The Efficacy of Closed Versus Open Kinetic Chain Exercise in Knee Osteoarthritis—A Systematic Review and Meta‐Analysis

ABSTRACT

Background/Purpose

Despite the prevalence of knee osteoarthritis, no consensus exists in the most effective exercise type for improving patient clinical outcomes. Therefore, we conducted a systematic review to compare the effectiveness of closed and open kinetic chain exercises (CKC and OKC) in reducing pain, increasing physical function and improving quality of life (QoL) in people with knee OA.

Methods

We searched PubMed, MEDLINE, EMBASE, CINAHL, PEDro, Scopus and Cochrane from inception to April 2024. Studies were included if they compared pain, physical function and QoL outcomes via a randomised controlled trial design in people with knee osteoarthritis. The Pedro Scale was used to assess the risk of bias, and the GRADE approach was used to determine the certainty of evidence. Outcomes were pooled in a random effects meta‐analysis. 13 randomised controlled trials evaluating the efficacy of CKC versus OKC exercises in the short term (≤ 13 weeks) were included.

Results

Low‐certainty evidence found that CKC exercises were superior to OKC exercises in the short term for improving pain (standardised mean difference (SMD) −0.18, 95% CI –0.37 to < 0.00; p = 0.048) and self‐reported physical function (SMD –0.24, 95% CI –0.44 to −0.03, p = 0.022), with very low‐certainty evidence of no difference in QoL outcomes (SMD –0.25, 95% CI –0.81 to 0.30, p = 0.370).

Discussion

To reduce pain and improve physical function in individuals with knee osteoarthritis in the short term, physiotherapists should prescribe CKC exercises. However, there is a paucity of information outlining the effects of these CKC versus OKC in the medium to long term.

Study Registration Number

PROSPERO (CRD42023470739)

1. Introduction

Osteoarthritis (OA) is the most common form of arthritis in Australia, often affecting the weight‐bearing joints of the hips, knees and spine (Australian Institute of Health and Welfare ²⁰²⁴). The knee joint is most frequently affected by OA (Steinmetz et al. 2023), impacting over 1.2 million Australians (Australian Commission on Safety and Quality in Health Care ^2024a). In 2021–2022, 53,000 total knee arthroplasties (TKA) were performed for knee OA in Australia, with four times as many people hospitalised for knee OA over the same period (Australian Institute of Health and Welfare ²⁰²⁴). Non‐surgical treatments, including exercise therapy, are recommended as first‐line treatments for knee OA (Australian Commission on Safety and Quality in Health Care ^2024b). While various exercise therapy types and modes have demonstrated positive effects on patient‐reported outcomes in people with knee OA (Lim and Al‐Dadah ²⁰²²; McAlindon et al. ²⁰¹⁴; Fan et al. ²⁰²⁴), no agreement exists regarding which exercise therapy type produces superior clinical outcomes (Moseng et al. 2024; Young et al. ²⁰²³; Kolasinski et al. ²⁰²⁰).

Exercise is safe and beneficial for people with knee OA, with reports of reduced pain, improved knee function, increased activity and participation, and improved quality of life (Ackerman et al. 2023; Chen et al. ²⁰¹⁹). Reported programs may include closed kinetic chain (CKC) and open kinetic chain (OKC) exercises, defined by the distal segment of the limb remaining fixed or not, respectively (Pamboris et al. 2024). Comparative evaluations of OKC and CKC exercises have been explored in people with other knee conditions, such as patellofemoral pain (Mustafa et al. 2022; Witvrouw et al. ²⁰⁰⁰) and following anterior cruciate ligament reconstruction (Pamboris et al. 2024), indicating greater pain reduction and improved function in those undertaking CKC exercises (Pamboris et al. 2024; Mustafa et al. ²⁰²²). It is proposed that increased knee strength and stability, achieved through the co‐contractions associated with the compound movements in CKC exercises (Pamboris et al. 2024; Mustafa et al. ²⁰²²; Witvrouw et al. ²⁰⁰⁰), might explain the observed difference. It is possible that similar findings might be observed in people with other knee conditions, including those further along the disease continuum, such as knee OA. Therefore, the aim of this systematic review and meta‐analysis was to determine whether CKC exercise is superior to OKC exercise in reducing pain, increasing physical function and improving quality of life (QoL) in people with knee OA. We hypothesised that CKC exercise would demonstrate superior outcomes in pain, function, and QoL compared to OKC.

2. Methods

The systematic review is presented in accordance with the Preferred Reporting Items for Systematic Review and Meta‐analyses (PRISMA) statement (Page et al. 2021) and was registered with PROSPERO (CRD42023470739).

2.1. Search Strategy

The electronic databases of PubMed, MEDLINE, EMBASE, CINAHL, PEDro, Scopus and Cochrane Library were searched from inception until April 2024. Search terms were based upon the three main concepts of (1) knee OA, (2) CKC exercise, and (3) OKC exercise. Synonyms and medical subjective headings (MeSH) were used for each concept and combined with the ‘OR’ Boolean operator (Table S1). No limits were placed on language, publication period, or publication type. Additional searching included citation tracking and reference checking of the included articles.

Search yields were imported into Covidence (Veritas Health Innovation ²⁰²³) for eligibility screening. At least two reviewers from a pool of six (H.J.M., H.D.P., M.G.K., L.S.K.L., M.H., M.J.S.) screened the titles and abstracts of all articles. Full texts of potentially eligible articles were sourced and independently evaluated by two reviewers from the same pool of six. Articles with uncertain eligibility were resolved through consensus during both screening phases. A third reviewer (A.I.S.) independently reviewed the paper if consensus could not be reached.

2.2. Selection Criteria

Studies were included in the review if they were published randomised controlled trials (RCT) in English (due to a lack of funding to support translation services) with participants aged ≥ 18 years and diagnosed with symptomatic primary (idiopathic) unilateral or bilateral knee OA. RCTs had to evaluate the interventions of CKC exercise therapy compared to OKC exercise therapy on patient‐reported outcomes of pain, physical function, or QoL. Knee OA could be diagnosed via recognised criteria (e.g., American College of Rheumatology criteria) (Altman et al. 1986), or National Institute for Health and Care Excellence (NICE) Diagnosis Guidelines (National Institute for Health and Care Excellence ²⁰²²), or by patient self‐report of a previous diagnosis of knee OA. Studies that investigated non‐land‐based interventions (e.g., hydrotherapy), included combined‐chain exercise programs, or involved a cohort where > 10% of the participants had previously undergone TKA (or partial) at baseline/enrollment were excluded.

2.3. Risk of Bias

The risk of bias of included RCTs was evaluated using the Physiotherapy Evidence Database (PEDro) (PEDro ¹⁹⁹⁹). The PEDro tool assesses an RCT's risk of bias over 11 different items (PEDro ¹⁹⁹⁹). When available, scores were directly transcribed from the PEDro online database (https://pedro.org.au/). Where scores were not available on the database, RCT quality was assessed using the PEDro scale performed by two independent reviewers (M.G.K. and M.J.S.). Discrepancies were settled based on consensus or a third independent reviewer (A.I.S.) if consensus could not be met.

2.4. Data Extraction

Data extraction was completed by two independent reviewers (H.J.M. and one of H.D.P., M.G.K., L.S.K.L., M.H.) and included information relating to population demographics, diagnosis, sample size, interventions undertaken, and outcomes relevant to the systematic review aims. Extracted data were collated in a bespoke Excel spreadsheet (Microsoft Corporation ²⁰¹⁸) developed specifically for this systematic review. All data were extracted as reported, irrespective of completeness. No missing data was imputed, and the corresponding authors were not contacted to clarify their reported results.

In the event an RCT presented data at multiple time points within one of our a priori defined follow‐up windows (short‐term, 0–3 months, medium‐term > 3 months to 6 months, and long‐term > 6 months), for example, 4 weeks and 8 weeks, the data from the longest follow‐up time point were extracted. Furthermore, if an RCT presented data on more than one patient‐reported outcome measure relating to pain, physical function, or QoL, data were extracted from the highest listed outcome based on the following hierarchy (Fransen and McConnell ²⁰⁰⁸; Jüni et al. ²⁰⁰⁶; Reichenbach et al. ²⁰⁰⁷).

2.4.1. Pain Outcomes Only

1. Average pain—via visual analogue scale (VAS) or numerical pain rating scale (NPRS)

2. Pain during walking

2.4.2. Pain, Physical Function, and Quality of Life Outcomes

• 3Western Ontario and McMaster University Osteoarthritis Index (WOMAC)

  • Pain subscale—Pain outcome

  • Physical function subscale ‐ Physical function outcome

  • Total score—Quality of Life outcome

• 4Short Form—36 Health Survey Questionnaire (SF‐36)

  • Pain summary—Pain outcome

  • Physical function summary ‐ Physical function outcome

  • Mental component summary—Quality of Life outcome

• 5Knee Injury and Osteoarthritis Outcome Score (KOOS)

  • Pain—Pain outcome

  • Activities of daily living ‐ Physical function outcome

  • Knee‐related quality of life—Quality of life outcome

• 6Any other patent‐reported outcome measure reported

2.5. Data Synthesis and Analysis

To assist with clinical interpretability, studies were pooled if deemed to be clinically homogeneous based on outcome (pain, physical function, or QoL) and follow‐up time period (short‐term, 0–3 months; medium‐term, > 3–6 months; and long‐term, > 6 months) (Fan et al. 2024). Post‐intervention means, standard deviations and sample sizes were used for analysis. Where studies reported post‐intervention outcomes as non‐parametric data, and sufficient information was reported (i.e., median and range (or interquartile range)), means and standard deviations were calculated based on conversion methods outlined by Wan et al. for analysis (Wan et al. 2014). Standardised mean differences (SMD) were calculated using corresponding 95% confidence intervals by dividing the differences between groups by the pooled standard deviation. SMDs were pooled in a random effects meta‐analysis, with the pooled SMD defined as small (SMD = 0.2), medium (SMD = 0.5) and large (SMD = 0.8) (Cohen 1992). Statistical heterogeneity was determined via the x ² statistic, with inconsistency evaluated with the I ² statistic and defined as low (I ² = 25%), moderate (I ² = 50%), and high (I ² = 75%) (Higgins et al. 2003). In the event that 10 or more studies were pooled, funnel plots were used to assess the presence of publication bias, with asymmetry assessed using Egger's test. Where studies could not be pooled, their outcomes were qualitatively described via a best‐evidence synthesis. All statistical analysis and data visualisation was conducted in R (R, R Foundation for Statistical Computing), using the ‘Meta’ and ‘Metafor’ packages.

The Grading of Recommendations Assessment, Development and Evaluation (GRADE) (Guyatt et al. 2008) approach was used to describe the certainty of the body of evidence for each meta‐analysis, consistent with previously defined methods (Fan et al. 2024; LeBel et al. ²⁰²⁵). In summary, the quality of evidence was rated as very low, low, moderate, or high based on the following criteria: risk of bias (downgraded if average PEDro scores across studies as < 7); inconsistency (downgraded if I ² > 50); indirectness (downgraded if heterogeneous intervention or population); imprecision (downgraded if CI > 0.25 in either direction); and publication bias present (Figure S1).

3. Results

3.1. Study Selection

After duplicates were removed, 7692 articles were identified using the search strategy and assessed for inclusion (Figure 1). Following abstract screening, 63 full‐text articles were evaluated against the inclusion criteria. 13 studies met the eligibility criteria and were included for final analysis (Adegoke et al. 2019; Bennell et al. ^{2014, 2020}; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Gbiri et al. ²⁰¹³; Girgin et al. ²⁰²⁰; Jan et al. ²⁰⁰⁹; Moreira et al. ²⁰²¹; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³; Shah ²⁰¹⁴; Verma ²⁰¹²). All 13 studies included short‐term follow‐up only, with no studies evaluating outcomes over a medium‐ or long‐term period.

FIGURE 1.

PRISMA flow diagram.

3.2. Participant Characteristics

Characteristics of the included studies are outlined in Table 1. In summary, 651 participants (166 men and 341 women from 9 studies (Bennell et al. 2014, 2020; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Girgin et al. ²⁰²⁰; Jan et al. ²⁰⁰⁹; Moreira et al. ²⁰²¹; Özüdoğru and Gelecek ²⁰²³; Verma ²⁰¹²) with 4 studies (Adegoke et al. 2019; Gbiri et al. ²⁰¹³; Olagbegi et al. ²⁰¹⁶; Shah ²⁰¹⁴) not reporting the male/female split of their sample) with knee OA were included in this review. The average study cohort age ranged from 53.05 (Bennell et al. 2020) to 65.9 (Gbiri et al. 2013) years with OA diagnosis confirmed through radiographic evidence (Bennell et al. 2014, 2020; Girgin et al. ²⁰²⁰; Verma ²⁰¹²), various clinical criteria (Adegoke et al. 2019; Bennell et al. ^{2014, 2020}; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Jan et al. ²⁰⁰⁹; Moreira et al. ²⁰²¹; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³; Shah ²⁰¹⁴; Verma ²⁰¹²); or clinical examination findings from Orthopaedic Surgeons (Jan et al. 2009).

TABLE 1.

Overview of included studies.

Study	Participant inclusion	Diagnosis	Recruitment	Closed kinetic chain (CKC) group	CKC intervention notes	Open kinetic chain (OKC) group	OKC intervention notes
Adegoke et al. (2019)	—Kellgren‐Lawrence classification grade 2 or 3, AND —American College of Rheumatology Clinical Criteria	Knee OA	Participants receiving treatment at University College Hospital in Ibadan, Nigeria	Eight‐week progressive WB exercise program (and 100 mg of NSAIDs daily) — n = 14 (F = not reported) —Age: 57.57 (8.72) years old	Program consisting of progressive exercise additions, commencing with quadriceps setting, adding mini squats, wall slides and step‐ups/downs with increasing weight over the 8‐week program duration	Eight‐week progressive NWB exercise program (and 100 mg of NSAIDs daily) — n = 15 (F = not reported) —Age: 60 (8.81) years old	Program consisting of progressive exercise additions; commencing with quadriceps setting, adding SLR and full arc knee extensions (at 10‐repetition maximum intensity), increasing weight over the 8‐week program duration
Bennell et al. (2014)	—Age > 50 years, AND —Knee pain > equal to 25 mm on 100 mm VAS, AND —Pain or tenderness over medial knee, AND —Kellgren‐Lawrence classification grade 2 or 3, AND —Varus malalignment, AND —Radiographic evidence of a medial tibiofemoral joint compartment narrowing and osteophytes	Varus‐malaligned medial knee OA	Recruited from the community via advertisements (Melbourne, Australia)	NEXA strengthening and balance exercises — n = 50 (F = 26) —Age: 62.7 (7.3) years old	Six strength and balance exercises with progressions (TheraBand, weight and others), including functional knee and hip strength series, step‐ups/downs, static and dynamic single‐leg balance, and leg slides	Quadriceps strengthening exercises — n = 50 (F = 26) —Age: 62.2 (7.4) years old	Five strength exercises with progressions (TheraBand, weight, ROM and duration of contraction); including knee extensions at increasing ranges of motion and SLRs
Bennell et al. (2020)	—Age > 50 years, AND —Knee pain most days of the month, AND —Knee pain > 3 months, AND —Pain severity ≥ 4 on 11‐point NRS scale over the previous week, AND —Radiographic evidence, AND —Obesity (BMI < 30 kg/m2), AND —Mobile phone use and consent to receive messages	Medial knee OA	Recruited from the community (Melbourne, Australia) using advertisement through consumer organisations, social media, community locations, media, and volunteer database	WB strength exercises with progressions — n = 62 (F = 45) —Age: 60.9 (6.8) years old	Five strength exercises with progressions (TheraBand, weight and complex movement); including leg slides, stepping forwards/sideways, squat and step up/down variations, and hip muscle exercises	NWB strength exercises with progressions — n = 66 (F = 41) —Age: 62.4 (6.7) years old	Five strength exercises with progressions (TheraBand, weight and complex movement); including knee extensions at increasing ranges of motion and SLRs
Daşkapan et al. (2013)	—Females aged between 30 and 65 years, AND —Kellgren‐Lawrence classification grade 2 or 3	Bilateral knee OA	Patients were recruited from the Department of Physical Therapy of University Hospital, Ankara, Turkey between February 2009 and February 2010.	Mini squat exercise — n = 20 (F = 20) —Age: 57.50 (SD not provided) years old	The MSE was performed with 3–4 s holds, 20 reps for 2 sets daily, for a 3‐week duration. The number of repetitions was increased by 5 every second day so that at program end (Day 15), 55 repetitions twice daily were performed for a further 4 weeks after program completion	Straight leg raises exercise — n = 20 (F = 20) —Age: 59.5 (SD not provided) years old	The SLR exercises was performed with 3–4 s holds, 20 reps for 2 sets daily, for a 3‐week duration). The number of repetitions was increased by 5 every second day so that at program end (Day 15), 55 repetitions twice daily were performed for a further 4 weeks after program completion
Desai et al. (2022)	—American College of Rheumatology Clinical Criteria	Unilateral knee OA	Community/Clinic recruitment (Dr D. Y. Patil College of Physiotherapy in India)	CKC exercise — n = 15 (F = 8) —Age: 55.6 (4.94) years old	Exercises of mini squats, step ups/downs, lunges, wall slides. Each exercise was progressed over the 4‐week program duration by increasing range of motion, duration of exercise and repetitions.	OKC exercise — n = 15 (F = 5) —Age: 54.07 (6.00) years old	Exercises of mini squats, step ups/downs, lunges, wall slides. Each exercise was progressed over the 4‐week program duration by increasing range of motion, duration of exercise and repetitions
Gbiri et al. (2013)	Not reported	Knee OA	Nigeria, no further details provided A total of 30 recruited participants, 25 of whom completed the study. It is unclear how many participants were allocated to each group	CKC exercise — n = 15 (F = not reported) —Age: 60.2 (8.3) years old	Exercises including static and dynamic components on both double and single leg (e.g., walking around wide and narrow circles, retro walking) were completed twice weekly for 30 min over the 8‐week program duration	OKC exercise — n = 15 (F = not reported) —Age: 65.9 (8.1) years old	Exercises including knee extension/flexion, fixed biking, hip abductor and adductor strengthening, and stretching of the quadriceps and hamstrings were completed twice weekly for 30 min over the 8‐week program duration
Girgin et al. (2020)	Radiographic evidence	Knee OA	Not reported	CKC exercise — n = 19 (F = 17) —Age: 59.84 (6.54) years old	Exercises included mini‐squats, knee flexion, single leg stance, calf raise, supine wall push, and hamstring active stretching. Each exercise (except stretching) was held for a 10‐s duration over 10 repetitions; stretching was held for a duration of 15 s over 5 repetitions. The program was initially completed daily for 30 min under supervision and continued at home for 3 weeks	OKC exercise — n = 19 (F = 15) —Age: 56.75 (7.8) years old	Exercises included knee extensions, SLR, prone hamstring curl, and hamstring active stretching. Each exercise (except stretching) was held for a 10‐s duration over 10 repetitions; stretching was held for a duration of 15 s over 5 repetitions. The program was initially completed daily for 30 min under supervision and continued at home for 3 weeks
Jan et al. (2009)	—American College of Rheumatology Clinical Criteria, AND —Diagnosis of bilateral knee OA by two orthopaedic surgeons after clinical history and physical exam, AND —Radiological evidence, AND —Age 50 years or older, AND —Chronic knee pain for 6 months or more, AND —Willingness to cease medication	Bilateral knee OA	Taiwan, no further details provided	WB knee exercise in sitting — n = 36 (F = 24) —Age: 62 (6.7) years old	Eight‐week program completed 3 times week, with each session consisting of 6 repetitions for 4 reps, with 1 min rest between sets. The exercise performed includes 1 foot fixed on the centre of the pedal of an EN0Dynamic resistance device while the participant is seated. The knee is fully extended and flexed to 90° of knee flexion at a speed of 90°/2 s.	NWB knee exercise in sitting — n = 35 (F = 25) —Age: 63.2 (6.8) years old	Eight‐week program completed 3 times week, with each session consisting of 6 repetitions for 4 reps, with 1 min rest between sets. Participants were positioned facing away from the EN‐Tree machine and instructed to sit and maintain the knee at a 90° flexion angle with the distal extremity free, and a cable attached to this distal segment. Participants extended their knee to 0° before returning to the starting position, the exercises was completed at a speed of 90°/2 s
Moreira et al. (2021)	—American College of Rheumatology Clinical Criteria, AND —Kellgren‐Lawrence classification grade 2 or 3	Knee OA	Recruitment through television adverts, posters in academic centres and referrals from rheumatologists in Brazil	WB exercise — n = 14 (F = 9) —Age: 65.6 (10.3) years old	The 50 min program consisted of a 10 min warm‐up phase (walking on flat ground), 20 min of strength training (four sets, with six repetitions per set on the WB leg press machine), and 20 min of an ice pack applied on the participant's knees. The exercise was begun at 50% 1RM and progressively increased by 5% every 2 weeks, with a working ROM of 90%–40% of knee flexion	NWB exercise — n = 16 (F = 9) —Age: 58.7 (7.2) years old	The 50 min program consisted of a 10 min warm‐up phase (walking on flat ground), 20 min of strength training (four sets, with six repetitions per set on the NWB knee extension machine), and 20 min of ice pack applied on the participant's knees. The exercise was begun at 50% 1RM and progressively increased by 5% every 2 weeks, with a working ROM of 90%–40% of knee flexion
Olagbegi et al. (2016)	—American College of Rheumatology Clinical Criteria, AND —Kellgren‐Lawrence classification grade 2 or 3	Knee OA	Patients with mild to moderate knee OA (primary and secondary) recruited from the Physiotherapy Department, FMC, Owo, Nigeria, between January 2013 and December 2014	CKC exercise — n = 27 (F = not reported) —Age: 61.19 (3.29) years old	Participants completed three exercise sessions a week comprising quadriceps setting, cycling in the air, SLR and full‐arc knee extensions set corresponding to each participant's 10 RM, with sets of 10 repetitions completed for one set; weight was used for progression according to a new 10 RM from a varying timepoint. Participants were also administered 3000 mg of paracetamol daily	OKC exercise — n = 28 (F = not reported) —Age: 63.57 (14.33) years old	Participants completed three exercise sessions a week comprising quadriceps setting, cycling in the air, SLR and full‐arc knee extensions set corresponding to each participant's 10 RM, with sets of 10 repetitions completed for 3 sets; the quadriceps setting exercise was only completed for one set. Each week exercises were progressed according to a new 10 RM Participants were also administered 3000 mg of paracetamol daily
Özüdoğru and Gelecek (2023)	—Kellgren‐Lawrence classification grade 2 or 3, AND —Males and females aged 45–75 years diagnosed with unilateral knee OA grade I or II, AND —Ongoing knee pain for ≥ 3 months	Knee OA	Not reported	CKC exercise — n = 20 (F = 10) —Age: 54.4 (7.92) years old	Participants completed a progressive program 3 times weekly for 12 weeks, with the first 6 weeks completed under supervision. Exercises included sit‐to‐stand, mini squats, lunges and three side step‐ups. Exercise sessions lasted an average of 45 min including warm‐ups and cool‐downs During the last 6 weeks of the program, participants visited the physiotherapy department every fortnight for their exercise progression	OKC exercise — n = 20 (F = 11) —Age: 53.05 (10.88) years old	Participants completed a progressive program 3 times weekly for 12 weeks, with the first 6 weeks completed under supervision. Exercises included terminal knee extension, concentric quadriceps, and four‐way straight leg rise. Exercise sessions lasted an average of 45 min, including warm‐ups and cool‐downs During the last 6 weeks of the program, participants visited the physiotherapy department every fortnight for their exercise progression
Shah (2014)	—Kellgren‐Lawrence classification grade 2 or 3, AND —American College of Rheumatology Clinical Criteria, AND —Men and women aged 45–75, with an OA Hx and 12 months of difficulty with squatting and sit‐to‐stand movements, AND —Knee symptoms of a chronic duration according to the International Association for the Study of Pain	Knee OA	Patients from Orthopaedic Out‐patient Department, Civil Hospital, Ahmedabad and Physiotherapy Department, Civil Hospital, Ahmedabad	CKC exercise and muscular stretching program — n = 15 (F = not reported) —Age: Not reported	Exercises of static quadriceps (in standing), step up/down and lunge variations, partial squatting and single leg stance exercises were completed alongside a stretching program hamstring, calf, tensor fascia lata), and short‐wave diathermy. 10 repetitions for 3 sets of each exercise was completed daily for 8 weeks	OKC exercise and muscular stretching program — n = 15 (F = not reported) —age: Not reported	Exercises of static quadriceps (in sitting), terminal knee extensions, supine SLR, and hip abductor strengthening exercises were completed alongside a stretching program hamstring, calf, tensor fascia lata), and short‐wave diathermy. 10 repetitions for 3 sets of each exercise were completed daily for 8 weeks
Verma (2012)	—‘Clinical and radiographic definition of unilateral knee OA’—no further details provided, AND —50‐ to 70‐year‐old women with a diagnosis of unilateral OA, AND —Ability to walk 100 m on uneven surface	Unilateral knee OA	Not reported	CKC exercise (+ heat pack and bike warm‐up) — n = 15 (F = 15) —Age: 61.27 (5.37) years old	The exercise program consisting of seated leg press, knee bends, rowing machine exercise, step up/down, progressive jumps using a mini trampoline and stationary biking were completed 3 times weekly for 5 weeks. Repetitions and sets were not reported. Prior to exercise, participants were treated with heat packs and completed a warm‐up using stationary bikes at a rate of 20 cycles/min for 5 min	OKC exercise (+ heat pack and bike warm‐up) — n = 15 (F = 15) —Age: 60.2 (7.24) years old	The exercise program, consisting of short arc knee extension, adduction exercises, and static quadriceps setting in full knee extension, was completed 3 times weekly for 5 weeks. Repetitions and sets were not reported. Prior to exercise, participants were treated with heat packs and completed a warm‐up using stationary bikes at a rate of 20 cycles/min for 5 min

Abbreviations: m, metre; min, minute; MSE, mini squat exercise; NRS, numerical rating scale; NSAID, non–steroidal anti–inflammatory; NWB, non–weightbearing; OA, osteoarthritis; RM, repetition maximum; ROM, range of motion; SLR, single leg raise; VAS, visual analogue scale; WB, weightbearing.

3.3. Risk of Bias

Results of the ROB assessment are outlined in Table 2 with GRADE assessment outcomes in Table S2. Bennell et al. (Bennell et al. 2020) demonstrated the lowest ROB with a total score of 8, and Adegoke et al. (Adegoke et al. 2019) deemed to have the highest ROB with a score of 2. All included studies were deemed to have low ROB for the results of their between‐group statistical comparisons, and all but one study (Adegoke et al. 2019) demonstrated a low ROB associated with their random group allocation of subjects. Conversely, all included studies demonstrated a high risk of bias associated with blinding of all subjects and therapists who administered the therapy.

TABLE 2.

Results of risk of bias assessment.

Study	Eligibility	Item 2	Item 3	Item 4	Item 5	Item 6	Item 7	Item 8	Item 9	Item 10	Item 11	Total score
Adegoke et al. (2019)	Yes	No	No	Yes	No	No	No	No	No	Yes	No	2/10
Bennell et al. (2014)	Yes	Yes	Yes	Yes	No	No	Yes	No	Yes	Yes	Yes	7/10
Bennell et al. (2020)	Yes	Yes	Yes	Yes	No	No	Yes	Yes	Yes	Yes	Yes	8/10
Daşkapan et al. (2013)	Yes	Yes	Yes	No	No	No	No	Yes	No	Yes	Yes	5/10
Desai et al. (2022)	No	Yes	No	Yes	No	No	No	Yes	Yes	Yes	Yes	6/10
Gbiri et al. (2013)	Yes	Yes	No	No	No	No	No	Yes	No	Yes	Yes	4/10
Girgin et al. (2020)	No	Yes	No	Yes	No	No	No	Yes	Yes	Yes	No	5/10
Jan et al. (2009)	No	Yes	No	Yes	No	No	Yes	Yes	Yes	Yes	Yes	7/10
Moreira et al. (2021)	Yes	Yes	Yes	Yes	No	No	Yes	No	Yes	Yes	Yes	7/10
Olagbegi et al. (2016)	Yes	Yes	No	Yes	No	No	No	Yes	No	Yes	Yes	5/10
Özüdoğru and Gelecek (2023)	Yes	Yes	No	Yes	No	No	No	Yes	No	Yes	No	4/10
Shah (2014)	Yes	Yes	No	No	No	No	No	Yes	No	Yes	No	3/10
Verma (2012)	No	Yes	No	Yes	No	No	No	Yes	No	Yes	No	4/10

Note: Item definitions: 1. Specified eligibility criteria; 2. Random group allocation of subjects; 3. Concealed group allocation; 4. Groups were similar at baseline regarding the most important prognostic indicators; 5. Blinding of all subjects; 6. Blinding of all therapists who administered the therapy; 7. Blinding of all assessors who measured at least one key outcome; 8. Measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups; 9. All subjects for whom outcome measures were available received the treatment or control condition as allocated or, where this was not the case, data for at least one key outcome was analysed by ‘intention to treat’; 10. The results of between‐group statistical comparisons are reported for at least one key outcome; 11. The study provides both point measures and measures of variability for at least one key outcome.

3.4. Characteristics of Interventions

Programs varied from 4 (Desai et al. 2022) to 13 weeks (Bennell et al. 2014) duration, with participants completing exercises independently at home (Bennell et al. 2014, 2020; Daşkapan et al. ²⁰¹³) and/or under supervision (Adegoke et al. 2019; Bennell et al. ^{2014, 2020}; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Gbiri et al. ²⁰¹³; Girgin et al. ²⁰²⁰; Jan et al. ²⁰⁰⁹; Moreira et al. ²⁰²¹; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³; Shah ²⁰¹⁴; Verma ²⁰¹²) The frequency of exercise sessions within the intervention varied widely between included studies, from once daily exercise (Daşkapan et al. 2013; Desai et al. ²⁰²²; Girgin et al. ²⁰²⁰; Shah ²⁰¹⁴; Verma ²⁰¹²) to two‐to‐three exercise sessions weekly (Jan et al. 2009; Moreira et al. ²⁰²¹; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³). Two studies did not adequately report the frequency of exercise completion (Adegoke et al. 2019; Desai et al. ²⁰²²) Program duration, where reported, ranged from 30 (Gbiri et al. 2013; Girgin et al. ²⁰²⁰; Moreira et al. ²⁰²¹) to 45 (Özüdoğru and Gelecek 2023) minutes. Similarly, exercise repetitions, where reported, varied greatly; the most common dosage being 10 repetitions per set, often either dosed for 2 to 4 sets per exercise (Bennell et al. 2014, 2020; Jan et al. ²⁰⁰⁹; Moreira et al. ²⁰²¹; Shah ²⁰¹⁴) or isometric holds of 5–10 s duration (Adegoke et al. 2019; Daşkapan et al. ²⁰¹³; Gbiri et al. ²⁰¹³; Girgin et al. ²⁰²⁰; Olagbegi et al. ²⁰¹⁶). Exercise intensity and progression, where reported and applied, were commonly prescribed using variants of repetition maximum or a rating of perceived exertion (Bennell et al. 2014, 2020). Common CKC exercises included step up/down, squat variations, lunge variations, single leg stance, leg press, calf raise, hip strengthening exercise, and balance training. Common OKC exercises include knee extension and flexion, single leg raise, and hip adductor and abductor strengthening.

3.5. Outcomes

3.5.1. Pain

Pain was investigated in 11 studies, of which the results of 9 (Adegoke et al. 2019; Bennell et al. ^{2014, 2020}; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Gbiri et al. ²⁰¹³; Girgin et al. ²⁰²⁰; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³) were able to be pooled. The results of the meta‐analysis demonstrated low‐certainty evidence of a small effect size that CKC exercise was superior in improving pain compared to OKC exercise in the short‐term (SMD –0.18, 95% CI –0.37 to < 0.00; p = 0.048) (Figure 2). The results of the two studies that were unable to be pooled (Moreira et al. 2021; Shah ²⁰¹⁴) found no differences in pain outcomes between CKC and OKC exercise. There were no included studies that evaluated pain over the medium‐ or long‐term.

FIGURE 2.

Results of the meta‐analysis for pain outcomes in the short term (< 3 months). CI, confidence interval; CKC, closed kinetic chain; OKC, open kinetic chain; SMD, standardised mean difference.

3.5.2. Physical Function

Physical function was investigated in 12 studies, of which the results of 10 (Adegoke et al. 2019; Bennell et al. ^{2014, 2020}; Daşkapan et al. ²⁰¹³; Desai et al. ²⁰²²; Gbiri et al. ²⁰¹³; Girgin et al. ²⁰²⁰; Jan et al. ²⁰⁰⁹; Olagbegi et al. ²⁰¹⁶; Özüdoğru and Gelecek ²⁰²³) were able to be pooled. The results of the meta‐analysis demonstrated low‐certainty evidence of a small effect size that CKC exercise was superior in improving physical function compared to OKC exercise in the short‐term (SMD –0.24, 95% CI –0.44 to −0.03, p = 0.022) (Figure 3). The results of the two studies, unable to be pooled (Moreira et al. 2021; Shah ²⁰¹⁴), found no differences in physical function outcomes between CKC and OKC exercise. No included study evaluated physical function over the medium‐ or long‐term.

FIGURE 3.

Results of the meta‐analysis for physical function outcomes in the short term (< 3 months). CI, confidence interval; CKC, closed kinetic chain; OKC, open kinetic chain; SMD, standardised mean difference.

3.5.3. Quality of Life

QoL was investigated in 5 studies, of which the results of 4 (Desai et al. 2022; Girgin et al. ²⁰²⁰; Özüdoğru and Gelecek ²⁰²³; Verma ²⁰¹²) were able to be pooled. The results of the meta‐analysis found very low‐certainty evidence demonstrating no significant differences in the improvement of QoL in the short‐term (SMD –0.25, 95% CI –0.81 to 0.30, p = 0.370) (Figure 4). The results of the study that was unable to be pooled (Moreira et al. 2021) were consistent with this finding, demonstrating no difference in QoL outcomes between CKC and OKC exercise. No included study evaluated QoL over the medium‐ or long‐term.

FIGURE 4.

Results of the meta‐analysis for quality of life outcomes in the short term (< 3 months). CI, confidence interval; CKC, closed kinetic chain; OKC, open kinetic chain; SMD, standardised mean difference.

4. Discussion

This review compared outcomes for people with knee OA undertaking CKC or OKC exercises. Studies evaluated pain, self‐reported function, and QoL in the short‐term only, with no studies evaluating outcomes in the medium‐ and long‐term. Low‐certainty evidence found that CKC exercises were superior to OKC exercises in the short term for improving pain and self‐reported function but not QoL. The observed effect sizes were small, potentially indicating modest clinical implications at an individual level (Tagliaferri et al. 2024).

Superior outcomes for pain and physical function in people with knee OA undertaking CKC compared with OKC exercises are consistent with reports from studies of other knee‐related conditions (e.g., ACL reconstructions and patellofemoral pain) (Pamboris et al. 2024; Mustafa et al. ²⁰²²). These findings may result from the propensity of CKC exercises to facilitate dynamic stability through co‐contraction of agonistic and antagonistic muscle groups across multiple joints (Desai et al. 2022; Krupa and Dinesh ²⁰²¹; Kisner and Colby ²⁰¹²), stimulating mechanoreceptors to improve muscle strength and reduce shearing forces on the knee joint (Desai et al. 2022; Kisner and Colby ²⁰¹²; Molla et al. ²⁰¹⁷). CKC exercises also often involve movement patterns replicating functional tasks (Kwon et al. 2013), likely contributing to the observed superior physical function outcomes. This is further supported by Liu et al. (2014), who found that functional exercise produces superior outcomes in activities of daily living compared to strength training alone.

Quality of life outcomes were comparable between CKC and OKC exercise for people with knee OA, despite findings of superior outcomes for pain and physical function in those undertaking CKC exercises. Our findings suggest that improvements in pain and function alone do not appear to directly translate to superior outcomes in QoL. This may be explained by QoL being a multi‐dimensional construct, influenced by psychological, social, and environmental factors of the individual (Skevington et al. 2021). Knee OA is a chronic, progressive condition for which there is no cure; interventions are therefore palliative, aiming to optimise pain management, alleviate symptoms, and maintain functional capacity. Demoralisation from unmet unrealistic expectations of exercise outcomes, combined with past experiences of failed interventions and the increasing chronicity of OA symptoms (Gillison et al. 2009; Fiorilli et al. ²⁰²²), may also contribute to the disconnect between improved pain and/or physical function with minimal change in QoL. Notably, improvements in QoL may be influenced more by the setting than the type of intervention (i.e., CKC vs. OKC), with group‐based therapy associated with greater improvements in QoL than individualised exercise (Gillison et al. 2009). This is primarily due to group‐based therapy being associated with greater social support (Duncan and McAuley ¹⁹⁹³) and removal of barriers to solo exercise in the community (Emslie et al. 2007). Further research could evaluate the setting type (i.e. group‐based vs. individual) in conjunction with exercise type to ascertain what combinations of settings and exercises provide the greatest benefit on QoL.

Shared decision making is the cornerstone of patient‐centred care (Young et al. 2023), where patient and clinician agree upon the exercise type and modality that are evidence‐based and relevant to the patient's goals (Australian Commission on Safety and Quality in Health Care ^2024b; Young et al. ²⁰²³). Although final exercise selection may be influenced by patient preference, functional ability and setting, our findings indicate that patients seeking to reduce pain and/or improve function may benefit more from CKC exercises than OKC. Prescribed exercises should be enjoyable (Young et al. 2023), functionally relevant (Australian Commission on Safety and Quality in Health Care ^2024b), and dosed at a therapeutic level that is individualised to each patient (Australian Commission on Safety and Quality in Health Care ^2024b). To ensure that the exercise program meets its intended goal for a specific patient, ongoing and longer‐term monitoring and outcome assessment should be completed by clinicians, acknowledging that the outcomes of this review were assessed over a short period of time (≤ 3 months) and evidence was of low certainty.

All included studies, independent of the outcomes evaluated, only followed participants up in the short term (the longest follow‐up time was 13 weeks) (Bennell et al. 2014). What cannot be gleaned from the review is the comparative efficacy of CKC and OKC in the medium and long‐term. Long‐term follow‐up is imperative in evaluating interventions in health, particularly as early outcomes may give distorted treatment outcomes (Cuzick 2023). Although the risk of loss to follow‐up increases as follow‐up becomes longer (Norrie 2023), tracking participants in the longer term allows for a more pragmatic evaluation of treatment efficacy as well as its associated risks (Cuzick 2023). In addition, longer follow‐ups allow for determining whether a treatment is more or less effective in different subgroups (Cuzick 2023; Assmann et al. ²⁰⁰⁰). Future studies should evaluate the efficacy of CKC versus OKC exercise over an extended follow‐up period to understand the effect of these interventions beyond its initial short‐term efficacy.

The included studies of this review were of a varying methodological quality, with risk of bias assessment ranging from 2 to 8 of 10 on the PEDRO scale. Consistent with previous systematic reviews on rehabilitation studies (Fan et al. 2024; LeBel et al. ²⁰²⁵), no studies blinded participants to the intervention received or the therapists delivering the intervention, albeit this is an inherent and unavoidable limitation associated with rehabilitation RCTs (Armijo‐Olivo et al. ²⁰¹⁷; Nüesch et al. ²⁰⁰⁹). Notwithstanding, the absence of blinding can still result in over‐estimating treatment effectiveness (Nüesch et al. 2009), and should be considered when interpreting the results. The varying methodological quality and risk of bias were considered in the GRADE assessment (Guyatt et al. 2008) to inform the strength of evidence and recommendations formed. As a result of this variability, all meta‐analyses conducted were downgraded as the average ROB of the studies within each meta‐analysis was < 7. Future RCTs evaluating the efficacy of differing kinetic chain exercises should consider strategies to mitigate potential biases, such as group allocation concealment during randomisation, assessor blinding, and analyses that are consistent with intention to treat principals (Mansournia et al. 2017).

The limitations associated with this review, and its included studies, require acknowledgement and should be considered when interpreting the results. Firstly, the included studies were limited to those published in English due to the absence of resources for translation and the authors' expertise. It is possible that additional RCTs published in languages other than English exist and were not included in the review. Secondly, there was a variety of diagnostic criteria used across the included studies, including various radiographic definitions (e.g., Kellgren and Lawrence grade ≥ 2 or ≥ 3) (Kellgren and Lawrence 1957), the American College of Rheumatology criteria (Altman et al. 1986), and clinical assessment from Orthopaedic surgeons (Jan et al. 2009), with some studies failing to report how knee OA was defined (Gbiri et al. 2013; Girgin et al. ²⁰²⁰; Verma ²⁰¹²). This heterogeneity in criteria used for knee OA diagnosis may influence the characteristics in the sample population (Bedson and Croft 2008), contributing to the indirectness observed in the results (Salanti et al. 2014). Thirdly, participant age was not an exclusion criteria associated with the systematic review; however, the majority of participants included were of older age (mean age range ‐ 53.05 (Bennell et al. 2020) to 65.9 (Gbiri et al. 2013) years), precluding subgroup analysis based on age. The results of our review may not be generalisable to those who develop knee OA earlier in life (e.g., 35–45 year olds) who may have different occupational, domestic, and recreational loads than older individuals (Bruder et al. 2023; Louzado et al. ²⁰²¹; King et al. ²⁰²⁵). Finally, although all interventions were categorised as CKC or OKC, the rehabilitation programs varied in frequency, dosage, and load, with some studies providing insufficient reporting to clearly determine the underlying aim of the program (e.g., strength, endurance, or hypertrophy). As a result, subgroup analysis and the potential mediating effects of these key exercise variables on the observed outcomes remain unclear and could not be investigated. This variability contributes to indirectness within our GRADE assessment and highlights the need for more detailed and standardised reporting of exercise prescription in future trials.

5. Conclusion

In conclusion, the review found low‐certainty evidence that CKC exercise demonstrated more efficacious outcomes on pain and physical function compared with OKC exercise in people with knee OA in the short term. Very low‐certainty evidence indicated that CKC and OKC exercises had comparable effects on QoL in the short term. Our findings indicate that patients who seek to reduce pain and/or improve function may derive greater benefit from CKC exercises than OKC in the short term. However, the review is not without its limitations; the included studies only evaluated the effects of CKC versus OKC exercise in older populations with short‐term follow‐up (≤ 3 months). Future studies should evaluate the efficacy of CKC versus OKC exercise in younger adults (i.e. 40 years old) with knee OA, and with long‐term follow‐up periods (> 6 months), to understand the effect of these interventions beyond their initial short‐term efficacy.

6. Implications for Physiotherapy Practice

This systematic review highlights that patients seeking to reduce pain and/or improve function may benefit more from CKC exercises than OKC. When the aim of rehabilitation is to reduce pain and/or improve function in knee OA, CKC exercises should be the first‐line treatment option, at least in the short term. Such CKC exercise could include (but not limited to) step up/down, squat variations, lunge variations, leg presses, and calf raises. Physiotherapists can use this information to inform their exercise selection for rehabilitation programs and tailor education to patients as to why CKC exercises are selected.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Permission to Reproduce Material From Other Sources

The authors have nothing to report.

Supporting information

Figure S1: Publication bias ‐ Function.

Table S1: Medline search strategy.

Table S2: Results of the GRADE Assessment.

Michael, Harrison J. , Pham Harrison D., Li Lok S. K., et al. 2025. “The Efficacy of Closed Versus Open Kinetic Chain Exercise in Knee Osteoarthritis—A Systematic Review and Meta‐Analysis.” Physiotherapy Research International: e70105. 10.1002/pri.70105.

Data Availability Statement

Data is available upon reasonable request to the corresponding author, subject to the execution of a data sharing agreement.