Osteoarthritis (OA) is a chronic, degenerative condition characterized by multifactorial etiology resulting in subchondral sclerosis, cartilage loss, and osteophyte formation [1]. This leads to pain and uneven loading between joint compartments [2] and the knees are the most affected anatomical sites [3]. Individuals with knee osteoarthritis (KOA) often adopt compensatory mechanisms to alleviate pain, which can manifest as reduced limb movements and a limited range of motion (ROM) in the knee joint, which contributes to a gradual loss of function and significant disabilities [4,5]. It is reported that the overall prevalence of KOA among people aged 50 and older in South Korea is 35.1% [6] and the average medical costs for both in-patients and outpatient care for KOA amount to 240 billion KRW [7].

As OA is an incurable and irreversible condition, its ultimate management goal is to maintain or improve functionality and quality of life [8]. Management of OA is divided into non-operational and operational methods. Non-operational options include patient education, lifestyle management, oral analgesics, intra-articular injections, exercise, and physical therapy [9]. According to the existing guidelines, supervised or unsupervised exercises are strongly recommended over no exercise to improve pain and function for treatment of KOA, although the particular types of exercise have not been specified [9-11].

Generally, cardiorespiratory-, resistance-, flexibility/stretching-, and/or neuromotor exercises are recommended in patients with KOA, not different from general population [10]. Among these, the beneficial effects of cardiorespiratory and resistance exercise in KOA have been established by previous research [12].

Stretching exercises are a form of flexibility training, contributing to the reduction of stiffness and an increase in joint ROM [13-15]. Furthermore, it is a simple, easily accessible and safe method with relatively low cost. As patients with KOA typically experience loss of joint motion [4], often exacerbated by avoidance of using an involved joint due to pain, stretching exercise can alleviate and slow down these conditions [10]. However, due to relatively small studies and variability of primary outcomes and interventions, the effects of stretching exercises on KOA have not been conclusively demonstrated. In addition, they are often incorporated in combination with other modalities or exercises. Recent systematic reviews revealed that various stretching exercises can effectively alleviate pain, stiffness or physical functional limitation in patients with KOA [16,17]. However, these reviews have not established the isolated effects of stretching, due to the inclusion of studies where stretching was part of other exercises or where the control group participated in different forms of exercise [16,17].

Therefore, this study purposes to provide a comprehensive and updated review of the studies with randomized controlled trials (RCTs) investigating the effects of stretching exercises in patients with KOA. Furthermore, we aim to synthesize the isolated effects of stretching exercises on pain and ROM through meta-analysis.

According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [18], a systematic search was conducted to investigate the effects of stretching exercises on KOA.

1. Literature search strategy

We systematically searched electronic databases including PubMed, Web of Science, Cochrane Library, and Embase for all papers published until October 2023. The search keywords were “knee osteoarthritis” and “stretching exercise.” We did not restrict the outcome variables for inclusion of the studies for comprehensive review of stretching exercises and KOA. Additionally, we performed a review of literature and examined reference lists of relevant articles to supplement the search.

2. Eligibility criteria

1) Inclusion criteria for systematic review

(1) Studies involve participants with KOA

(2) Interventions consist of stretching exercises

(3) Control groups do not involve stretching exercises

(4) Study designs are RCTs

2) Exclusion criteria for systematic review

(1) Animal studies

(2) Where both the intervention and control groups involve stretching exercises

(3) Focused on post-operative interventions for KOA patients

(4) Written in non-English language

(5) With unavailable data extraction

3. Risk of bias and quality assessment

The included papers underwent a risk of bias assessment by two independent reviewers using the tool outlined in the “Cochrane Handbook for Systematic Reviews of Interventions,” developed by the Cochrane Collaboration [19]. This tool comprises two sections and covers seven specific domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other sources of bias. In cases of disagreement, consensus was reached through discussion, and if necessary, a third reviewer was consulted for resolution.

4. Data extraction and synthesis

Extracted data included publication date, author information, study design, inclusion and exclusion criteria, number of participants and demographic data, interventions in each group, duration of follow-up, and outcomes of interventions. In trials with multiple treatment groups, each treatment group was compared with the control group independently. Two reviewers independently used a standardized data extraction form to retrieve relevant data from the articles and discrepancies were resolved through discussion until a consensus was reached.

We classified the studies according to the types of intervention and control groups: 1) Stretching exercises vs. no exercise group, 2) Stretching and another type of exercises vs. another type of exercise, 3) Stretching exercises vs. another type of exercise/therapy, 4) Stretching exercises as part of various exercises vs. no exercise or another type of exercise/therapy. Furthermore, we specifically considered the studies falling into categories 1 and 2 from above classification as studies involving in isolated stretching exercises.

5. Statistical analysis

Main results of each variable of individual articles were presents as effect size (ES), calculated mean values and a pooled standard deviation by Cohen [20]. We synthesized the isolated effects of stretching exercises (the studies in category 1 and 2) of outcome variables of pain and ROM, using standardized mean differences (SMD) with 95% confidence intervals (CI). Outcome measures were computed as the mean change before and after intervention, with P<0.05 denoting significant difference. In addition, we assessed heterogeneity based on Higgins I²-statistics [21]. A significance level of P<0.05 and I²>50% was considered indicative of significant heterogeneity, warranting the use of a random-effects model; otherwise, a fixed-effects model was employed. Synthesized effects for this meta-analysis are presented as forest plots. For publication bias, the asymmetry of the effect size was visually assessed using a funnel plot. This meta-analysis was conducted using Review Manager statistical software version 5.3 (Cochrane collaboration, Copenhagen, Denmark).

1. Study selection

A flow chart according to PRISMA guideline [18] for study selection is shown in Figure 1. Initially, 500 articles were identified from the preliminary search across four electronic databases (PubMed: 28, Web of Science: 3, EMBASE: 266, Cochrane Library: 203) and 6 articles were included by manual searches of reference lists in selected articles and reference reviews. Following deduplication and title/abstract screening, 89 records underwent full-text review, resulting in the assessment of 104 articles. Finally, 19 RTCs were included in the systematic review. In addition, 6 articles (comprising 7 trials due to multiple treatment groups) with intervention involving isolated stretching exercises were included in the quantitative synthesis of meta-analysis.

Figure 1. Flow chart of the selection process for studies, adopted from PRISMA guidelines

2. Risk of bias and quality assessment

According to the intervention system Cochrane Handbook for Systematic Reviews, an assessment was conducted on seven aspects related to bias risks in randomized controlled trials [19] (Figure 2). All studies were randomized, with the most common shortcomings being a lack of description regarding allocation concealment, incomplete information about randomization procedures, and inadequate blinding of participants and personnel.

Figure 2. Risk of bias. (A) Risk of bias summary: review authors’ judgements about each risk of bias item for each included study. (B) Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies

3. Study characteristics

Nineteen eligible studies were included in this systematic review and a detailed overview of the characteristics of the studies is described in Table 1.

Table 1 . Characteristics of the studies included in this systematic review.

Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
I. Stretching exercise vs. no exercise
Aoki et al. [8] (2009)	67 to 82 years All females	Diagnosed with severe KOA by radiographs	Home based self-stretching (n=17) Stretching method: seated quadriceps stretching and prone rectus femoris stretching	No intervention (n=17)	Pain (VAS) [0.20*] Supine-ROM [0.20*] Gait-ROM [0.11*] Gait-speed [0.23*]
Gao et al. [26] (2023)	≥65 years Male: 16 Female: 11	Diagnosed with unilateral or bilateral KOA based on the ACR clinical criteria; grade≥ 2 based on the K-L classification; normal or corrected-to-normal vision; ability to step over obstacles without assistance	PNF stretching (n=13) Stretching method: diagonal and spiral patterns and stretching techniques (contract-relax, hold-relax, reversal of antagonists and repeated stretch)	No intervention (health lecture) (n=14)	Pain (VAS) [0.26*] Step length [0] Crossing velocity [0.32*] Foot clearance [0]
Song et al. [36] (2020)	≥65 years Male: 11 Female: 18	Diagnosed with medial KOA in one or both legs	PNF stretching (n=13) Stretching method: flexion-abduction-internal, extension-adduction-external, flexion-adduction-external and extension-abduction-internal rotations	No intervention (n=16)	Pain (WOMAC) [0.83*] Hip extension ROM [0.51*] Hip flexion ROM [0.73*] Knee flexion ROM [0.63*] Ankle plantar flexion ROM [0.53*] Ankle dorsal flexion ROM [0.41*]
Kannas et al. [34] (2023)	≥50 years	Diagnosed with KOA based on the K-L radiographic evaluation criteria	Passive stretching (n=10) Stretching method: passive stretching in an extended prone position, lumbar L2-L4, maintaining lordosis, immobilizing pelvis for full joint flexion	No intervention (n=10)	Knee ROM [0.43*] Mean torque [−0.36] Maximum torque [0.58*] Max torque angle [0.62*]
Ahmed [37] (2010)	45 to 70 years	Diagnosed with severe KOA scheduled to have primary total knee arthroplasty	Self-stretching (n=10) Stretching method: seated quadriceps stretching (knee flexion assisted by hand, or by the opposite leg while the patient in a prone position); supine hamstring stretching (fully extended knee and lifted foot with a towel around the foot, in a supine position)	No intervention (n=10)	Pain (VAS) [0.98*] ROM [0.47*] Pain (WOMAC) [0.82*] Stiffness (WOMAC) [0.74*] Physical function (WOMAC) [0.66*] Total WOMAC [0.72*]
II. Stretching exercise and another type of exercise vs. another type of exercise
Weng et al. [38] (2009)	46 to 78 years Male: 26 Female: 106	Diagnosed with moderate KOA based on the Altman grade II	Static stretching and isokinetic exercises (n=33) Stretching method: active-assistive stretching on the quadriceps and biceps femoris in a supine position	Isokinetic exercises (n=33); no intervention control (n=33)	Pain (VAS) [0.60*]† ROM [0.31*]† Lequesne’s index [0.74*]†
Weng et al. [38] (2009)	46 to 78 years Male: 26 Female: 106	Diagnosed with moderate KOA based on the Altman grade II	PNF stretching and isokinetic exercises (n=33) Stretching method: stretching on the quadriceps and biceps femoris; hold-relax, contract-relax, contract-relax agonist contract and hold-relax agonist contract	Isokinetic exercises (n=33); no intervention control (n=33)	Pain (VAS) [0.73*]† ROM [0.61*]† Lequesne index [0.79*]†
Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
III. Stretching exercise vs. another type of exercise or therapy
Zhang et al. [27] (2022)	45 to 65 years Male: 13 Female: 37	Diagnosed with KOA based on the criteria in the “Guidelines for the Diagnosis and Treatment of Osteoarthritis 2018 Edition.” grade 1 or 2 based on the K-L classification; body mass index ≤36 kg/m2; knee pain with VAS≥2 (0-10 points)	Stretching training exercise (n=25) Stretching method: 5 items of quadriceps training and neuromuscular training	YiJinJing (Qigong) exercise (n=25); 5 items compiled by the State Sports General Administration of China	Pain (VAS) [0.23*] Pain (WOMAC) [−0.21] Stiffness (WOMAC) [−0.33] Physical function (WOMAC) [0.11*] Mental component summary [−0.59] Physical component summary [0.55*] Beck depression inventory [0.79*] Perceived stress scale [0.93*] Berg balance scale [−0.18] Stride velocity [−0.54] Stride length [0] Stance phase time [0.07*] Swing phase time [0.15*]
Nafees et al. [23] (2023)	≥40 years Male: 11 Female: 37	Diagnosed with KOA based on the ACR clinical and radiological criteria	PNF stretching and isometric quadriceps strengthening exercises (n=25) Stretching method: hold-relax on tight hamstring muscle in a supine lying position	Dynamic soft tissue mobilization technique and isometric quadriceps strengthening exercises (n=25)	Pain (VAS) [0] Right AKET [0.12*] Light AKET [0.15*] Symptoms (KOOS) [−0.06] Pain (KOOS) [0.19*] ADLs (KOOS) [0.15*] Sports and recreational (KOOS) [0.11*] Quality of life (KOOS) [0.02*]
Masekar et al. [29] (2020)	40 to 60 years Male: 10 Female:26	Radiologically and clinically diagnosed with KOA by certified orthopedic surgeons or physiotherapists; grade 2 or 3 based on the K-L classification	PNF stretching and other physical therapies (n= 18) Stretching method: in a supine position with hip flexed, hamstring stretch with knee extension performed by a therapist	Muscle energy technique and other physical therapies (n=18)	Pain (NPRS) [0.97*] Function (WOMAC total) [0.88*] AKET [0.35*]
IV. Stretching exercise as part of various exercises vs. no exercise or another type of exercise/therapy
Suzuki et al. [31] (2018)	50 to 70 years Male: 23 Female: 29	Pre-radiographic KOA (grade 0 or 1 based on the K-L classification); ability to walk independently on a flat surface without an ambulatory assistive device Bilateral KOA cases were not considered separately from unilateral cases	10 programs of multiple exercises including hamstring and quadricep stretching (n=28) Stretching method: hamstring stretching in a chair-sitting position with one knee extended, followed by the upper body tilting forwarded; quadriceps stretching in a side-lying position with knees flexed, followed by pulling one foot behind the body Other exercises: multiple exercises involving training the knee and hip muscles	Single program of quadriceps exercises (n=24)	Pain (VAS) [0.15*] Pain and stiffness (JKOM) [0.14*] ADLs (JKOM) [0.08*] Participation in social activities (JKOM) [0.16*] General health (JKOM) [−0.23] Total (JKOM) [0.15*] Strength of knee extension [0.05*]
Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
Mazloum et al. [22] (2018)	≥40 years Male:28 Female:13	Diagnosed with KOA based on the ACR clinical and radiological criteria; knee pain on most days of the previous month (mean 4); osteophyte in radiography	Conventional therapeutic exercise including isometric strengthening and stretching (n=14) Stretching method: in a supine position, knee flexion to straighten the soles, followed by lifting non-healthy limb with full knee extension, using a Thera-band on foot for ankle dorsiflexion	No intervention (n=13) Pilates training exercise (n=14)	Compared to no intervention controlLequesne index [0.35*] The time required to complete four functional activities [0.72*] Joint position sense [0.86*] Compared to Pilates training controlLequesne index [−0.18] The time required to complete four functional activities [0.09*] Joint position sense [−0.29]
Assar et al. [35] (2020)	≥40 years All females	Diagnosed with KOA based on the ACR clinical criteria; ≥grade 2 based on the K-L classification; self-reporting knee instability	Water based strengthening and aerobic exercises, step and proprioceptive exercises and core training; static stretching as warm up and cool down (n=12) Warming up stretching exercise included quadriceps, hamstrings, triceps surae, abductors and adductors of hip and gluteal muscles	No intervention (n=12) Total resistance exercise (n=12)	Compared to no intervention control Knee instability [0.78*] Stiffness (WOMAC) [0.55*] Pain (VAS) [0.57*] Berg Balance Scale [0.78*] Quadriceps strength [0.47*] Knee flexion ROM [0.88*] Compared to total resistance exercise control Knee instability [−0.39] Stiffness (WOMAC) [0.03*] Pain (VAS) [0.08*] Berg Balance Scale [0.33*] Quadriceps strength [−0.07] Knee flexion ROM [−0.48]
Nahayatbin et al. [28] (2018)	45 to 65 years	Diagnosed with KOA, grade 2 or 3 based on the K-L classification 3; ≥grade 3 of lower limb muscles strength based on the Oxford scale	Closed kinetic chain exercise with routine physical therapy; static stretching exercise was included as warm-up (n=16)	Tai Chi exercise with routine physical therapy (n=16) Routine physical therapy (n=16)	Compared to Tai Chi exercise with routine physical therapy control Symptoms (KOOS) [−0.5] Pain (KOOS) [−0.41] ADLs (KOOS) [−0.38] Sport (KOOS) [−0.31] Quality of life (KOOS) [0.04*] Total score (KOOS) [−0.49] 6MWT [−0.19] Compared to routine physical therapy control Symptoms (KOOS) [0.64*] Pain (KOOS) [0.51*] ADLs (KOOS) [0.26*] Sport (KOOS) [0.40*] Quality of life (KOOS) [0.57*] Total score (KOOS) [0.60] 6MWT [0.16*]
Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
Arslan and Kul [30] (2022)	40 to 70 years Male: 6 Female: 45	Diagnosed with primary KOA based on the ACR’s clinical and radiologic criteria; grade 2 or 3 based on the K-L classification	Transcutaneous electrical nerve stimulation combined with therapeutic home exercise program including knee stretching (n=25)	Radial extracorporeal shockwave treatment combined with therapeutic home exercise program including knee stretching (n=26)	Pain (VAS) [0.20*] Pain (WOMAC) [0.23*] Stiffness (WOMAC) [0.19*] Physical function (WOMAC) [0.05*] Total (WOMAC) [−0.03] Right knee Flexion-active [0.04*] Right knee flexion-passive [0] Left knee flexion-active [0.04*] Left knee flexion-passive [−0.06] TUG [0.12*]
Lai et al. [24] (2018)	50 to 70 years Male: 2 Female: 32	Diagnosed with KOA by an orthopedic surgeon based on the ACR clinical and radiologic criteria	Strength exercise including squat training and stretching (n=18) Stretching exercise was included as warm-up and cool-down covered the joint motion and muscle stretch.	No intervention (n=16)	Plantarflexion of ankle [0.003] Dorsiflexion of ankle [0.002] Varus of ankle [0.017] Valgus of ankle [0.004] Flexion of knee [0.124*] Extension of knee [0.004]
Ha et al. [39] (2018)	Middle-aged All females	Diagnosed with KOA; received a doctor's consent to participate in the exercise program; did not participate in regular exercise or other exercise programs for the past 6 months	Aquatic exercise (n=9) including stretching exercise as warm up	No intervention (n=8)	Weight [0.12*] Body mass index [0.007*] Body fat [0.012*] Waist circumference [0.13*] Systolic blood pressure [0.16*] Diastolic blood pressure [0] Fasting blood glucose [0.27*] Hemoglobin A1C [0.27*] Triglyceride [0.06*] High density lipoprotein cholesterol [0.05*] C-reactive protein [0] VO2 max [0.34*] Right knee flexion peak torque/body weight [0.32*] Left knee flexion peak torque/body weight [0.27*] Right knee flexion peak torque/body weight [0.12*] Left knee flexion peak torque/body weight [0.12*] Pain (WOMAC) [0.27*] Stiffness (WOMAC) [0.37*] Physical function (WOMAC) [0.18*] Total (WOMAC) [0.23*]
Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
da Silva et al. [25] (2015)	≥18 years or above Male: 4 Female: 26	Diagnosed with chronic KOA based on the ACR clinical criteria; moderate to very severe knee pain according to the Lequesne algofunctional index; knee pain on most days at least 3 months within a year	Group rehabilitation program including strengthening, function task-oriented and balance exercise (n=15); stretching exercise was included as warm up Stretching method: active stretching of quadriceps, hamstrings, and calf	No intervention (n=15)	Pain (Lequesne index) [0.41*] Distance (Lequesne index) [0.41*] Function (Lequesne index) [0.10*] Total score (Lequesne index) [0.34*] Physical functioning (SF-36) [0.38*] Physical role (SF-36) [0.64*] Pain (SF-36) [0.39*] General health (SF-36) [0.35*] Vitality (SF-36) [0.40*] Social function (SF-36) [0.03*] Emotional role (SF-36) [0.50*] Mental health (SF-36) [0.33*] Chairstand test [0.43*] SitandReach test [0.28*] TUG [0.6*] 6MWT [0.38*]
Oliveira et al. [32] (2012)	50 to 75 years Male: 6 Female: 92	Diagnosed with KOA based on the ACR criteria; ≥grade 2 of the K-L classification	Femoral quadriceps strengthening exercise including stretching as warm up (n=50) Stretching method: hamstring muscles stretching with an elastic band in a seated position	No intervention (n=50)	TUG [0.32*] Lequesne Index [0.21*] Pain (WOMAC) [0.09*] Stiffness (WOMAC) [0.27*] Function (WOMAC) [0.18*]
Braghin et al. [33] (2018)	45 to 75 years Male: 11 Female: 31	Diagnosed with KOA based on the K-L classification grade 1 to 3	Symptomatic KOA (n=15); asymptomatic KOA (n=11)	No intervention (n=16)	Symptomatic vs. control Total (WOMAC) [0.26*] Pain (WOMAC) [0.34*] Stiffness (WOMAC) [0.22*] Function (WOMAC) 0.21*] Step up/over: Liftup index, left [0.20*] Liftup index, right [0.12*] Movement time, left [0.21*] Movement time, right [0.21*] Impact index, left [0.01*] Impact index, right [0.18*] Semistatic balance: Eyesopen stable surface [0.27*] Eyesclosed stable surface [0.15*] Eyesopen unstable surface [0.12*] Eyesclosed unstable surface: [0.25*]
Author (year)	Participants (age，sex)	Inclusion criteria of participants	Intervention		Outcome variables (ES of relevant outcomes)
			Experimental group (n)	Control group (n)
			Physical exercises of strengthening, aerobic exercise and functional training including stretching as warm up		Asymptomatic vs. control Total (WOMAC) [0.39*] Pain (WOMAC) [0.52*] Stiffness (WOMAC) [0.22*] Function (WOMAC) [0.35*] Step up/over: Liftup index left [0.36*] Liftup index right [0.27*] Movement time–left [0.21*] Movement time–right [0.29*] Impact index left [0.17*] Impact index right [0.09*] Semistatic balance: Eyes-open stable surface [0.28*] Eyes-closed stable surface [0.03*] Eyes-open unstable surface [0.06*] Eyes-closed unstable surface [0.12*]

*Favoring experimental group, ^†compared to isometric strengthening group. ES, effect size; KOA, knee osteoarthritis; ACR, American college of rheumatology; VAS, visual analogue scale; K-L, Kellgren and Lawrence; JKOM, Japanese Knee Osteoarthritis Measure; PNF, proprioceptive neuromuscular facilitation; ROM, range of motion; WOMAC, Western Ontario and McMaster Universities Arthritis Index; AKET, active knee extension test; KOOS, Knee Injury and Osteoarthritis Outcome Score; ADL, activities of daily living; NPRS, Numeric Pain Rating Scale; SF-36, 36-Item Short Form Survey; TUG, Timed Up and Go Test; 6MWT, 6-Minute Walk Test..

1) Participants

A total of 734 participants with KOA involved in the studies, sample size ranged from 17 to 100. The participants’ age ranged from 18 to 78, although the majority of the studies included individuals aged 40 and above. Patients participating in these studies were usually diagnosed with KOA mainly according to the American College of Rheumatology clinical criteria [22-25] or the Kellgren-Lawrence radiographic criteria [26-35] for disease staging. In some studies [8,36-39], KOA was diagnosed based on symptoms and imaging findings, without specifying precise diagnostic criteria (Table 1).

2) Intervention

We categorized the studies according to their types of intervention vs. control groups as follows: 1) Stretching exercises vs. no exercise group (n=5) [8,26,34,36,37], 2) Stretching and another type of exercises vs. another types of exercises (n=1) [38], 3) Stretching exercises vs. another type of exercises/therapy (n=3) [23,27,29], 4) Stretching exercises as part of various exercises vs. no exercise or another type of exercise/therapy (n=10) [22,24,25,28,30-33,35,39] (Table 1).

For studies involved in stretching exercise with no exercise control group, self-stretching [8,37] and proprioceptive neuromuscular facilitation (PNF) stretching [26,36] were used as interventions. For studies in stretching and another type of exercises compared with another types of exercises control group [38], static stretching combined with isokinetic exercises group and PNF stretching combined with isokinetic exercise group were compared with isokinetic exercise control group (Table 1).

3) Outcomes

The main outcome variables in 19 individual studies were diverse: pain, ROM, function, activities of daily living, quality of life, gait etc. Among these, the Visual Analog Scale (VAS) [8,23,26,27,30,31,35,37,38] and Western Ontario and McMaster Universities Osteoarthritis (WOMAC) index [27,30,32,36,37,39] serve as the most commonly used pain indicators. Measurement of ROM primarily focuses on the measurement of knee joint flexion angle in most studies [8,24,27,35-37], although extension ROM [24] and active knee extension test [23,29] were measured in small studies. Functional status assessment primarily includes the WOMAC index, the Knee injury and Osteoarthritis Outcome Score (KOOS), and the Lequesne’s index. In gait analysis, various parameters were employed, including gait speed [8] and crossing velocity [26], as well as stride velocity, length and stance and swing phase time [27]. In addition, functional gait performance was assessed by the 6-minute Walk Test [25,28] and the Time Up-and-Go Test [25,30,32].

4. Effects of isolated stretching exercise by meta-analysis

To synthesize the effect of isolated stretching exercise on KOA, we specifically considered studies involving in isolated stretching exercises in categories 1 and 2 from above classification (6 studies and 7 trials). As the outcome variables used in these studies were mainly on pain and ROM, we synthesized the effect of those variables by meta-analysis. Figure 3 shows the forest plots of the analysis.

Figure 3. Effect of isolated stretching exercises on knee joint pain (A) and range of motion (B). CI, confidence interval; SD, standard deviation

1) Pain

Five studies [8,26,36-38], reported in six trials, indicated a significant improvement in pain among patients with KOA following stretching interventions (Figure 3A) (SMD −0.74; 95% CI −0.96 to −0.52; I²=72%; P<0.00001; n=356 combined sample size).

2) Knee joint flexion ROM

Four studies [8,34,36,38] reported in five trials demonstrated a significant improvement in knee flexion ROM among patients with KOA (Figure 3B) (SMD 0.60; 95% CI 0.38 to 0.83; I²=57%; P<0.00001; n=333 combined sample size).

5. Publication bias

As shown in Figure 4, large-sample effect sizes typically appear at the top of the funnel plot, clustering around the average effect size, indicating a less severe publication bias. Due to increased sampling bias, small-sample effect sizes tend to cluster towards the bottom of the funnel plot. However, the test power of these funnel plots is lower since the number of studies is less than 10.

Figure 4. Funnel plots of standard errors for knee pain (A) and range of motion (B) in the meta-analysis. SE, standard error; SMD, standardized mean difference

This study comprehensively reviewed the studies on stretching exercise interventions involved in patients with KOA. Stretching exercises were used not only as independent interventions but also as part of and/or combination with other exercise regimens. In addition, there were a limited number of studies focusing solely on isolated stretching exercises, revealing the significant improvement of pain and knee joint ROM by meta-analysis.

Stretching, in a broader sense, extends beyond muscle stretching, involving ligaments, joint capsules, and more [8]. Patients with KOA usually intend to minimize pain triggers, which consequently restricts knee joint movement, impairs walking abilities, and results in loss of functionality [4,40]. However, full ROM in the knee joint is crucial for optimal functionality [41] as well as preoperative knee joint mobility determines postoperative mobility and outcomes in total knee arthroplasty [42]. Studies have indicated that increasing joint flexibility and multi-muscle strength are more crucial than solely enhancing quadriceps strength [43-45]. Thus, whether considering surgery or opting for conservative treatment in patients with KOA, knee joint ROM remains a critical factor. Stretching exercises can address limited ROM arising from the joint, muscles, ligaments, and joint capsules, thereby improving knee joint flexibility [46]. Incorporating regular stretching exercises in knee joint treatment is advisable [38], with a treatment duration of no less than 3 weeks [13]. Therefore, the American College of Sports Medicine recommend daily stretching exercises for patients with arthritis, emphasizing a more frequent routine compared to the typical recommendation for healthy adults, which is generally two or three days per week [10].

According to our systematic review, stretching exercises primarily involved static and PNF stretching, focusing on the quadriceps, biceps femoris, and hamstring muscles. The intervention duration ranged from 12 weeks to 3 months. These exercises were conducted solely or combination with other exercises, in supervised or unsupervised setting. In a supervised setting, experienced therapists typically guided and instructed patients through complex training routines, often involving passive stretching exercises. In an unsupervised environment, individuals usually engaged in self-training within their own homes, performing simpler exercises and self-stretching routines. According to the clinical management guidelines of KOA, both supervised and unsupervised exercises are recommended, as no established benefit has been identified for supervised over unsupervised exercises [1].

In studies encompassing stretching exercises within their scope, a diverse range of outcome variables is utilized. Most studies assessed pain and function in patients with KOA, with VAS and WOMAC index being the most frequently used tools. This usage of variables aligns with the findings of existing systematic reviews on KOA exercise therapies [12]. The WOMAC index is a widely used outcome assessment tool for patients with OA, consisting of 24 self-administrated questions covering three domains of pain, stiffness and physical function [47]. Another commonly used tool was the KOOS scale, which includes five subscales of pain, symptoms, activities of daily living, sports and recreational function, and knee-related quality of life [48]. The Lequesne’s index, comprising 11 questions about pain, discomfort and function in patients with knee and hip OA [49] was also used in various studies.

In addition, studies typically included the measurement of ROM, as the primary impact of stretching exercises is on flexibility. This measurement was predominantly focused on knee joint flexion ROM, most affected by KOA. In addition, small number of studies measured using the active knee extension test, measuring hamstring tightness and extension ROM [23,29].

In contrast to studies assessing the effects of other types of exercises on KOA [12], research on the effect of stretching exercises is limited in measuring variables on strength [28,34,35], as stretching exercises are presumed to have minimal effects on strength. Previous systematic review revealed static stretching showed no significant effect on isometric strength, and there was even a negative effect stretching before resistance training [50].

This study specifically selected the articles that focused on the isolated stretching intervention to synthesize the effect of exercise. Therefore, we selected the studies in which the exercise regimen was compared with and without stretching exercises. The interventions in these studies were either stretching exercises compared to no exercise control group (Category 1) or stretching combined with another type of exercise compared to another type of exercise (Category 2). This emphasizes how stretching can effectively contribute to improving the condition of patients with KOA. Isolated stretching interventions exhibited a significant improvement in knee joint pain and ROM by meta-analysis, consistent with other previous study [17]. In addition, Weng’s study showed that combination of stretching was more effective than isokinetic exercise alone in functional improvement in individuals with KOA, specifically PNF stretching over static stretching [38].

There were studies comparing stretching exercises to another types of exercises, classified into Category 3 in this study. In this category, stretching exercises with/without combined exercises or therapies were compared with Qi Gong [27], dynamic soft tissue mobilization therapy (DSMT) [23], and muscle energy technique (MET) [29]. The comparison with Qi Gong [27] yielded conflicting results in pain levels, depending on assessment scales by WOMAC and VAS measurements, which should be proved in further trials However, the PNF stretching combined with isometric strengthening exercise or physical therapies exhibited superior therapeutic effects compared to DSMT [23] or MET [29].

This systematic review reveals that various combined exercise/therapy regimens (e.g., isometric strengthening, aquatic training, closed kinetic chain exercise, transcutaneous electrical nerve stimulation, etc.), including stretching exercises were integrated as parts of regimens or warm-up and cool-down routines, appear feasible for alleviating knee joint pain in KOA [22,24,25,28,30-33,35,39] . These regimens might reflect real world practical settings and suggest that the inclusion of stretching exercises in clinical settings can be more expansive. However, the intervention effects observed in these studies cannot be considered solely as the effects of stretching exercises themselves.

In addition, Luan’s research suggests that combining stretching exercises with other forms of exercise might not be helpful in managing pain for KOA patients [17]. There could be loading and pressure leading to discomfort during squatting and leg stretching exercises [51]. Furthermore, quadriceps contraction during lower limb strengthening programs might increase pressure and load on the patellofemoral joint cartilage [52-54], potentially triggering symptoms of KOA [55]. Therefore, individualized exercises with meticulous care should be applied to the patients with advanced KOA.

As stretching exercises can reduce muscle tension, enhancing the range of motion in the knee joint [46], this could potentially result in an increased space within the knee joint cavity, indirectly reducing pressure on the knee joint. However, the long-term effects of whether stretching can alter patellofemoral joint pressure remains uncertain and requires further research.

The limitations of this study are as follows: (1) Small sample sizes in synthesizing the effect of stretching exercises due to emphasis on highlighting isolated effects; (2) Detection of publication bias with too low power to distinguish change from true asymmetry due to small number of included articles; (3) Differences in the quality of included studies due to limited selection of articles; (4) No established long-term effects of stretching exercises such as modification of patellofemoral joint pressure; and (5) No establishment of the optimal stretching regimens in patients with KOA. Despite the limitation, our study conducted an in-depth review of studies on stretching exercises in patients with KOA and addressed their synthesized isolated effects in a novel manner.

In conclusion, stretching exercises can be implemented as individually or as part of combined regimens. Additionally, isolated stretching interventions can alleviate pain and improve flexion ROM in individuals with KOA, laying the groundwork for further treatment. More studies involved isolated stretching exercises with large sample sizes are needed to establish the long-term effects and the optimal stretching exercises regimen for patients with KOA.

This work was supported by the Pukyong National University Research Fund in 2023 (202315360001).

No potential conflict of interest relevant to this article was reported.