medicine.

To locate relevant articles, we conducted a search by filtering papers based on 3 primary categories of MeSH (Medical Subject Headings) terms: “Knee Osteoarthritis,” “Behavioral Therapies,” and “Digitization.” According to MeSH terminology, “Behavioral Therapy” is divided into “Behavioral Therapy,” “Cognitive Behavioral Therapy,” and “Dialectical Behavioral Therapy.” In this review, in order to understand the categorization of all behavioral therapies, we collected information on the subcategories of these 3 categories related to behavior. In relation to the subject of “digital” content, we gathered relevant material from the report by Safari et al [19] on digital literature, encompassing topics, such as “Telehealth,” “Email,” “Smartphone,” “Computer Systems,” “Digital Technologies,” and “Mobile Applications,” and other forms of digitization. Table 1 illustrates sample search strategies used for the PubMed digital library. The article titles, keywords, and abstracts were searched. Similar search strategies were applied to the remaining 3 databases. Relevant articles published between January 2013 and July 2023 were gathered. We included journal papers and peer-reviewed conference proceedings. Only articles written in English were considered.

The authors DZ and JZ were assisted in the literature search by an experienced librarian well versed in medical database searching. This literature review was guided by the question of how behavioral therapies can be integrated with digital applications in the rehabilitation of patients with KOA. On this basis, we anticipated that this review would (1) generalize and summarize the digital applications used in behavioral therapy, and (2) describe the overall research status and research trends of these digital applications.

Data relevant to the purpose of the study were extracted independently by the authors MW, WZ, and BC, and any misunderstandings and disagreements were resolved through negotiation. Extracted data included study context details, study population, and digital application details. In more detail, the template included the following categories: (1) basic information (author, year, origin, study population, sample size, presence of a physiotherapy intervention, and duration of the intervention); (2) digital application details (digitalization of behavioral therapy, interactive device function, study outcomes, and application deficiencies); and (3) type of study (randomized controlled trial, cohort experiment, experimental protocol, and qualitative study).

The Mixed Methods Assessment Tool (MMAT) was used to evaluate the methodological quality of the included studies [30]. This tool was initially created in 2006 through a comprehensive analysis of systematic evaluations that integrated qualitative and quantitative evidence. In 2018, a revised version of the MMAT was developed by assessing its usefulness, reviewing key assessment tools in the literature, and conducting a modified e-Delphi study involving methodology experts to determine the essential criteria (Multimedia Appendix 2). The MMAT evaluates the caliber of research employing qualitative, quantitative, and mixed approaches. The primary emphasis is on methodological standards, which encompass 5 fundamental quality criteria for 5 distinct study designs: (1) qualitative, (2) randomized controlled, (3) nonrandomized, (4) quantitative descriptive, and (5) mixed methods.

Figure 1 provides a summary of the outcomes at various phases of article selection. The search results of the databases are provided in Multimedia Appendix 3. Based on the search strategy, 2975 articles were found initially, and 2507 articles remained after title and abstract screening and removal of duplicates. Next, full-text articles were chosen based on the inclusion and exclusion criteria. We included a total of 131 articles explicitly related to digital behavioral therapy for KOA. Moreover, 4 articles were identified following a manual search of the references for articles that were cited. Final consideration was given to 36 articles for systematic evaluation (Table 2).

Among the included studies, 12 met 100% of the quality assessment criteria, 15 fulfilled 60%-80% of the quality assessment criteria, and 2 met 40% of the quality assessment criteria (Multimedia Appendix 4 [11,21,22,31-63]). The remaining 7 studies could not be evaluated for their quality owing to the absence of results. Nevertheless, the application description portion involved in the studies was highly valuable for analysis.

Although researchers and developers typically validate their digital applications with end users, it is uncommon for relevant studies to include end-user participation in the design phase. Of the 36 studies in our review, 8 (22%) reported end-user participation in the design phase [31,33,37,40,41,48,50,56], and 6 (75%) of these 8 studies also reported the participation of stakeholders other than end users [33,37,40,41,50,56]. These stakeholders also included caregivers, coaches, physiotherapists, and other individuals who provide care or services to the target population.

Heuristic assessments were used in a study to explore usability of the intervention for patients with KOA. Based on the outcomes of the interviews and heuristic evaluations, the program’s time structure was modified to be more flexible. In the most recent iteration, users had the option of repeating modules and adjusting module difficulty. The strategy also addressed improper website design and placement of multiple icons [31]. In addition, a study involved 3 patients with KOA who provided feedback on the prototype to inform the final design. Regarding how participants perceived interventions used outside the study setting, the majority suggested that health professionals, particularly general practitioners or physiotherapists, could deliver interventions to enhance or improve care [48].

Additional stakeholders were included in the qualitative study analysis. One study referred to experimental websites where patients with KOA and physiotherapists provided feedback on prototypes to inform the final design [40,41,50]. In a separate study, the app was also developed in collaboration with physiotherapists, physicians, and patient representatives. Named members of the project team submitted a list of 30 SMART (Specific, Measurable, Achievable, Relevant, and Time-bound) objectives related to OA treatment [41].

A total of 23 experimental studies, 7 experimental protocols, and 6 qualitative studies were included in this literature review. The primary outcomes of using digital behavioral therapy for rehabilitation of patients with KOA were (1) pain, (2) symptoms, (3) physical functioning, and (4) PA. Moreover, the secondary outcomes included (1) self-efficacy, (2) usability and user needs, (3) health-related quality of life, (4) satisfaction, (5) negative emotions, (6) quality of sleep, (7) adherence, (8) surgical intent, and (9) understanding of the condition (Table 5).

A total of 24 studies that aimed to reduce pain in patients with KOA were ultimately identified in this review [11,21,22,31,32,34,36,38,41,42,45,46,49-56,59-61,63]. Moreover, 12 studies reported statistically significant decreases in pain [11,21,31,32,36,38,41,42,49,59-61]. Of the 12 studies that did not demonstrate improvement, 6 involved randomized controlled trial protocols and 2 involved qualitative studies. In the studies that showed improvement, intervention durations ranged from 9 weeks to 9 months, and in those that did not show positive results, intervention durations ranged from 8 weeks to 6 months.

Overall, 21 studies assessed physical dysfunction [11,21,22,31,32,34,36,38,39,41,44-46,49,50,52-55,59,63], and of these, 6 reported statistically significant improvements [21,34,41,44,45,59]. Among the 15 studies that did not demonstrate improvement, 7 were randomized controlled trial protocols and 2 involved qualitative research. In studies that demonstrated improvement, intervention durations ranged from 3 weeks to 9 months, while in studies that demonstrated no improvement, intervention durations ranged from 4 weeks to 4 months.

A total of 12 studies measured PA outcomes [21,22,31,34,35,39,41,44,46,47,49-51,53-56,58], and 5 of them reported statistically significant improvements [34,41,47,49,60]. Among the 15 studies that did not demonstrate improvement, 6 were randomized controlled trial protocols and 2 involved qualitative research. In studies that demonstrated improvement, intervention durations ranged from 3 weeks to 9 months, while in studies that demonstrated no improvement, intervention durations ranged from 4 weeks to 12 months.

Physical symptoms were assessed in 11 studies [31,35,36,41,44,47,49-52,60], and of these, 4 studies reported statistically significant improvements [31,35,44,49]. In studies that demonstrated improvement, the intervention duration was 2 months, while in studies that demonstrated no improvement, intervention durations ranged from 2 to 9 months.

Self-efficacy was examined in 11 studies [11,21,22,32,36,46,52,55,58-60], and of these, 3 studies reported statistically significant improvements [11,36,58]. A total of 15 studies reported health-related quality of life [11,21,22,34,36,46,48-51,53-55,59,63], and of these, 3 studies reported statistically significant improvements [11,49,59]. Satisfaction was assessed in 8 studies [21,42,45,46,55,59,61,63], and of these, 1 study reported significant improvements [61]. Seven studies reported improvements in self-reported negative affect after the intervention [22,32,42,44,52,53,61]. Adherence was evaluated in 7 studies [22,36,42,43,49,55,58], and of these, 2 studies reported statistically significant improvements [22,58]. Sleep quality was assessed in 2 studies [42,61], and of these, 1 study [61] reported significant improvements. Three studies examined surgical intent [38,55,59]; however, none of the studies reported significant changes in patients’ surgical intent. Patients’ understanding of their condition was examined in 2 studies [38,51]; however, no improvement was identified.

The design recommendations of users and stakeholders for the app were reported at different stages of the final app study design. Two studies identified the negative emotions associated with the app [31,56]. For instance, Östlind et al [56] found that a wearable activity tracker facilitated PA in various ways and increased the awareness of the optimal number of steps to treat OA symptoms. However, not all participants found the wearable activity tracker to be motivating, and in some cases, if they missed a weekly PA, the app’s prompts about PA caused them to feel anxious and frustrated [31].

Two studies examined the efficacy of applying various characteristics. For instance, the study by Lindberg et al [50] demonstrated the combined efficacy of education, exercise therapy, and internet-based CBT, but was unable to distinguish between these interventions individually. Moreover, Dunphy et al [49] suggested that some participants viewed the physiotherapist’s participation as positive, customizing the digital program and monitoring their progress. Others described it as restrictive, particularly if the physiotherapist did not understand how the digital program operated.

In terms of how to promote PA, the studies suggested various potential solutions. For instance, setting and achieving a daily step objective would motivate individuals to walk more than usual [56]. In addition, there was a strong desire for support from health care professionals who could monitor and guide progress; reinforce health messages; and offer reassurance, motivation, and encouragement. In the interviews, peer support through online communities (eg, forums and blogs) was also mentioned as a positive feature where people could share their experiences and learn from and support others experiencing joint pain, although users felt that it needed to be supervised to prevent inaccurate and inappropriate posts [37].

This review presents an analysis of 36 papers that examined the use of digital behavioral therapies for patients with KOA. The analysis provides insights into 3 key areas: (1) the effectiveness of digital and behavioral therapies, (2) the role of digital technologies in these treatments, and (3) the importance of involving users and stakeholders throughout the design phase.

Several digital applications designed to provide behavioral therapy–based approaches for KOA were reviewed. Various theories of behavior modification and cognition were incorporated into these applications to increase their applicability and efficacy. In this review, we discuss behavioral therapy elements that are frequently combined to form applications.

Physiotherapists or applications, goal setting based on the user’s current physical status, and BCTs derived from a control theory framework can enhance long-term sustained exercise in patients with KOA [74]. Positive reinforcement of progressive PA is the most essential element of goals and planning [75]. Gradual increases in PA alter the perception that PA is associated with pain and increase confidence in improving PA performance, resulting in favorable physical (eg, physical ability, muscle strength, and joint flexibility) and psychological (eg, self-esteem, pain perception, and anxiety) changes. Repetition and substitution are intended to boost application engagement and health-related behaviors. A “severe” to “extremely severe” program and specific dosages are designed to stimulate strength gains, resulting in enhanced function. Physiotherapists assisted participants in devising a PA program intended to increase PA, with exercises targeting the hips, knees, and ankles, including sit-to-stand exercises and seated knee extensions. Our findings indicate that the aforementioned 2 BCTs were covered in virtually all of the applications we reviewed.

It has been demonstrated that reminders increase adherence to unsupervised home strengthening exercises [76]. This feature provides personalized behavior change messages to help patients with KOA surmount barriers to exercise participation. The physiotherapist or the application sends reminder messages based on completion to assist and remind the user to reach their exercise objectives. In the mobile app or wearable device, graphical displays were used to monitor workout adherence and provide feedback. These approaches provided the ability to register the completion of weekly workout sessions and send regular messages to encourage weekly workout participation. Despite the prevalence of wearable devices, their effectiveness in enhancing PA has been questioned. These devices frequently incorporate motivational techniques, such as self-monitoring and real-time feedback, but rarely address skills such as action planning and problem solving, which are essential for altering PA behaviors. Some studies also examined common CBT topics, such as catastrophizing, positive coping strategies, and fear avoidance [42]. Specifically, supervised exercise therapy through patient education enables users to access information regarding OA treatment modalities as well as topics such as the advantages of a healthy lifestyle, PA, vitality, and nutrition.

Our analysis revealed that OA digital management applications may be an alternative to traditional therapy and may further assist with the implementation of OA standards in the wider community [77,78]. However, the engagement of a physiotherapist is a vital aspect. Most participants had favorable experiences with their assigned physiotherapist and were motivated by the daily contact and the support and encouragement provided [79].

The use of digital technologies in communication by the patient care team can contribute to the enhancement of information flow, facilitation of patient information retention, improvement of information accessibility and portability, customization of information based on individual needs, and provision of tools for patients to actively participate in their health care [75,80]. In our analysis, the process of digitization manifested through a convergence of several technological mediums, including apps and websites, text messaging, phone calls, emails, and wearable gadgets.

New patient self-management programs have emerged in recent years, demonstrating the importance and efficacy of eHealth interventions such as websites and mobile apps. Current topics include PA, exercise, goal setting, action plans, pacing, medication management, nutrition, at-home exercise, pain comprehension, pain management, and relaxation. They may be combined with multimedia technologies to facilitate the sharing of content and with human-computer interface technologies to enhance accessibility.

Most health and medical apps fail to retain users beyond 90 days. This is due to their missing potential for facilitating disease management and provider-patient communication. The development of teleconferencing software or telematic messaging technology is supported by reminders and distinct objectives [81]. The development of text messaging program functionality and message libraries (including message types, message frequency, and program interaction levels) has the potential to increase adherence to unsupervised home exercise.

Wearable electronic devices aid in tracking the user’s daily activities and offer continuous visualization support. However, such devices typically carry a greater risk of privacy invasion and social stigmatization. Our findings indicate that none of the studies specifically addressed privacy concerns. Therefore, designers of pertinent systems should be encouraged to consider these factors more thoroughly. In addition, current wearable electronic devices collect limited data, and their primary function is still to provide feedback on physiological signals (steps, consumption, asymmetry, etc), lacking the accumulation of electrical signals unique to KOA. In the future, machine learning can be introduced into ubiquitous devices to predict patient activity in order to improve patient care.

In the reviewed studies, the combination of app/website plus SMS text message/telephone/email was the most common. It has been shown that text messaging programs combined with unsupervised web-based exercise can reduce pain and dysfunction in patients with KOA [22,40]. Although the addition of wearable electronic devices would improve the intervention process by providing more accurate monitoring data, the experience would not be enhanced. However, the cost of ubiquitous electronic devices and the complexity of their operation continue to prevent their widespread adoption.

Less than half of the reviewed studies reported structured user or stakeholder participation in the design phase of their systems, according to our analysis. Nonetheless, there is a distinct trend indicating that an increasing number of studies are emphasizing the significance of involving users in the design and development phases (and not just the validation or deployment phases). Specifically, we summarize a set of emerging user or stakeholder involvement design trends in emerging research on the application of digital behavioral therapy for KOA. Based on the interviews conducted for the qualitative study, issues were identified with the digital app user experience of KOA patients [52,56,62].

First, direct patient feedback was not taken into account when developing the content of the apps, and the content was based solely on the knowledge and experience of hospital staff. In future studies, patients should play a more important role in the development of the content of patient-specific apps, evaluating and further optimizing the preferable mode of information for outcomes [26]. Moreover, there is a need for a structured collaborative design involving patients, physicians, and researchers in order to establish multiple collaborative processes based on shared concepts, mutual learning, and respect for diversity and divergent opinions [82].

Second, digital support is regarded as an integral component of OA care. An ideal approach would be one that combines traditional OA care with digital OA care to offer solutions for personalized, comprehensive, straightforward, dependable, and continuous hybrid care [56,81]. The experience of digital applications is divided into 4 subcategories [79]: (1) simplicity of implementation, (2) flexibility in choosing time and location, (3) significance of interaction with health care professionals, and (4) additional motivating factors. Consequently, our research contends that digital applications must incorporate and differentiate between various user experience stages.

The functional requirements of apps that patients, physicians, and researchers deemed most essential, convenient, desirable, and actionable differed significantly. Participants in the studies agreed, despite their differences, that minimum viable products should be electronic, should monitor patients’ symptoms and activities, and should include features tailored to factors identified by patients and physicians as well as self-management strategies based on international guidelines. Over the course of the study, participants came to a consensus regarding the order of their functional requirements. Visual symptom mapping, goal setting, exercise programs, daily monitoring, and self-management strategies had the highest priorities.

After conducting a thorough study, a set of recommendations were summarized to improve the quality of applications. These recommendations are focused on making the process more efficient and improving future design and development endeavors. Suggestions for future studies are presented below.

This research has some limitations. Initially, our evaluation thoroughly examined the present state of digital behavioral therapy applications using qualitative analysis. However, it is indisputable that quantitative analysis yields more robust clinical value. Hence, a complete quantitative meta-analysis will be conducted to assess the efficacy of digital behavioral therapy applications. Future research will integrate both qualitative and quantitative analyses to provide a more thorough evaluation. In addition, our search was limited to English-language research publications, and it is possible that there are significant findings in other languages. The search for research papers was restricted to 5 databases (Web of Science, ScienceDirect, PubMed, Ovid, and Embase). As algorithms are added to digital intervention tools, additional computer science databases, such as IEEE and ACM, could be added to increase the comprehensiveness of the assessment. Furthermore, this review was limited by the search criteria employed and the time period during which the papers were published. However, a focus on the last 10 years went a long way in ensuring that this systematic review includes the most recent research.

This review provides an overview of digital behavioral therapy applications for patients with KOA. In this systematic review, 36 studies were examined. The results demonstrate 14 BCTs and show that behavioral cognitive therapies are frequently combined when developing digital applications. The most prevalent areas were “goals and planning” and “repetition and substitution,” which were frequently used to develop PA goals and PA behavior adherence. The most prevalent combination strategy was app/website plus SMS text message/telephone/email, which has tremendous potential. Consequently, this research provided results for digital applications in terms of pain relief, physical function improvement, self-confidence, and improvement in health-related quality of life among patients with KOA. Nevertheless, there was a shortage of evidence indicating enhanced surgical intention, compliance, and disease knowledge. Subsequent quantitative analysis of this phenomenon is required in the future. Moreover, the incorporation of several stakeholders in the design and development stages might enhance the user experience, considering the distinct variations in their requirements. Based on the findings, digital applications should incorporate various stages of user experience and should include a combination of traditional and digital solutions for OA care.