Stretching, Bracing, and Devices for the Treatment of Osteoarthritis-Associated Joint Contractures in Nonoperated Joints: A Systematic Review and Meta-Analysis

Abstract

Context:

Many patients with osteoarthritis (OA) develop range of motion (ROM) restrictions in their affected joints (contractures), associated with worse outcomes and rising healthcare costs. Effective treatment guidance for lost ROM in OA-affected joints is lacking.

Objective:

A systematic review and meta-analysis evaluating the effectiveness of stretching and/or bracing protocols on native (nonoperated) joint ROM in the setting of radiographically diagnosed OA.

Data Sources:

Seven databases, English-language.

Study Selection:

Studies including participants with radiographically diagnosed OA in any native joint evaluating the effect of stretching or bracing on ROM.

Study Design:

Systematic review and meta-analysis.

Level of Evidence:

Level 2.

Data Extraction:

Two reviewers independently screened articles for inclusion and assessed risk of bias in included trials. Primary outcomes were ROM, pain, and adverse events (AEs).

Results:

We identified 6284 articles. A total of 9 randomized controlled trials, all evaluating the knee, met eligibility criteria. For stretching, 3 pooled studies reported total ROM, which improved by mean difference (MD) of 9.3° (95% CI 5.0°,13.5°) versus controls. Two pooled studies showed improved knee flexion ROM (MD 10.8° [7.3°,14.2°]) versus controls. Five studies were pooled for knee extension with mean improvement 9.1° [3.4°,14.8°] versus controls. Seven pooled studies showed reduced pain (standardized MD 1.9 [1.2,2.6]). One study reported improved knee extension of 3.7° [2.9°,4.5°] with use of a device. No studies used orthoses. One study reported on AEs, with none noted. Performance bias was present in all included studies, and only 3 studies clearly reported blinding of outcome assessors. Strength of evidence for primary outcomes was considered moderate.

Conclusion:

There was moderate-quality evidence that stretching is an effective strategy for improving knee total, flexion and extension ROM, and pain. Our findings suggest that stretching to regain joint ROM in OA is not futile and that stretching appears to be an appropriate conservative intervention to improve patient outcomes as part of a comprehensive knee OA treatment plan before arthroplasty.

Osteoarthritis (OA) is the most common arthritis and a major cause of disability globally.^{53,56,68,77,79} Those who are engaged in impact sports or who have suffered knee injuries are at increased risk of developing OA.^8,51,76 Patients with OA suffer not only from pain associated with degenerative joint changes but also from stiffness and loss of function in the affected joints, thus limiting their ability to participate in physical activities.^38,71 OA is associated with the development of joint contractures.^{9,17,25,49,64,71,75} A joint contracture, or loss of passive joint range of motion (ROM), results in considerable morbidity, including pain, stiffness, and loss of function such as walking, climbing stairs, running, or participating in exercise-related activities.^1,9 For those with knee OA, at least one-third of individuals will develop a contracture in the OA-affected joint.^{9,17,25,49,64,71,75} A joint contracture and OA leads to significant morbidity beyond that of OA alone.^11,12,14 In addition to more pain and loss of function, those with an OA-associated contracture are at higher risk of falls and increased mechanical stress on proximal and contralateral joints.^{14,21,29,40,41,52,57,74,75,78,82} Further, contractures may accelerate the OA disease process toward end-stage disease and the need for joint replacement surgery.^11,71 Even after arthroplasty, a preoperative joint contracture is the top risk factor for poor outcomes postarthroplasty.^61,64 This presents major systemic healthcare concerns given the number of hip and knee replacements exceeds 1 million annually in the United States alone. ¹⁸ Globally, OA accounted for 2.2% of years lived with disability, with an annual cost of OA to the United States healthcare system of >$16.5 billion, accounting for >4% of the costs for all hospitalizations.^22,59

Stretching is an inexpensive and commonly used intervention for the treatment and prevention of joint contracture associated with many health conditions. ⁹ Orthotic devices are also well described as a treatment for the reversal of joint contractures and have the advantage of providing a stretch to 1 or more joints over long periods without the need for direct manual intervention during their use.^9,20 While OA-associated contractures likely share common elements of pathophysiology with other contracture-associated conditions (eg, immobility, neurologic disorders, burns), contributing pathognomonic features of the OA-affected joint may be unique. ¹⁵ For example, changes such as cartilage loss, bone marrow lesions, capsular and ligamentous alterations, joint effusions, and osteophytes are associated with loss of knee extension,^{15,16,24,36,55} suggesting the possibility that response to stretching in OA-affected joints may not mirror that of other conditions.^16,38,46,63 In addition, little is known regarding the effects of stretching on such OA-associated structural changes. For example, contractures associated with mature bony osteophytes may be resistant to improvement through stretching due to established mechanical block, whereas the presence of early osteophytes, characterized by more malleable gelatinous cysts, may be more amenable to stretch treatment before their ossification. ⁶ Similarly, aggressive stretching on a joint with a large effusion may cause pain and hinder ROM recovery, while initiating stretching before large effusion development may yield better results. Indeed, established contractures are notoriously difficult to treat,^13,30,39 and evidence-based recommendations for their treatment in patients with OA is lacking. Correspondingly, a previous high-quality systematic review showed no benefit from stretching for the prevention or treatment of contractures for any musculoskeletal condition. ³⁰ However, this review included only a single study evaluating patients with OA before joint replacement, and participants with OA were combined with those living with rheumatoid arthritis, making it difficult to draw any conclusions about stretching effects for OA-affected joints. ³⁰ Another recent systematic review concluded that stretching could be used to assist in the management of knee OA pain, but this study did not evaluate the impact of stretching on joint ROM or functional outcomes nor did it evaluate the possible role of stretching devices such as orthoses. ⁴⁵

Given the large potential benefits to patients and the healthcare system from regaining ROM in OA-affected joints, as well as the limited guidance and evidence available regarding the impact of stretching on the OA joint, the aim of this review was to determine the effects of stretching and/or bracing on joint ROM, pain, and function in people with radiographically diagnosed OA before joint arthroplasty.

Methods

Study Reporting and Protocol Registration

This systematic review adhered to the statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions [Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)]. ⁴⁴ The review protocol was prospectively registered on PROSPERO (CRD42019127244) and has been published. ¹⁰

Search Methods for Identification of Studies

The initial comprehensive search of the following electronic databases was done in March 2019, then repeated in September 2021, searching for studies from inception to September 30, 2021: Cochrane Central Register of Controlled Trials (CENTRAL), The Database of Abstracts of Reviews of Effects (DARE) and The Health Technology Assessment Database (HTA); MEDLINE (Ovid); Embase (Ovid); CINAHL (Ovid); SCI-EXPANDED (ISI Web of Knowledge); and PEDro. We also searched the World Health Organization International Clinical Trials Registry Platform (https://www.who.int/clinical-trials-registry-platform/network) and https://clinicaltrials.gov/ to identify unpublished and ongoing trials. Electronic searches were complemented with a search of the reference lists of included studies, relevant systematic reviews, and textbooks. Regarding gray literature, we searched the Abstract archives from the American College of Rheumatology/Association of Rheumatology Health Professionals, European League Against Rheumatism, Canadian Physiotherapy Association and American Physical Therapy Association conferences from 2012 to 2021. See Online Appendix Table A1 for the MEDLINE full search strategy.

Study Selection and Participants

Study design inclusion included randomized controlled trials (RCTs), controlled clinical trials (CCTs), controlled before-and-after studies, cohort studies, case-control studies, studies using parallel-group designs, within-subject or cross-over designs, published in the English language, and for which full-length articles were available.

We included participants ≥18 years of age with radiographic evidence of OA of any peripheral joint. Studies including participants with inflammatory arthropathies (eg, seropositive or seronegative) were excluded. To maintain our focus on conservative preoperative OA contracture treatment, participants who received treatment postarthroplasty were excluded.

Two review authors independently screened titles and abstracts of the search output to identify potentially relevant studies. Full reports were obtained for all potentially relevant studies that appeared to meet the inclusion criteria. We recorded the selection process in alignment with the PRISMA flow diagram (http://prisma-statement.org/PRISMAStatement/Default.aspx). ⁴⁴

Types of Interventions and Comparisons

Types of interventions included therapist- or patient-administered stretch, use of orthosis (static or dynamic) or serial casting that provided a stretch, and adjunctive treatment with modalities of any duration (eg, ultrasound, inferential therapy, thermal, etc). We included all studies for which the effects of the intervention could be isolated. That is, if they compared a stretching or bracing intervention alone or in combination to any other treatment (eg, stretching plus ultrasound, vs ultrasound only) or placebo, such that differences between the groups could be attributed solely to the applied stretching intervention.

Types of Outcome Measures

Major (primary) outcomes included passive and active joint ROM, pain at rest or with activity, and number of patients experiencing any adverse event (AEs) (eg, increased pain or injury due to stretching or bracing). Because different anatomic structures may limit joint ROM in one direction but not another (eg, knee extension may be limited by the posterior capsule, while knee flexion may instead be limited by quadriceps shortening), and may therefore be impacted differently by stretching or bracing interventions, we segregated ROM analysis by joint movement direction. In addition, different joint anatomy (eg, the knee hinge joint vs the hip ball and socket joint) may respond differently to interventions, and thus ROM was also segregated by joint. We grouped outcomes by 3 main timing categories: short-term (≤3 months), intermediate-term (>3 but <6 months), and long-term (≥6 months) intervention.

Minor (secondary) outcomes included stiffness, functional performance, self-reported activity limitation or participation restriction, quality of life, patient satisfaction, radiographic changes, postarthroplasty outcomes (if stretching was performed preoperatively), and patients who withdrew because of AEs.

Data Extraction and Management

Data for each included trial were extracted by 2 independent reviewers using a standardized data extraction form. Discrepancies were resolved through discussion. We extracted the following study characteristics:

• Methods: study design, total duration of study, details of any ‘run-in’ period, number of study sites and location, study setting, withdrawals, and date of study;

• Participants: N, mean age, age range, biological sex, disease duration, severity of condition, diagnostic criteria, important baseline data, inclusion and exclusion criteria;

• Interventions: following the recommendations outlined in the Consensus on Exercise Reporting Template, ⁶⁹ we recorded characteristics of the intervention and comparison including details of treatment and control interventions; duration, frequency, and stages of intervention; details of cointerventions and their effects on stretched tissue (eg, temperature); compliance with and adherence to treatment;

• Outcomes: details of the major and minor outcomes - methods used to measure outcomes, mean scores and SDs of outcomes, direction of effect for each outcome;

• AEs (eg, stretch-related injury such as tendon rupture or muscle tear, or orthosis-related injury such as skin ulcer);

• Characteristics of the design of the trial as outlined in the following section;

• Notes: funding for trial, ethics approval/informed consent, and notable declarations of interest of trial authors.

Data were reported as means and CI. Missing SDs were computed from other statistics such as SEs or CIs or P values. ³⁴

Assessment of Risk of Bias in Included Studies

Two review authors independently assessed the risk of bias of the included studies using the Cochrane Risk of Bias tool. ³⁴ We assessed the following methodological domains: sequence generation, allocation sequence concealment, blinding of participants and treating clinicians, blinding of outcome assessors for objective outcomes, blinding of participants for self-reported outcomes, incomplete outcome data, selective outcome reporting, and other potential threats to validity. ³³ We judged these domains explicitly using the following criteria: Yes, low risk of bias; No, high risk of bias; Unclear, either lack of information or uncertainty over the potential for bias. We summarized the “Risk of Bias” judgements across different studies for each of the domains listed above. Disagreements were resolved by discussion. We attempted to contact authors of included studies to clarify any relevant information or request additional data when required through either email (2 attempts), ResearchGate, or their academic affiliation to inform our assessments.

Data Synthesis

For continuous outcomes, we analyzed mean differences (MDs) with SD and 95% CIs. The between-groups MD and 95% CIs were calculated using the mean final score for the intervention and control groups. We performed meta-analyses when treatments, participants, and the underlying clinical question were similar enough for pooling to make sense. We used a random-effects model to conduct meta-analyses and analyze data. We assessed heterogeneity by visual inspection of forest plots and used the I² statistic to quantify the heterogeneity of outcomes. ³² Positive values of the MD represent an effect in favor of the stretching group. For studies that included multiple treatment arms, we selected the treatment arm that allowed us to isolate the effect of the stretching intervention.

Summary of Evidence

Two authors independently assessed the quality of the evidence. We used the 5 Grading of Recommendations, Assessment, Development and Evaluations (GRADE) considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to the studies that contribute data to the meta-analyses for the prespecified outcomes, and reported the quality of evidence as high, moderate, low, or very low.^5,7,27,28 We summarized the strength of the evidence separately for the major outcomes using GRADE. We justified decisions to downgrade or upgrade the quality of studies using footnotes to aid the reader’s understanding of the review. ⁶⁶

Results

Selection of Studies and Study Characteristics

We identified 6284 articles through database searching. We excluded 5 duplicate articles. After screening titles and abstracts, we excluded 5513 citations and identified 766 studies as potentially eligible. We excluded a total of 757 studies after full-text review and identified 9 RCTs that met our inclusion criteria (Figure 1, Online Appendix Table A2).^{3,4,19,48,50,65,70,72,80} No non-RCTs were included. Of the included studies, all considered the knee joint; 8 studies enrolled patients with therapist-administered or patient-administered stretching intervention, 1 study used a stretching device. ⁷² Of the studies included in the final analysis, there was a total of 600 combined intervention and control participants enrolled, of whom 248 had an exercise/stretching intervention and 46 used a device as an intervention. The characteristics of the included studies and the types of interventions are summarized in Online Appendix Table A2. The summary of findings for our primary outcomes and quality of evidence is shown in Table 1.

Figure 1.

PRISMA flow diagram of search results. PRISMA, preferred reporting items for systematic reviews and meta-analyses; ROM, range of motion.

Table 1.

Summary of findings for primary outcomes and quality of evidence assessment (GRADE) ^a

Outcome	No. of Participants in Intervention Arm	MD/SMD [95% CI]	GRADE
Primary Outcomes
Stretching
ROM
Total Knee ROM (AROM and PROM)	1263,65,80	MD 9.3° [5.0°,13.5°]	Moderate b
Knee Flexion	5665,70	MD 10.8° [7.3°,14.2°]	Moderate b
Knee Extension	1514,19,48,50,65	MD 9.1° [3.4°,14.8°]	Moderate b
Pain	1883,4,48,50,65,70,80	SMD 1.9 [1.2,2.6]	Moderate b
ROM - Device
Knee Extension	46 72	MD 3.7° [2.9°,4.5°]	Moderate b
Secondary Outcomes
Stretching
Function/disability
WOMAC	43 65	MD 11.8 [8.5,15.1]	Moderate b
TUG	15 4	MD 3.3 [2.0,4.7]	Low c
Lequesne	33 80	MD 0.8 [0.3,1.3]	Low c

AROM, active range of motion; GRADE, Grading of Recommendations, Assessment, Development and Evaluations; MD, mean difference; PROM, passive range of motion; ROM, range of motion; SMD, standardized mean difference; TUG, timed-up-and-go test; WOMAC, Western Ontario and McMaster Universities Osteoarthritis.

^aData provided are for short-term outcomes (≤3 months postintervention).

^bDowngraded for risk of bias (1 point).

^cDowngraded for risk of bias (1 point) and imprecision (1 point).

Primary Outcomes

All included studies focused on stretching of the knee joint, therefore all reported outcomes pertain to the knee.

Short-Term Outcomes (≤3 Months After Initiation of Intervention)

Stretching versus control

Range of motion

The review studies reported different outcomes for knee ROM including active ROM (AROM) and passive ROM (PROM), knee flexion, and extension (Table 1 and Online Appendix Table A2). One study provided the intra-rater reliability for measuring ROM, ⁵⁸ and 3 studies described measurement methods in sufficient detail to replicate (eg, patient positioning, landmarks used).^19,48,80

• PROM versus AROM total knee ROM: 1 study reported passive total ROM for the knee with an MD of 8.4° improvement in total PROM for stretching over control that did not reach statistical significance [95% CI, -1.3°, 18.1°] (P = 0.09; Table 1) ³ ; the same study reported total AROMfor the knee during gait, with a mean increase of 4.9° [-0.11°,9.9°] (P = 0.06; Table 1). ³ A second study reported an assisted AROM improvement of 12.0° [8.5°,15.5°] (P < 0.001; Table 1). ⁸⁰

• Unspecified total ROM: 1 study reported improvement in ROM for the knee with stretching versus controls without describing whether it was PROM or AROM of 6.1° [1.3°,10.9°] (P = 0.02; Table 1) versus controls. ⁶⁵

• Combined total ROM: 3 studies reported on total knee ROM and were combined in meta-analysis.^3,65,80 Total knee ROM improved by 9.3° [5.0°,13.5°] (I² = 48%; P < 0.001; Table 1 and Figure 2A) versus controls.

• Flexion ROM: pooled data from 2 studies showed improved knee flexion ROM versus controls (MD 10.8° [7.3°,14.2]; I² = 0%; P < 0.001; Table 1 and Figure 2B).^65,70

• Extension ROM: 5 studies reported knee extension ROM.^{4,19,48,50,65} There was a mean improvement of 9.1° [3.4°,14.8°] (I² 99%; P = 0.001; Table 1 and Figure 2C) versus controls.

• The overall quality of the evidence for combined total knee ROM, knee flexion, and knee extension was each considered moderate (Table 1).

Figure 2.

Forest plots showing meta-analysis results. (A) Knee total ROM. (B) Knee flexion. (C) Knee extension. (D) Knee pain (SMD). IV, inverse variance; ROM, range of motion; SMD, standardized mean difference.

Pain at rest or with activity

For stretching studies that evaluated knee pain, 1 study reported improved Western Ontario and McMaster Universities Osteoarthritis (WOMAC) knee pain score of 2.8 [1.7,3.8] after proprioceptive neuromuscular facilitation (PNF) stretching (P < 0.001). ⁷⁰ Six other studies reported numeric rating scale or visual analog scale for knee pain, all of which reported a greater pain reduction after stretching intervention when compared with controls.^{3,4,48,50,65,80} Meta-analysis of all 7 studies showed moderate-quality evidence of reduced pain with stretching (SMD 1.9 [1.2,2.6]; I² 85%; Table 1 and Figure 2D).^{3,4,48,50,65,70,80}

Braces and Devices Versus Control

Range of motion

One study reported an improvement in knee extension after use of a dynamic knee extension aid device compared with control by 3.7° [2.9°,4.5°] (P < 0.001). ⁷² The overall quality of the evidence for knee extension using a device was considered moderate (Table 1).

Adverse events

A single study, ⁷² which evaluated the effectiveness of a dynamic knee extension aid device, reported AEs. In that study, none of the reported AEs were attributed to the use of the dynamic knee extension aid device. ⁷² No study described AEs related to stretching without the use of a device.

Intermediate-Term Outcomes (>3 to < 6 Months After Initiation of Intervention)

No studies reported intermediate-term outcomes.

Long-Term Outcomes (≥6 Months After Initiation of Intervention)

Range of motion

One study reported an improvement in total knee ROM at 1 year compared with control (MD 22.0° [18.4°,25.6°]; P < 0.001). ⁸⁰

Function

One study reported an improvement in the Lequesne index, an 11-item questionnaire evaluating pain, function, and mobility designed to obtain subjective information from patients about their knee OA, ⁴³ at 1 year compared with control (MD 3.3 [2.6,4.0]; P < 0.001). ⁸⁰

Secondary Outcomes

Stretching Versus Control

Stiffness, functional performance, quality of life, assessment of quality of life

All but 1 study reported short-term secondary outcomes (≤3 months). ⁸⁰ For short-term secondary outcomes, 1 study reported an improvement in WOMAC function after static stretching, MD 11.8 [8.5,15.1] (P < 0.001). ⁶⁵ The GRADE rating was considered moderate (Table 1). One study reported Lequesne index for the knee, mean improvement of 0.8 [0.3,1.3] (P = 0.001) for stretching versus control, ⁸⁰ and 1 reported the timed-up-and-go test with improvement versus control of 3.3 [2.0,4.7] (P < 0.001). ⁴ These findings were considered low-quality (Table 1).

For long-term outcomes, 1 study reported an improvement in the Lequesne index at 1 year compared with control (MD 3.3 [2.6,4.0] (P < 0.001). ⁸⁰

Postarthroplasty outcomes

One study evaluated the preoperative use of a device on postarthroplasty knee extension and found an MD favoring device versus control of 3.3° [2.5°,4.1°] (P < 0.001). ⁷²

Risk of Bias

Randomized Controlled Trials

We summarized the risk of bias of all included studies (Figure 3).

Figure 3.

Risk of bias summary for included studies.

• Allocation and concealment: 3 of 9 studies used adequate methods for generating the randomization sequence while 2 of 9 used adequate methods to conceal allocation.

• Blinding of participants and personnel: 3 of 9 of studies had high risk of performance bias for unblinded participants, and all studies had high risk of performance bias for unblinded personnel who delivered the interventions.

• Incomplete outcome data: 3 of 9 studies had high risk of attrition bias.

• Selective reporting: 1 study was at high-risk of selective reporting bias and 6 of 9 of studies had unclear risk of reporting bias.

• Other potential risk of bias: for 8 of 9 studies, the risk of bias for differences of cointerventions between arms was rated as unclear.

Discussion

In this study, we sought evidence-based stretching, bracing, and/or device use interventions to improve ROM and important clinical outcomes in OA-affected joints before undergoing arthroplasty. After meta-analysis combining 3, 2, and 5 studies, our results showed statistically significant improvement in total knee ROM, knee flexion, and knee extension after stretching, respectively. The GRADE quality of the evidence for each of these findings was considered moderate. After intervention with a device, a single RCT showed moderate-quality evidence of a superior effect for preoperative device use versus standard care for knee extension before and after arthroplasty. Finally, we found moderate-quality evidence for an improvement in knee pain and WOMAC function with the application of stretching versus control.

Our results demonstrated improvement in knee ROM after stretching OA-affected knees while a previous high-quality systematic review by Harvey et al ³⁰ showed no benefit of stretching on ROM for OA-affected joints. A systematic review by Raposo et al ⁶² evaluated the effect of exercise on knee OA and could not reach a clear conclusion regarding the effect of stretching on OA outcomes. Dissimilarities between our results and those reported previously may be due to different methodology. For example, while our review focused on participants who had not undergone arthroplasty, Harvey et al ³⁰ included studies evaluating participants both pre- and postarthroplasty. Our analysis builds on that by Harvey et al ³⁰ by extending the prearthroplasty population analysis. In addition, Harvey et al ³⁰ included studies with mixed inflammatory and noninflammatory (OA) arthritis populations, while we excluded studies if the data pertaining to participants with OA could not be segregated. Our analysis approach also distinguished ROM by joint movement direction (eg, flexion vs extension), rather than consolidating all ROM into a single meta-analysis. We did this on the basis that the effects of stretching could differentially affect different joint movements. For example, the knee is a modified hinge joint between the tibia and the femur and also includes articulation between the femur and patella.^46,54 The quadriceps tendon, patellar ligament, and expansions from the extensor muscles provide anterior stability, while the knee flexors help to provide posterior stability. ⁵⁴ The joint capsule that encloses both the tibiofemoral and patellofemoral joints is reinforced by tendons and expansions from the surrounding muscles and ligaments. ⁵⁴ Thus, different joint structures and tissue types (ligaments, tendons, capsule) limit opposing joint movements, and may respond differently to stretching. In OA, animal,^73,75,83 clinical,^36,71 surgical,^26,42,47 and imaging studies have revealed that pathologic changes contributing to joint ROM limitations include osteophyte formation, posterior capsule fibrosis, synovial inflammation, and degenerative soft tissue changes (eg, meniscal tears).^37,55 Thus, in addition to the anatomic tissues noted above, additional OA-associated structural limiters of ROM may be present in different joint regions, differentially affecting response to stretch.

As compared with the review by Raposo et al, ⁶² a further methodological difference was our exclusion of studies for which the effects of stretching could not be isolated (eg, studies incorporating stretching, strengthening, and balance exercises as an undivided intervention). ¹⁰ An additional factor may be that our literature search included more recent studies that were not published at the time of previous reviews and which used different stretching techniques.^4,48,65,70 At the time of the review by Harvey et al ³⁰ (published in 2017), only a single study using the PNF technique was included, ⁵⁰ whereas this review included 6 studies using PNF, 4 of which were published after 2017. The PNF technique involves a hold-relax stretching approach that includes a short isometric contraction of the agonistic muscles that are targeted to be stretched (eg, the hamstrings for knee extension).^48,67 This method of slow sustained stretching has been postulated to cause firing of the Golgi tendon organs, which lie in series with the extrafusal muscle fibers, resulting in muscle relaxation. ⁸⁰ This may reduce active resistance to stretch by the patient and allow a more effective stretch of other soft tissues implicated in contracture, such as the ligaments and capsule.^17,35,81 Although this effect on soft tissue is described, it is difficult to know how this technique would impact bony OA alterations, such as mature osteophytes. ¹⁵ The effect of a stretching technique such as PNF may therefore be more potent earlier in the OA disease, when such bony changes are less well established. OA disease duration could therefore play a role in the expected stretch outcome, but these data were rarely provided in our included studies. For those that did provide such data,^3,80 disease duration was heterogeneous and any differential stretching effect based on this factor was not reported. Finally, with 1 exception, ⁷⁰ our included studies using the PNF technique evaluated AROM rather than PROM, making it unclear whether ROM limitations before intervention represented a true joint contracture (loss of PROM), ⁹ or simply ROM limited by other factors such as pain inhibition or muscle weakness that improved with treatment.

Though further high-quality research in this area is needed before recommendations can be made to include stretching as a guideline-based component of OA clinical care, our findings are encouraging, and suggest that stretching the knee using techniques such as static and PNF stretching may indeed restore flexion, extension, and total ROM, and that efforts toward this goal may not be futile.

Interestingly, other than the single RCT describing a knee extension device before arthroplasty, ⁷² we did not identify any studies that used orthotic interventions to induce stretch, such as static progressive splinting or serial casting, and therefore the role of these interventions for restoring ROM in the OA joint is uncertain.

Our analysis revealed an improvement in pain after stretching. Consistent with our results, Raposo et al ⁶² found that the inclusion of stretching within a general therapeutic or exercise program, such as hydrotherapy and aquatic-based exercise, had a positive impact on pain. Similarly, Ebnezar et al ²³ observed pain reduction in knee OA with a comprehensive Hatha yoga therapy program. However, in these reviews,^23,62 the isolated impact of stretching was not evaluated, making it challenging to determine the effect size of the stretching component of their comprehensive intervention program. More recently, Luan et al ⁴⁵ reported that, when stretching exercises were used in isolation, the reduction of pain was both significant and clinically meaningful, with a weighted MD on a VAS of 1.9 out of 10, identical to our results and exceeding the MCID of 1.4 points on a 10-point scale, as described by Hawker et al. ³¹ Stretching may therefore represent an inexpensive and effective strategy for pain management of the OA-affected joint.

For our secondary outcomes, we found 3 studies that individually reported an improvement in function after either static stretching or PNF.^4,65,80 Our findings are consistent with previous work showing an association between knee ROM and function in the OA joint.^11,12,14

The findings in this review are clinically relevant. Necessary joint ROM is often described in terms of thresholds for normal function, for example 90° of knee flexion for sitting, ⁴⁰ 94° for climbing stairs, ⁶⁰ and 60° for normal ambulation. ¹ Once these thresholds are crossed, normal function can no longer be achieved. ⁹ Knee extension PROM is an important factor for determining patient satisfaction postarthroplasty.^61,64 As well, even mild (1-5°) knee flexion contractures were associated with increased pain, stiffness, and loss of function, and were found to be a risk factor for more rapid progression to arthroplasty.^11,12 The mean improvement in knee ROM after stretching (~9-10°, depending on whether it was total ROM, flexion, or extension) seen in our meta-analysis could therefore contribute positively to reducing patient symptoms, achieving a clinically meaningful benefit (eg, reducing a flexion contracture from 15° to 5° could have notable positive impact for the patient).^2,31 Vigilant clinical monitoring of knee ROM in patients with OA could trigger stretching interventions before exceeding ROM thresholds to maintain function and avoid developing “irreversible or fixed” contractures with magnitudes beyond these thresholds.

While our review protocol did not exclude any studies based on the joint studied (eg, studies evaluating the hip or ankle), ¹⁰ our literature search did not identify any studies that met our inclusion criteria evaluating joints other than the knee. Studies including other joints were often excluded due to lack of segregation of the stretching intervention (eg, stretching was combined with other interventions such as strengthening and soft tissue release techniques), precluding our ability to assess the effect of stretching in isolation. Regarding the use of devices for ROM restoration, we identified only a single study using a manually controlled stretching device for this purpose. ⁷² Despite being commonly described anecdotally for the correction of joint contractures,^9,20 studies using orthotic devices that were screened in our literature search described orthosis use for joint protection, rather than ROM restoration. Further research evaluating joints other than the knee and that evaluating the use of orthotic devices with the goal of ROM restoration is needed.

Review limitations include the small number of studies, trial heterogeneity, and limiting primary studies to English only. A single study was included evaluating a device to improve knee ROM, ⁷² precluding meta-analysis for device or orthosis use and limiting our findings to the data presented in that study. The moderate quality of the evidence in this study limits the strength of our findings.

In conclusion, there was moderate-quality evidence that stretching is an effective strategy for improving total knee ROM, knee flexion, and knee extension as well as clinically important outcomes such as pain and function. A device that assists the patient in providing self-administered knee extension appears to be safe and may also improve OA outcomes. Stretching is an appropriate conservative treatment to consider when designing a comprehensive knee OA treatment plan before arthroplasty.