Exergames to improve rehabilitation after knee arthroplasty: a systematic review and grade evidence synthesis

Abstract Objective We aimed to systematically review and synthesise the impact of rehabilitation with games in people after knee arthroplasty. Methods We conducted a systematic review following the Preferred Reporting Items for the declaration of Systematic Reviews (PRISMA – Preferred Reporting Items for Systematic Reviews and Meta-Analysis). The summary of evidence was developed using the Grading of Recommendations, Assessment, Development, and Assessment (GRADE). The review included randomised controlled trials that used characteristics of games in rehabilitation. Results Eight articles from a total of 1289 identified articles were included after duplicates were removed. In total, 239 participants participated. There were no statistically significant changes between the groups using the exergames and control groups. The level of evidence was rated using GRADE and was very low or moderate. The difference in grouped means was not significant for Knee Flexion, Knee extension, Range of Motion, WOMAC (Western Ontario and McMaster Universities Arthritis Index), AKSS (American Knee Society Score), Self-Efficacy, Five Times Sit-to-Stand Test time, Pain, or Proprioception. Conclusions The results of the different studies did not find significant changes in the intervention groups with exergames in the physical domains, especially in studies with shorter interventions. Therefore, further investment in future studies on developing and evaluating games is suggested to enhance training during the recovery process.


Introduction
Healthy knees are essential for optimal performance of daily activities [1].Knee arthroplasty is one of the most commonly performed orthopaedic procedures [1,2] and has become a common elective surgical procedure [3].A significant contributor to success is post-surgical rehabilitation [4].Therefore, patients should perform rehabilitation exercises at home, where effective postoperative rehabilitation remains crucial for optimal functional recovery [3,4].However, motivation to exercise may be low [5].
The postoperative rehabilitation process can be uncomfortable, tedious, and long, creating challenges that the patient needs to overcome.Patients generally show some resistance to participate in the rehabilitation process, resulting in more muscle atrophy [6,7].
Inadequate rehabilitation leads to a poor range of motion of the knee, muscle weakness, limited movement, functional disability, and patient dissatisfaction [4].Therefore, rehabilitation should begin immediately after surgery to facilitate the timely recovery of muscle activation, range of motion, and function [3,4].In addition, motivating rehabilitation tools are needed, given the importance of continuous exercise [8].
In the last decades, new technologies in rehabilitation have been developed, from the use of games to highly specialised systems with specific hardware and software platforms [9][10][11].Many studies have used gaming technologies that use movement as intervention tools in rehabilitation [7], and have observed that using games, like exergames, in rehabilitation increases patient's motivation [6].Rehabilitation powered by games in the format of exergames and virtual reality is increasingly used, and these aspects fall under the great technological development applied to the health area during the last two decades [12].
Exergames add exercise to a game, allowing the performance of reciprocal physical movements to interact with virtual environments, and are widely used in rehabilitation [13].However, in the context of rehabilitation, this utilisation is much higher in neurological disorders; in contrast, for the rehabilitation of musculoskeletal diseases, only a few publications can be found [12].This number is even lower when we specifically address the rehabilitation of the person undergoing knee arthroplasty, which justifies this review's undertaking.
When we analysed whether there were studies carried out on the topic, we observed that there were no reviews on this topic.The review studies identified in this population focussed only on virtual reality [9,11,14] or were not exclusively aimed at people after knee arthroplasty [10]; that is, they did not focus on the use of exergames.Despite the growing popularity of health games, there is no evidence of their use in rehabilitation after knee arthroplasty.Specifically, no review to date focuses on the impact of using games for rehabilitation after knee arthroplasty, which gives this review its unique character.This systematic review aimed to assess the impact of rehabilitation with games in people after knee arthroplasty.

Review design
This systematic review report elaboration follows the preferred report items for the declaration of systematic reviews PRISMA (PRISMA -Preferred Reporting Items for Systematic Reviews and Meta-Analysis) (Registration 10.17605/OSF).The summary of evidence was developed using the Grading of Recommendations, Assessment, Development, and Assessment (GRADE) framework to grade and present the strength and quality of the evidence.The online tool GradePro GDT was used to facilitate this process.

Criteria for study inclusion
This systematic review included randomised controlled clinical studies conducted in adults, in which there was both an intervention with games and a control group in people after knee arthroplasty.In some of the studies, the type of prosthesis is not specified, whether total or revision; therefore, all studies concerning knee arthroplasty were considered.Studies were excluded if they were not randomised, were uncontrolled, or involved participants in which the use of a strategy with game characteristics was not explicit in the intervention.All data resulting from the effect of the intervention with the game were included.Studies that did not include only patients who underwent knee arthroplasty were excluded.Unpublished studies and protocols of randomised clinical trials were also excluded.Publications in English, Portuguese, or Spanish were considered for inclusion, with no limitation on publication date.
While reviewing the numerous existing publications on the topic, the need to specify the concept of games used in this study was identified.Indeed, articles published in recent years show a broad interpretation of game elements, including different terminologies to define the way a game strategy is developed, integrating some common items, feedback, levels, leaderboards, challenges, badges, avatars, competition and cooperation [15]-which were considered in this review.In other words, throughout this review, only strategies that used game characteristics in their application were considered.These applications should include common game attributes structured with playful objectives to which therapeutic value has been added.

Search strategy
To conduct the literature search, the respective descriptors suitable for each database were identified using search syntaxes, illustrated in the Boolean phrases (Supplementary Appendix 1).The following databases were used: MEDLINE V R (Medical Literature Analysis and Retrieval System Online), CINAHL V R (Cumulative Index to Nursing and Allied Health Literature), SPORTDiscus, Psychology & Behavioural Sciences Collection, SCOPUS, SciELO (Scientific Electronic Library Online), and PEDRo (Physiotherapy Evidence Database).Free terms were used in addition to the descriptors of each database to make the search as comprehensive as possible.
Combinations of descriptors/medical subject headings (MeSH), subject headings, and subject terms were used for each database using the Boolean operators 'OR', 'AND' and the tool ' � ', which enhances a search by creating new variations of the same word.A search for additional studies was also carried out in the reference lists of all publications included in the review ('Backward citation searching').The research was carried out until April 2022.Two authors independently conducted the selection of the studies, and any disagreements were resolved by consensus with a third author to confirm the eligibility of publications.The Rayyan QCRI V R platform (the Systematic Reviews web app) (Qatar Computing Research Institute, Catar) was used for the first data selection stages.
The extraction of the selected data was carried out by two authors using a data extraction form.Disagreements were resolved with the help of a third author.

Data analyses
First the results were evaluated and selected regarding their relevance for inclusion based on the information provided in the title and summary.Subsequently, the selected articles were subjected to a full reading that preceded their integration in the final sample.Figure 1 shows the process of identification and inclusion of articles presented through PRISMA.Although, not all results of the studies included in this review could be compared and included in the meta-analysis, only 6 of the studies were included, but all studies were included in the GRADE outcome analysis.
To systematise the data extracted from the articles, they were first compiled descriptively in a table previously prepared by the researchers, including the place of the study, study objective, study design, participants, intervention, type of game, and results.Furthermore, this table facilitates the presentation and discussion of the results (Table 1).

Quality assessment
The risk of bias for the trials was assessed with the Cochrane risk of bias tool.The GRADE methodology was used to evaluate the quality of the body of the retrieved evidence.The concluded GRADE level of evidence ranged from moderate to low (Table 2).We collated the results and conducted a meta-analysis whenever possible.Standardised mean differences (SMD) between the participants, the intervention group, and the control group, and the respective 95% confidence intervals were calculated.Given the methodological heterogeneity between studies, a random-effects model was used to calculate the pooled effect estimate.The results of each analysis are presented in a forest plot (Figure 2), showing the joint effect estimate for each outcome along with the respective confidence interval.p < 0.05 was considered statistically significant.All meta-analysis calculations were obtained using the Revman 5.4 software (Cochrane Collaboration, Oxford, England).

Study characteristics
Table 1 summarises the characteristics of the 8 studies included in the review regarding the authors, year, type of study, study objectives, participants, type of intervention game, control group and outcomes measures.
In total, 239 participants participated.Regarding the characterisation of these participants, in one study [6], only the experimental group was characterised, meaning it is not possible to specify the total number of men and women who participated in the study.In another study [19], only women were included.The average age of the participants ranged from 65.58 to 73.58 years.All studies resorted to games in an  -American Knee Society Score-AKSS--Scale SER (Self-Efficacy for Rehabilitation).
-System usability scale (SUS) scale of 10 items.
Until the date of discharge from the physiotherapy services Followed up on an outpatient basis after surgery for an average of 54 days -Active range of motion of the knee (ROM) -Proprioception. (continued) exercise dimension falling on the concept of Exergames, or integrating virtual reality.Most interventions took place in the postoperative period while still in an acute setting, starting on the second postoperative day [6,7,18,20].The methodological quality of the studies is shown in Figure 3.The most common methodological limitation was the lack of blind evaluation by the evaluator or the implementation of the intervention.Another limitation of the included studies was the small sample size and the lack of long-term follow-up in some studies [6,7,20].Using the GRADE 'summary of findings' table (Table 2), the quality of evidence is seen as mostly moderate and low, essentially due to limitations of the studies related to the risk of bias (detection and performance).
Table 2, using the meta-analysis of pooled data, allowed us to identify the outcomes for constructing the GRADE evidence profile of the included studies.Not all results of the studies included in this review could be compared and included in the meta-analysis presented in Figure 2.

Games
Among the studies included, 3 used exergames that had already been developed, namely the Nintendo Wii Fit and motion control using the Wii Balance Board V R (WBB) [16,17,19].Wii Fit V R Games require the participant to be involved in postural control and balance.For example, in one of these studies [16], the games used were Ski Slalom V R , Tightrope Walk V R and Penguin Slide V R , for lateral weight change, and Table tilt V R , Hula Hoop V R , and Balance Bubble V R games for multidirectional balance.Other Nintendo Wii resources were also used but not as games.These included Deep Breathing V R , Half Moon V R and Torso Twist V R for static and dynamic postural control.
In another study [17], the application of games was progressively mentioned, with the following objectives: Bilateral Posture (Ultimate Balance Challenge , and Lunging (Yoga V R , Strengthening: Lunge V R ).In the third study [19], the authors always use the same game (Ski Slalom V R ) with or without restriction-induced movement (CIMT).Specific rehabilitation systems were developed in the remaining five studies [6,7,18,20,21], bringing together the characteristics of virtual reality and exergames.
In one study, the authors used the image processing technology of Kinect to detect movements while the patient rode an exercise bike and used the data to direct an avatar who runs with the virtual bike in the game [7].In another study, the patients stood or sat on chairs and a Kinect sensor was placed near a computer monitor in front of the person to capture their body movements [6].Another study involved patients being asked to row a boat using knee flexion in an immersive virtual environment [18].Finally, in the last study [21], the authors used a system that digitises patient's movement and provides real-time feedback on performance via a mobile app.This also included a web-based platform that  EUROPEAN JOURNAL OF PHYSIOTHERAPY allowed the clinical team to prescribe, monitor, and adapt the rehabilitation process remotely [21].

Outcome measures
Table 1 shows the heterogeneity of instruments used regarding both objectives and timing of applications, making it difficult to group results.Some of these instruments are concerned with issues such as physical function, balance, range of motion, and pain.Several instruments were used, namely: ROM (Range of Motion), WOMAC (Western Ontario and McMaster Universities Arthritis Index), AKSS (American Knee Society Score), Self-Efficacy, FTSST Time (Five Times Sit-to-Stand Test), Knee Flexion, Knee extension, Pain and Proprioception.
The evaluation of game strategies is much less frequent.In two studies [6,7], the SUS (System usability scale) was used to assess usability.In another [7], scoring and playing time were also used.While in another [21], a question was included asking participants whether they would recommend the system to friends and neighbours.

Outcome results
Three of the studies investigated the impact of exergames using the WOMAC scale [18][19][20] With regard to the total score, the difference in grouped means was 1.09 (95% CI 0.02-2.17;p ¼ 0.05), favouring the exergame group although with no statistically significant result.The GRADE of evidence was categorised as low quality.In the pain dimension of the same scale, the difference in grouped means was À 0.36 (95% CI À 0.74 to 0.02; p ¼ 0.06), favouring the control group, although this was not statistically significant.
Two studies assessed the impact of exergames using the ROM scale [18,20].The difference in grouped means was 1.25 (95% CI 0.72-1.78;p ¼ 0.42); while not statistically significant, the values were higher in the exergame group.
Regarding the impact of exergames on knee mobility, our meta-analysis did not demonstrate a significant standardised mean difference for the results of knee flexion (0.31: 95% CI À 0.06 to 0.69; p ¼ 0.10) or knee extension (0.40: 95% CI À 0.48 to 1.28; p ¼ 0.38) at the end of the intervention [7, 16,21], although the values were higher in the exergame group.
At the end of follow-up, as shown in Table 2, there was a significant standardised mean difference for outcomes resulting from individual studies; namely, the application of the AKSS scale, the Knee score dimension (p ¼ 0.002), the application of the scale Self-Efficacy (p ¼ 0.03), the application of the FTSS time (p ¼ 0.02), and proprioception (p ¼ 0.002), favouring the exergame group.
The evaluation of satisfaction and pleasure regarding the use of exergames did not receive much attention by any of the authors.This is undoubtedly an aspect to be highlighted in this review study, especially as this was an aspect that we intended to explore.The two studies that sought to assess the satisfaction, motivation, and usability of the systems, found statistically significant results [6,7].In one of these studies [7], they state that satisfaction affected rehabilitation and game performance: knee flexion angles (p < 0.01), game scores (p < 0.01), and game times (p < 0.01) [6].Similarly, in the other study [6], there were significant differences in the rehabilitation achievements in the knee score (p ¼ 0.016) and in self-efficacy (p ¼ 0.03) between the experimental group and the control group [6].On the other hand, in a different study [16], the authors report that some participants in the study group expressed enjoyment in using the Wii Fit V R , statements that the control group did not make during participation in the study activities [16].
In this study [21], patients were asked to answer the following question: 'On a scale of 0 to 10, how much would you recommend the system to one of your friends or neighbours?' Twenty-seven patients evaluated the system with 10, one patient evaluated the system with 9, and two evaluated the system with 8 [21].

Principal findings
This study aimed to systematically synthesise the impact of rehabilitation with Exergames on people after knee arthroplasty.This review identified eight studies with a total of 239 participants.Regarding the main results of the analysis, the mean difference for the total score of the WOMAC scale [18][19][20], and its pain dimension, showed a mean difference with no statistically significant effect, although the values were higher in the exergames group.Similarly, the mean difference in the ROM scale [18][19][20] had no statistically significant results.Furthermore, the impact of exergames on knee mobility (knee flexion, knee extension) [7, 16,21] and pain [16,[18][19][20][21] also showed no statistically significant relevance, although the values were also higher in the exergame group.
However, there was a statistically significant difference in the results of individual studies for the application of the AKSS scale [6], the Self-Efficacy scale [6], the FTSS time [17], and for proprioception [20], all favouring the exergame group.Overall, based on the systematic review and synthesis of GRADE evidence performed and according to the data that could be pooled, our study did not suggest a significant association between the exergames and the outcomes analysed.However, these results highlight that there was no aggravation or complications arising from the use of exergames in the rehabilitation of the person after knee arthroplasty.
The evidence included in this review is limited by the complexity of the interventions and assessment mechanisms.In particular, the results of these studies could not be grouped for the following reasons.Heterogeneity was present in the type, phase, and time of intervention, as well as in the outcome measures of the included studies.Diversity is seen in the phase in which the intervention is applied, ranging from the immediate postoperative period to six months postoperatively.This variability is also visible in the intervention's frequency and duration, which is reflected in the numerous measurement instruments applied at variable moments from study to study.
The results of the different studies found no significant changes in the intervention groups in the physical domains, especially in studies with shorter interventions.This is especially noted in those studies [6,7,16,20] restricted to the first postoperative days.This is probably because short hospital stays are reduced, and longer additional gaming therapy may not show effectiveness [22].
Individual observation of each study suggests that interventions over a longer time span may have better results, and it is important to consider this in future studies.This is the case for the study [18] in which the intervention takes place over six months, with the authors obtaining the following results: reduction of postoperative pain, improvement in functional recovery, and increased motivation for rehabilitation [18].Similarly, another study [21] using an 8-week intervention showed an impact on functionality and balance, a finding that is in line with the clinical window of these patients' recovery [21].Indeed, the ideal window for rehabilitation in the person undergoing knee arthroplasty is in the first three to six months after the operation [4].
When considering the results of this review, it is important to highlight the conceptual difficulty of a 'game' that the authors identified throughout the consulted studies.Throughout the research, it was possible to identify in some studies a reference to the use of games, or consoles, when in fact, they refer to the use of rehabilitation technologies without the use of games [22][23][24][25].
Similarly, despite the substantial development of these resources in the scope of rehabilitation, we observed that many studies analysed games as resources for rehabilitation and not as games per se.Few explore the real effects of games on patients undergoing knee arthroplasty, forgetting primordial aspects such as satisfaction with the use of the game.In the few studies investigating the user experience, there is a tendency for a positive impact on the rehabilitation process, particularly on adherence and motivation for physical activities and adherence to rehabilitation programs [6,7].This aspect is corroborated in this review study on technologies in rehabilitation in hip and knee arthroplasty [9].The evaluation of games is a primordial step, and a standardised structure must be used to legitimise the enthusiasm observed in the application of the game [26].Especially as this is possibly the greatest contribution of exergames through intrinsic and extrinsic motivation in the rehabilitation process [6,7], congregating repeatability, feedback, and motivation [16].
The impact of games on the rehabilitation process needs to be highlighted and recognised.Contextual, personal, and motivational factors can influence and affect the effectiveness of rehabilitation (both the process and the result) [20].These aspects were highlighted in one of the studies [7], where the author demonstrated that the system was useful in reinforcing motivation [7].Similar aspects were mentioned in another study with positive outcomes regarding rehabilitation self-efficacy and the system's usability [6].Unfortunately, these two studies were the only ones that evaluated the resources used in the experimental groups.This is important to bear in mind, as, in addition to the hardware, the design of rehabilitation games also plays a significant role.Although the aforementioned studies include some games in their systems, few explore the potential of game design [4], nor do they evaluate it.
The potential of games' characteristics was also mentioned in a previous study where the authors proposed changes to the design and operation of the system, including feedback coded by colours, performance scores, goals, among others, after the application of the rehabilitation system for patients undergoing knee prosthesis [24].However, in the studies analysed, there is a critical gap in the results and evaluation of the games that should go beyond its impact on physical outcomes.Especially as many questions could be addressed: What is being evaluated?Rehabilitation?The game?The system? Teleworking?Surgical procedures?The implanted material?
All the above reinforces the idea that future health games should expand the duration and repetition of games and increase the duration of follow-up assessments to provide evidence regarding long-term effectiveness [27].Particularly as game systems have proven to be more fun than conventional exercises in a rehabilitation environment [2].Therefore, there is a need to develop more games that can be used for rehabilitation, which can be helpful to attract and maintain the user's attention [7].
Overall, the authors make some recommendations, noting that although rehabilitation systems for different diseases have been developed, some devices still have restrictions that must be addressed and improved [6,7].For example, some equipment requires the user to hold the remote control, and motion detection is limited to the upper extremity, making it difficult to perceive lower limb exercise feedback [7].In addition, more diversified rehabilitation games should be designed, offering users various attractive features of their choice to improve interest in rehabilitation [6,20,28].
Most authors also comment that larger study samples are needed for more robust data and results and to provide more evidence regarding the impact of these strategies [6,16,17,20].Among other limitations, the impossibility of carrying out a double-blind trial due to the nature of the interventions is noted [16,17,20].

Limitations
The present review has some limitations.Only published randomised controlled trials were included; however, this criterion was chosen to improve quality and provide more reliable information on the effect of the intervention used.In addition, one of the existing difficulties was the lack of specification regarding the game strategies used in studies and some confusion regarding game concepts.On the other hand, it should be noted that the systematic review was registered late and after it had started.The evidence included in this review is also limited by the complexity of the interventions, duration, and evaluation mechanisms.

Conclusion
These results fall short of the potential of the games, namely, to promote the involvement of participants in rehabilitation.This review shows no conclusive evidence that interventions with games are more effective than standard treatment in the rehabilitation of patients after knee arthroplasty.Although, the evidence regarding interventions is superior for longer programs.Future research, using randomised clinical trials with a well-defined research protocol, is necessary to determine the effectiveness of using games in the functional domain as well as in the patient's involvement in the rehabilitation process.

Figure 1 .
Figure 1.Process of identification and inclusion of articles -PRISMA Flow Diagram.
interval; MD: mean difference.a Some studies with risk of bias arising from the randomisation process.b Different intervention time.c Some studies did not blind participants and no blinding of outcomes assessment-Risk of bias.d No blinding of participants.e Risk of bias arising from the randomisation process.f No blinding of outcomes assessment-Risk of bias.

Figure 3 .
Figure 3. Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Table 1 .
Characteristics of included studies (n ¼

Table 2 .
GRADE evidence Profile of included studies.