Patients with unilateral or bilateral knee OA were included, with a diagnosis based on the physician’s assessment or radiography. Self-reported cases were excluded. There were no age or sex restrictions. Patients were included if their disease severity corresponded to grades I-IV of the Kellgren-Lawrence classification system.

Studies using apps with features that fit one or more of the following criteria were eligible for inclusion in this scoping review: (1) documenting or self-managing knee OA-related pain and other symptoms, (2) providing patient education, and (3) instructing or recording activities of daily living (such as exercise and diet). According to a previous study, self-management activities include maintaining good health and preventing adverse events, interacting with health care providers, improving self-monitoring, managing symptoms of knee OA, developing problem-solving skills, making decisions, using resources, forging partnerships with providers, and taking action [4]. Patient education was defined as content (videos and documents) that provided patients with knowledge on the pathogenesis of OA, treatment information, specific strategies to deal with pain, and appropriate exercise [16]. Studies on decision-making related to knee OA or assessing joint function were excluded. Additionally, studies were excluded if patients with diseases other than knee OA (such as hip OA) were included, as the results of knee OA could not be isolated from those of other diseases in such reports.

Many studies have investigated the effects of conservative management for knee OA using pain scales, functional assessments, and quality of life (QoL) measurements. In other words, the main goals for the management of knee OA should be pain relief, improvement in physical function, and enhancement of QoL [17]. Therefore, this scoping review summarizes the results of the included studies by using the 3 categories of pain, physical function, and QoL.

There were no restrictions based on region, race, or sex in the study selection. The search results were limited to papers published in peer-reviewed journals in English. Protocol papers, conference abstracts, interventional studies, and observational studies, including exploratory studies, were included. Systematic reviews or meta-analyses, case series, and case reports were excluded.

The following databases were used to conduct an electronic search: PubMed, Web of Science, the Cochrane Central Register of Controlled Trials, and CINAHL. A comprehensive search strategy for each of the 4 databases was developed using the words contained in the titles and abstracts of the relevant articles and the indexed terms from the reports (see Multimedia Appendix 1). The search period was from January 2007 (approximately the start of the smartphone era) to April 2021. The primary search was conducted in May 2021, followed by an updated electronic search and a manual search (mostly a citation search) in January 2022.

Citations were collated and uploaded to the Qatar Computing Research Institute, Ar Rayyan, Qatar [18], and duplicates were removed. Following a pilot test, 2 independent reviewers conducted a screening based on the eligibility criteria. This process was carried out in two stages: (1) during the first screening stage, titles, and abstracts were screened for inclusion or exclusion and (2) during the second screening stage, the full text was screened and evaluated. For studies excluded in the second screening stage, the reasons for exclusion were recorded. An independent third reviewer resolved any disagreements between the 2 reviewers. A PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) [19] format flow diagram shows the search results and study inclusion process (Figure 1).

The data extraction was performed by 2 independent reviewers using a spreadsheet. The extracted data included information on the first author, year of publication, country of origin, study design, population, sample size, intervention type, comparator, outcomes, time points of follow-up assessment, key findings of relevance to this scoping review, and the conclusion. Any discrepancies between the 2 reviewers were discussed and finalized by a third reviewer.

The outcomes identified in the literature were analyzed in 3 categories of pain, physical function, and QoL.

Articles published within the past 5 years were found to be relevant to this scoping review, suggesting that the majority of relevant literature is concentrated in recent years. Of the 8 studies included, 3 studies were RCTs [21,23,24], 1 was a conference abstract [20], and the remaining 4 were RCT protocols [22,25-27]. Using pain, physical function, and QoL as outcomes, mHealth was shown to be almost as effective as standard therapy in all RCTs. Effective mHealth interventions included exercise therapy, patient education, and dietary advice. The interventions varied in frequency, intensity, duration, and type, but most mHealth-enabled interventions improved associated outcomes effectively.

The effectiveness of mHealth for improving pain was examined in 5 studies, including the 3 RCTs and 1 conference abstract that examined the differences in effectiveness compared with a control group. In 1 RCT, there was no significant difference in pain scores between the 2 groups (mHealth versus conventional therapy). However, subgroup analysis in patients who actively used the mHealth app showed improvements in pain scores [23]. In the conference abstract, outcomes were compared between a home exercise training group that used mHealth and a brochure. Both groups showed significant improvement in the visual analog scale; however, there was no significant difference between the 2 groups [20]. Another RCT provided instructions on daily therapy without using mHealth in the control group. In this study, there was a greater improvement in WOMAC pain scores in the intervention group using mHealth [24]. However, the follow-up periods of the 2 included RCTs and 1 conference abstract were 1 month, 3 months, and 3 weeks, respectively; studies that examine outcomes for more extended periods are warranted.

Previous studies on mHealth with a patient, intervention, comparison, outcome (PICO) model, analogous to this review, have also reported improved pain outcomes for patients using mHealth compared with the control group [14,28,29]. Although we note that the patients and interventions differ slightly from those in our review, no significant differences in pain outcomes between mHealth and control groups have been reported [10,12]. Future systematic reviews should more precisely define their PICO models in order to deliver more objective assessments of efficacy.

All 8 studies examined the effectiveness of mHealth in improving physical function, and the 3 RCTs and 1 conference abstract examined the difference in effectiveness between the mHealth and control groups. In one RCT that compared mHealth with conventional therapy, there was no significant difference in KOOS-PS scores between the 2 groups. However, in a subgroup analysis of patients who actively used mHealth, there was a significant improvement in KOOS-PS scores [23]. In the conference abstract that compared home exercise training groups using a mobile phone and a brochure, there was a significant improvement in the Berg Balance Scale and WOMAC scores in both groups before and after the intervention; however, there was no significant difference between the 2 groups [20]. Another RCT reported significant improvements in the TUG test and the Katz Index of Independence in Activities of Daily Living in the group using mHealth [24]. The remaining RCT involved 3 treatment groups: diet, exercise, and a combination of diet and exercise. All 3 intervention strategies were associated with significant improvements in WOMAC scores and the TUG test [21]. However, the follow-up periods of the included 3 RCTs and 1 conference abstract were 1 month, 3 months, 12 weeks, and 3 weeks, respectively. Future studies should examine the long-term effectiveness of mHealth interventions in improving physical function. It should also be noted that one 3-arm RCT did not have a strict control group [21]. As such, the effectiveness of diet control and exercise therapy cannot be compared.

In another study, WOMAC scores improved after 24 weeks of mHealth intervention [28]. Another report on concomitant hyaluronate treatment showed an increase in walking speed and activity after 90 days of mHealth intervention [29]. Conversely, in a similar study on hip OA, the mHealth intervention group showed almost no improvement in physical function compared with that of the control group [10,12]. Due to the variety of outcomes associated with physical function, researchers should delineate outcomes carefully before conducting a systematic review. Through this review, we have identified KOOS-PS and WOMAC as the common measures used in the assessment of knee joint function. In studies with these outcomes, rather than simply assessing statistical significance, it is essential to consider whether there is an effect beyond the minimal clinically important difference.

The efficacy of mHealth in improving QoL was examined in 5 studies. One RCT and a conference abstract examined the difference in QoL between an mHealth group and a control group. The RCT compared mHealth with conventional care. There was no significant difference between the 2 groups in the results of the EQ-5D questionnaire. In a subgroup analysis of patients who actively used the app, there was also no improvement in the EQ-5D results [23]. The conference abstract compared a home exercise training group between a mobile phone and a brochure. Both groups showed significant improvements in the SF-36 questionnaire before and after the intervention; however, there was no significant difference between the 2 groups [20]. The follow-up periods for the included RCT and the conference abstract were 1 month and 3 weeks, respectively, so the effects of the intervention may have been temporary. As with the other outcomes described above, it would be appropriate to conduct future studies to examine the long-term effects of mHealth on QoL.

Although participants and interventions were not the same, other similar studies on mHealth have reported no significant differences in QoL outcomes between mHealth and control groups [12,30]. A systematic review with a specific PICO model should be used to determine the effectiveness of mHealth on QoL.

In recent years, the number of RCTs on mHealth in the management of knee OA has increased. Some protocol papers have also been published [31,32]. Although not included in this review, there are a number of other studies that have recruited participants through websites, and the presence or absence of knee OA was self-reported in the studies [12,33]. The recent data generated from the extant literature can guide the direction of future RCTs and systematic reviews.

There are 3 primary limitations to this study. First, the definition of mHealth as a form of medical intervention was not presented in detail. As a result, the scope of mHealth in the included studies was heterogeneous. In the future, mHealth interventions should be more rigorously defined. Second, the risk of bias and the quality of the reviews were not assessed. Although these evaluations are not essential in scoping reviews, readers should be aware of this limitation. Third, the studies did not consider the severity of knee OA in participants, and as a result, this aspect was not uniform in this review. By considering the severity of knee OA, it may be possible to examine the efficacy of studies in terms of population and heterogeneity.

Most of the outcomes included in this review were followed up only in the short to medium term. Long-term follow-up, such as up to 12 to 24 months, would help expand our findings with respect to the effectiveness of mHealth.

Studies on the effectiveness of mHealth in patients with knee OA are increasing. Our review suggests that mHealth is as effective as conventional therapy for pain, physical function, and QoL outcomes. Although the results of this review suggest that mHealth does not have a more significant effect on clinical outcomes than standard rehabilitation or conservative management, this finding is not necessarily negative. mHealth may still be more cost-effective, as it can be as effective as standard care without medical staff supervision or direct face-to-face instruction. In light of the importance of health care affordability, researchers should continue to include cost-effectiveness indicators in future study outcomes.