Gamification in knee osteoarthritis rehabilitation: A systematic review of clinical evidence

Eligibility criteria

For inclusion, peer-reviewed English-language studies, randomized clinical and controlled trials with participants aged 40 years and above, physician-based diagnosis or with K-L grade 1 and above, or American College of Rheumatology criteria evaluating the effects of digital interventions (Gamified exercise, exergaming, VR-mediated rehabilitation, AVG, sensor-augmented feedback, wearable-supported programs, or digital-coaching interventions) for at least 3 weeks on KOA symptoms, focusing on pain and functional disability. Studies that concentrated on non-OA circumstances, non-balance therapies, secondary studies (reviews), gray literature, case reports and series, qualitative studies, opinion pieces, or animal research were excluded.

Literature search

Articles were searched from PubMed, Embase, and PEDro by searching for terms “Gamification” OR “Virtual reality” OR “Gamified Exercise” OR “Digital gaming” OR “Virtual Exercise” OR “Game-based training” OR “Digital intervention” OR “Digital health” OR “Exergaming” AND “Knee Osteoarthritis” OR “KOA” OR “Knee OA.” This article employs a randomized controlled trial (RCT) and a clinical trial (CT) study design. The filters applied to the search are detailed below: Results from 2015 to 2025, English-language articles, human species, and both sexes [Supplementary material 1].

• P: Population – KOA

• I: Intervention – Gamified exercise, exergaming, VR-mediated rehabilitation, AVG, sensor-augmented feedback, wearable-supported programs, or digital-coaching interventions

• C: Comparator – Conventional group, usual care, or no therapy

• O: Outcome – Visual Analog Scale (VAS), Western Ontario and McMaster Universities OA Index (WOMAC), Knee Injury and OA outcome score (KOOS)

• S: Study Design: Experimental studies – Clinical and controlled trial

A reference list screening of the full-text articles was conducted to ensure that no relevant articles were missed.

Study selection and data extraction

Based on the search strategy employed, duplicates across databases were manually removed using the Rayyan software. Reviewers independently screened studies first by title and abstract, discussing any disagreements together until consensus was reached. Full-text articles were then evaluated for inclusion, with reasons for exclusion noted.^[11] Two independent reviewers evaluated all included studies; any conflicts that occurred were resolved through discussion.

The reviewers independently extracted the data using a standardized Excel template to reduce bias. Data collected included study title, design, authors, participant details, sample size, intervention/comparator specifics, outcomes, duration, and key results.

Assessment of quality of methods: PEDro scale

The PEDro scale provides a validated, widely accepted method for evaluating the quality of CT methods. It includes 11 criteria that cover external validity (item 1), internal validity (items 2–9), and statistical reporting (items 10–11). Each item receives a rating of “Yes (Y),” “No (N),” or “Uncertain (?).” Trials are then classified by total score calculated manually: ≤3 indicates poor quality, 4–5 suggests fair quality, 6–8 denotes good quality, and 9–10 reflects excellent quality, with Item 1 not included in total scoring.^[12]

Risk of bias (ROB) assessment: ROB 2.0 tool

RCTs were evaluated using the RoB 2 tool, which assesses key sources of bias across domains such as randomization procedures, deviations from planned interventions, missing outcome data, outcome measurement, and selection of reported results. Response options include “Yes,” “Probably yes,” “Probably no,” “No,” or “No information.” Each domain receives a rating of low ROB, some concerns, or high ROB, with an overall judgment derived from these individual assessments.^[13,14]

Study selection

Based on the database search, 943 relevant records were selected. After removing 117 duplicates, 819 records were screened based on their abstracts and titles. Records were excluded when they were unrelated to the research question or were of a different study type. Finally, 14 full-text studies among the eligible records were assessed for eligibility based on inclusion and exclusion criteria. Of 14 studies, 6 met the eligibility criteria and were included in the systematic review [Figure 1].

Close

Figure 1: Flowchart.

Export to PPT

Characteristics of the study

The six included trials ranged in sample size from 42 to 80 participants (median n = 60) and publication year from 2015 to 2025. Three studies used VR or exergaming platforms.^[15-17] One compared AVG to traditional exercise.^[18] Two employed wearable and digital-coaching approaches.^[19,20] All were conducted in clinic-based or hybrid clinic-home settings. Intervention durations ranged from 10 days to 12 weeks; follow-up assessments extended from immediate post-intervention to 12 weeks. Most studies included KOA with Lawrence grades II–III, with only one study that included grade I as well [Supplementary material 2].

Assessment of quality of methods: PEDro scale

PEDro scores ranged from 6 to 8 out of 10. Common limitations were allocation concealment not reported (n = 4 studies), blinding of participants and therapists not reported or not possible (n = 6 studies), intention-to-treat analysis not clearly described (n = 3 studies), and incomplete follow-up in some studies. Two studies received a score of 8/10 (high-quality methods). Overall, the included trials demonstrated good methods’ objectivity, though several lacked robust reporting of blinding and allocation concealment [Table 1].

Risk of bias

Figure 2 presents the ROB assessment of five method domains (D1–D5) for the included studies. Overall, studies demonstrated low bias in the domains of randomization (D1), outcome measurement (D4), and missing outcome data (D3). Although concerns about bias due to deviations from intended interventions (D2) and selective reporting (D5) were frequent, several studies were rated as having some concerns or high risk, which negatively influenced their overall judgment. Two studies reported a high ROB in the selection of reported results (D5).^[16,17] But one study, among the six, achieved low scores across all domains.^[20] Taken together, these findings suggest that while the overall quality of the methods of the included evidence is moderate, caution is warranted when interpreting results from studies with identified high-risk domains, especially those related to intervention adherence and outcome completeness.

Close

Figure 2: Risk of bias plot.

Export to PPT

Pain outcomes

Two included trials assessed pain using a VAS.^[15-17] One of the included studies specifically assessed pain using the KOOS pain subscale.^[19] In another study, pain assessment was performed using the WOMAC pain subscale.^[18,20]

Disability outcomes

All 6 studies assessed disability using WOMAC.^[15-18,20] One study assessed disability using KOOS.^[19]

Technology-enhanced programs

The present review discusses the KOA rehabilitation delivered using (a) Telerehabilitation mode, (b) VR mode, (c) AVG with exergaming mode, and (d) wearable-based technology mode.

Plavoukou et al.^[20] evaluated the effectiveness of a sensor-augmented telerehabilitation platform (knee exercise postural assessment dataset [KneE-PAD]) using surface electromyography (sEMG) and inertial measurement units (IMUs). It involves delivering a structured exercise protocol through remote monitoring technology, dynamically adjusted based on patients’ biomechanical input, and providing real-time auditory and visual feedback during home-based exercise, compared with supervised in-person physiotherapy with similar exercise content over 8 weeks, with a 12-week follow-up.^[20] Both groups showed clinically relevant improvements in WOMAC pain and disability that exceeded the established minimal clinically important difference (MCID) of 9–12 points for total WOMAC score;^[21] however, the telerehabilitation group showed significantly greater pain and disability improvement at post-treatment and at 12-week follow-up, with larger effect sizes (Cohen’s d > 0.8), suggesting that sensor-augmented feedback may improve and maintain analgesia.

The effects of VR game-based exercises on pain and functional disability in KOA patients were investigated by Yilmaz and Baki.^[15] The use of interactive simulations generated by a computer, which may be game-based, provides an opportunity for patients. The use of interactive simulations generated by a computer, which may be game-based, provides an opportunity for patients to participate in an environment that resembles the real world through visuals. At the end of treatment, a significant difference was observed between the two groups on VAS and WOMAC. Based on the minimal clinically important improvement, the changes in VAS and WOMAC (function) are clinically significant at 1.08 cm and 5.3 points for knee OA.^[21] In this study, improvement in balance was also demonstrated only within the group. Ozlu et al.^[16] studied the effect of VR gamified rehab added to conservative physiotherapy compared to conservative rehab alone, in a 3-week (15-session) course. VR-based game allows users to perform exercises by interacting with motion sensors and receiving auditory and visual feedback.^[16] The VR group showed a significant improvement in VAS scores compared to the control group. The progress was maintained at the 7-week follow-up. For WOMAC scores, a significant improvement was observed at 3 weeks, but it did not persist through 7 weeks, as indicated by MCII values.^[21] The analgesic effect was attributed to increased engagement and adherence due to gamification. Mete and Sari examined exergaming combined with conventional physiotherapy, compared with conventional physiotherapy alone, over 6 weeks (30 sessions, 5 times/week). Game-based exercises are performed in a non- to semi-immersive VR-based environment through auditory and visual stimuli.^[17] Both groups showed significant improvements in pain, but the exergaming group had greater improvements in VAS and WOMAC (p < 0.05), with moderate effect sizes. This difference is clinically significant too, based on MCII values for VAS (10.8 mm)^[21] and the MCID value of WOMAC (9–12 points).

Lin et al. investigated the impact of AVG on pain and disability as measured by WOMAC in patients with KOA. Both therapeutic exercises and AVG showed equal efficacy in pain relief in patients with KOA.^[18] An active video game, a form of exergaming, where patients perform physical actions while receiving visual or auditory cues. However, AVG showed greater efficacy in dynamic balance, physical function, and overall physical health than therapeutic exercises.

A 12-week multifaceted wearable-based intervention delivered immediately (Fitbit tracking, FitViz application, and motivational counseling by a physical therapist) compared to a delayed intervention (control by newsletter) at 12 weeks was investigated by Li et al.^[19] It involves the use of a wearable device (Fitbit Flex 2 wristband) and counseling, patient education with visual feedback in both groups. The multifaceted wearable-based intervention did not lead to significantly greater improvements in the KOOS pain and disability subscale at 13 weeks in either group, indicating that although pain might be alleviated over time, the primary benefit of the wearable intervention lay in its impact on activity behavior rather than joint pain.^[20]

Brief conventional programs

Conventional physiotherapy for KOA typically involves strengthening exercises, as well as patient education and self-management techniques.^[22] The exercise protocol included knee range-of-motion exercises, quadriceps and hamstring stretching, isometric quadriceps strengthening, resisted active knee flexion and extension, and balance training (e.g., on a balance board or trampoline).^[15] In some exercise programs, electrotherapy modalities such as hot or cold packs, therapeutic ultrasound, and transcutaneous electrical nerve stimulation were incorporated.^[16] Other components included terminal knee extension, elastic-band leg press, functional lower-limb and gait exercises, cycling at 40–60% of heart rate reserve, and a stretching cool-down.^[17] Strengthening exercises were progressed with sandbags and Thera-Bands, usually done for 5–10 repetitions in 3–5 sets.^[18]

Safety and feasibility

The trials reported some adverse events. The most common were transient events: Mild nausea or headache in VR groups (reported by 2–3 participants as mild dizziness, fall, muscle cramp, muscle pain in early VR sessions (reported anecdotally), and minor increases in pain during the first 1–2 sessions of active video game or exergaming (expected transient responses).^[16,19] No serious adverse events (hospitalizations, falls, fractures) were reported.^[15,18] Two studies did not report whether patients experienced side effects.^[17,20] All studies reported that interventions were feasible and acceptable in patients with KOA.