Nonsurgical Treatment Options for Muscle Contractures in Individuals With Neurologic Disorders: A Systematic Review With Meta-Analysis.

OBJECTIVE: To investigate whether nonsurgical treatment can reduce muscle contractures in individuals with neurologic disorders. The primary outcome measure was muscle contractures measured as joint mobility or passive stiffness. DATA SOURCES: Embase, MEDLINE, Cumulative Index to Nursing and Allied Health, and Physiotherapy Evidence Database in June-July 2019 and again in July 2020. STUDY SELECTION: The search resulted in 8020 records, which were screened by 2 authors based on our patient, intervention, comparison, outcome criteria. We included controlled trials of nonsurgical interventions administered to treat muscle contractures in individuals with neurologic disorders. DATA EXTRACTION: Authors, participant characteristics, intervention details, and joint mobility/passive stiffness before and after intervention were extracted. We assessed trials for risk of bias using the Downs and Black checklist. We conducted meta-analyses investigating the short-term effect on joint mobility using a random-effects model with the pooled effect from randomized controlled trials (RCTs) as the primary outcome. The minimal clinically important effect was set at 5°. DATA SYNTHESIS: A total of 70 trials (57 RCTs) were eligible for inclusion. Stretch had a pooled effect of 3° (95% CI, 1-4°; prediction interval (PI)=-2 to 7°; I 2=66%; P<.001), and robot-assisted rehabilitation had an effect of 1 (95% CI, 0-2; PI=-8 to 9; I 2=73%; P=.03). We found no effect of shockwave therapy (P=.56), physical activity (P=.27), electrical stimulation (P=.11), or botulinum toxin (P=.13). Although trials were generally of moderate to high quality according to the Downs and Black checklist, only 18 of the 70 trials used objective measures of muscle contractures. In 23 trials, nonobjective measures were used without use of assessor-blinding.
CONCLUSIONS: We did not find convincing evidence supporting the use of any nonsurgical treatment option. We recommend that controlled trials using objective measures of muscle contractures and a sufficiently large number of participants be performed.

Muscle contractures are a common complication of neurologic disorders such as stroke, spinal cord injury, multiple sclerosis, and cerebral palsy. The prevalence has been reported to range from 36%-60%.1, 2, 3, 4, 5 Muscle contractures represent a unique muscle adaptation characterized by increased passive stiffness of the muscle and limited mobility of the joint with little or no active force production. Muscle contractures lead to joints fixated in abnormal positions and limited use of the affected limbs. Furthermore, muscle contractures can cause considerable pain, strength loss, and muscle atrophy.,

To restore the mobility of affected joints, surgical procedures such as various forms of tendon lengthening and intramuscular aponeurotic recession are used., These procedures may increase the range of motion for some time, but because they rarely have lasting effects, other effective treatment approaches should be considered also. A variety of other treatment options currently exists. A few of these have previously been reviewed (stretching and shockwave therapy,), but a systematic evaluation of the effectiveness of all the available nonsurgical treatment options in a single review has so far not been conducted. A critical and comprehensive evaluation of the effect of all treatment options in 1 single study may help clinicians to obtain a better overview of the field. It may also help to clarify where the existing knowledge needs to be strengthened by further research and point to new therapy options.

Therefore, the aim of this systematic review was to provide an overview of the evidence supporting the use of current nonsurgical treatment options for reduction of muscle contractures in individuals with neurologic disorders. We included randomized controlled trials (RCTs) and controlled clinical trials (CCTs) of nonsurgical interventions administered with the aim to treat muscle contractures in individuals with neurologic disorders. We decided to include not only RCTs but also CCTs because we wanted to review all available treatment options. The primary outcome measure was muscle contractures measured as either joint mobility or passive stiffness.

Methods

Study design

We conducted this systematic review with meta-analyses of RCTs and CCTs using a protocol based upon Cochrane Collaboration recommendations and reported it according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.,

Eligibility criteria

Published trials fulfilling the following patient, intervention, comparison, outcome (PICO) criteria were included.

Participants

Individuals of all ages and sexes with muscle contractures due to a neurologic disorder.

Interventions

Nonsurgical interventions administered to treat muscle contractures.

Comparisons

Trials that compared the intervention with a control condition. Control condition included no intervention, usual care, and placebo/sham treatment.

Outcomes

The main outcome was muscle contractures measured as either passive range of motion (PROM) or passive stiffness.

Search strategy

Relevant articles were identified by searching the databases of Embase, MEDLINE, Cumulative Index to Nursing and Allied Health, and Physiotherapy Evidence Database, using a combination of subject headings and free-text terms. The search string was initially developed for MEDLINE and adapted for use in the other databases. Search strings used in all databases can be found in the supplemental table S1 (available online only at http://www.archives-pmr.org/). Publications were limited to the English language. Publications were not limited by year of publication. We performed the search in June-July 2019. We additionally searched all databases in July 2020 to detect any eligible trials published during the review process.

Data extraction

Two review authors (C.S., J.L.) screened title and abstracts of all records obtained from the searches and excluded irrelevant articles. Full texts of the remaining articles were then obtained and screened for eligibility based on our PICO criteria by the 2 review authors (C.S., J.L.). Through subjective judgment, the reviewers doing the data extraction decided whether the intervention was administered to treat muscle contractures. Disagreements were solved by discussion and, when necessary, arbitrated by a third review author (J.B.N.) deciding whether to include or exclude the disputed.

Data synthesis

C.S. extracted short-term joint mobility data (up to 1wk after intervention). Preferably, change scores and SDs were extracted. If change scores were not available, postintervention scores were used instead. Change scores/postintervention scores and SDs were not available for all trials. In trials where this information was not available, we contacted the corresponding author of the article in an attempt to retrieve the information. Several trials investigated the effect of the intervention on multiple joints and/or both sides. In these cases, we used data from a single joint on the right side of the body. In prioritized order, we chose to use data from the ankle joint, the elbow joint, the knee joint, or the wrist joint. This order was based on our experience of where muscle contractures are frequent and severe and is in accordance with literature on muscle contracture prevalence in different neurologic disorders.1, 2, 3

We identified 6 types of interventions with multiple trials: stretch, shockwave therapy, physical activity, botulinum toxin (BTX) treatment, electrical stimulation, and robot-assisted rehabilitation interventions. Based on the recommendations by Valentine et al, we conducted individual meta-analyses for these 6 intervention types. Because very few trials used passive stiffness as an outcome measure, the meta-analyses were performed based on PROM results. The primary outcome measure was set as the pooled PROM from RCTs. For all intervention types, we conducted sensitivity analyses to examine the effects of randomization on joint mobility. Similarly to Harvey et al, we did not consider a treatment effect of <5° PROM as clinically important. Because we considered the included trials to have varying effect sizes, all meta-analyses were performed using a random-effects model. In accordance with the Cochrane Handbook for Systematic Reviews of Intervention, we reported the effects using mean differences in the meta-analysis in cases where the outcome was reported using comparable measures. In 1 case with robot-assisted rehabilitation, the outcome was not measured using comparable methods. Here, we reported the effect of the intervention using standardized mean differences in the meta-analysis. In forest plots, randomized and nonrandomized trials are presented separately. Subgroup analyses were used to explore possible differences between types of stretch. In studies with several relevant experimental groups (2 types of stretch protocols), we combined the experimental groups in to 1 single group. Prediction intervals were calculated in accordance with the method described by Borenstein. Meta-analyses were conducted using Review Manager 5.3.a

We assessed trials for risk of bias using the Downs and Black checklist. Initially, 2 review authors (C.S., J.L.) scored the first trials together to synchronize the interpretation of the checklist. Subsequently, C.S. and J.L. scored the remaining trials independently. The maximum score attainable using the Downs and Black checklist is 33 points. The quality of included trials was ranked as high if the total score was >75% of the maximum, moderate if 60%-74% of the maximum, and low if <60% of the maximum., In question 20 we focused on whether the primary outcome measure was objective. We defined an objective measure as a measure not easily influenced by the rater. All torque-controlled goniometric measures were defined as objective, whereas noncontrolled goniometric were not. As we were interested in whether joint mobility was measured objectively and by use of blinded assessors to not introduce bias, we focused in particular on question numbers 15 and 20.

Ethics and registration

This study did not require ethical approval. The systematic review protocol was prospectively registered in the PROSPERO international prospective register of systematic reviews under registration number CRD42019140424.

Results

Study selection

The review process is explained in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram (fig 1). We excluded 243 full-text articles because the trials did not fulfill our PICO criteria (211); because the full text was not available (12), not accessible (14), or was a duplicate (3); or because the primary data/summary statistics was not presented (3). The remaining 70 articles were included in this systematic review. Of the 70 articles included in the review, 57 were RCTs (see fig 1).

Fig 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Forest plot showing the mean difference with 95% CI for short-term effects of stretch on joint mobility.

Forest plot showing the mean difference with 95% CI for short-term effects of physical activity on joint mobility.

Forest plot showing the mean difference with 95% CI for short-term effects of BTX on joint mobility.

Forest plot showing the mean difference with 95% CI for short-term effects of electrical stimulation on joint mobility.

Forest plot showing the mean difference with 95% CI for short-term effects of robot-assisted rehabilitation on joint mobility.

Of the included trials, there were 22 trials (19 RCTs) on stretch interventions, 6 trials (2 RCTs) on shockwave interventions, 8 trials (7 RCTs) on BTX interventions, 9 trials (5 RCTs) on electrical stimulation interventions, 10 trials (8 RCTs) on physical activity interventions, and 5 trials (5 RCTs) on robot-assisted interventions. We performed meta-analyses for all of these intervention types. Additionally, we found 10 trials investigating other interventions. These trials are described in the section “Other interventions.”

Study characteristics

Table 1 depicts the characteristics of the included studies, including information about the intervention, the number of participants, and the measure of muscle contractures.

Table 1

Characteristics of the included trials (n=71)

Study	Participants	Intervention	Intervention details	n (Experimental Group)	n (Control Group)	Primary Outcome
Stretch
Fox et al19	Elderly persons with cognitive and functional impairment	Bed positioning	Bed positioning for 40 min, 4×/wk for 8 wk	12	12	PROM of knee extension measured using a goniometer
Maas et al20	Children with CP	Orthosis	Foot orthosis for 1 y	13	11	PROM of ankle DF measured using a single digital inclinometer attached to a torque wrench
Copley et al21	Adults with acquired brain injury	Splinting	Individualized, thermoplastic resting mitt splint for 3 mo	6	4	Wrist and finger PROM measured using a goniometer
DeMeyer et al22	Adults with stroke	Casting/orthosis	Bivalve cast group wore custom fiberglass castPRAFO group wore off-the-shelf AFO.Wearing schedule of 8-12 h every night for ∼4 wk.	PRAFO 14Bivalve cast 13	19	Ankle DF PROM measured using a standardized torque application
Beckerman et al23	Adults with stroke	Orthosis	AFO for 15 wk	16	14	PROM of ankle joint measured using a goniometer
Harvey et al24	Adults with stroke/SCI/TBI	Splinting	Experimental thumbs splinted into abduction. 8 h per night for 12 wk	29 thumbs	29 thumbs	PROM of palmar measured using a standardized torque measure
Kerem et al25	Adults with CP	Splinting	Johnstone pressure splints. 5 d/wk for 3 mo	17	17	PROM of the lower extremity measured using a goniometer
Harvey et al26	Adults with SCI	Passive movements	Passive ankle for 10 min in the morning and 10 min in the evening, 5 d/wk for 6 mo	20	20	PROM of ankle DF measured through application of standardized torque
Theis et al27	Children with CP	Passive stretch	15 min (60-s repetitions) of ankle DF stretch 4 d/wk for 6 wk	7	6	Passive stiffness of triceps surae
Harvey et al28	Adults with SCI	Passive stretch	Passive hamstring stretch for 30 min/d, 5 d/wk for 4 wk	14	11	Hamstring muscle extensibility measured using a torque-controlled measure
Cheng et al29	Children with CP	Repetitive passive movements	Knee repetitive passive movement intervention, 3/wk for 8 wk	18	18	PROM of knee joint measured using an electric goniometer
Lannin et al30	Adults with stroke	Splinting	Static, palmar resting mitt splint on a daily basis, for max 12 h/night for 4 wk	18	11	PROM of wrist extension measured using a torque-controlled measure
Basaran et al31	Adults with stroke	Splinting	Static volar or dorsal splints for 5 wk	Volar 13Dorsal 13	12	PROM of wrist extension measured using a goniometer
Moseley32	Children and adults with TBI	Casting	Below-knee cast for 7 d	9	9	PROM of the ankle joint measured using a torque-controlled measure
Pradines et al33	Adults with chronic hemiparesis	Passive and active stretch	Guided self-rehabilitation Contract program, consisting of daily self-stretch exercises for 1 y	12	11	Maximal extensibility (XV1 of the Tardieu Scale) of several muscles (PROM) measured with a goniometer
Lee et al34	Adults with stroke	Posterior talar glide	DF of the ankle joint for 10 glides of 5 sets/d, 5 d/wk for 4 wk	17	17	PROM of ankle joint measured using a digital goniometer
Harvey et al35	Adults with tetraplegia	Splinting	One thump of each participant was splinted each night for 3 mo	20	20	Extensibility of the flexor pollicis longus muscle measured with a standardized torque application
Hill36	Children and adults with brain injury	Casting	Casting for 1 mo	15	15	PROM of casted joints measured using a goniometer
Lannin et al37	Adults with stroke	Splinting	Hand splints positioning wrist in 0-10° extension (neutral splint group) or 45° wrist extension (extension splint group) at night for 4 wk	Neutral splint 20Extension splint 21	21	Muscle extensibility measured using a standardized torque measure
Smedes et al38	Adults with stroke	Manual mobilization	10-min manual mobilization of the wrist 2 d/wk for 6 wk	9	9	PROM of wrist extension measured using a goniometer
Horsley et al39	Adults with stroke	Passive stretch	30 min of self-assisted stretch of the wrist and finger flexors, 5 d/wk for 4 wk	20	20	PROM of wrist extension measured using a torque-controlled measure
An and Jo40	Adults with stroke	Talocrural mobilization	Talocrural mobilization 3 sessions/wk for 5 wk. Each session consisted of 6 sets of 10 repetitions.	13	13	DF PROM measured using a dynamometer
Electrical stimulation
Pool et al41	Children with CP	FES	8-wk FES intervention, FES used at least 1 h/d 6 d/wk	12	12	PROM of ankle DF measured using a goniometer
Pool et al42	Children with CP	FES	FES device, which dorsiflexes the ankle during the swing phase of gait for at least 4 h/d, 6 d/wk for 8 wk	16	16	PROM of ankle DF measured using a goniometer
Sabut et al43	Adults with stroke	FES	FES for 20-30 minutes to the TA muscle of the paretic limb 5 d/wk for 12 wk	27	24	PROM in the ankle joint measured using a goniometer
Bakaniene et al44	Children with CP	Transcutaneous electrical nerve stimulation/Mollii suit	Electrical stimulation through the Mollii suit for 1 h/d, 3/wk for 3 wk	8	8	PROM of ankle and knee joint measured using a goniometer
Malhotra et al45	Adults with stroke	NMES	30 min sessions of NMES to the wrist and finger extensors at least 2 times/d, 5 d/wk for 6 wk	45	45	PROM at slow stretchPassive stiffness at slow stretch
Nakipoglu Yuzer et al46	Adults with stroke	FES	FES for 30 min/d, 5 d/wk for a total of 20 sessions per patient	15	15	PROM of wrist extension measured using a goniometer
Leung et al47	Adults with TBI	Electrical stimulation	The intervention group received 30-min tilt table standing with electrical stimulation to the ankle dorsiflexor muscles 5 d/wk and ankle splinting 12 h/d, at least 5 d/wk.Control group only received tilt table standing for 30 min, 3 times/wk.	17	18	PROM of ankle DF measured with a torque-controlled measure
Sabut et al48	Adults with stroke	FES	FES of the TA muscle for 30 min, 5 d/wk for 12 wk	16	14	PROM of the ankle joint
Beaulieu et al49	Adults with stroke	Repetitive peripheral magnetic stimulation	Single session of repetitive peripheral magnetic stimulation	9	9	PROM of ankle DF
Shockwave therapy
Manganiotti and Amelio50	Adults with stroke	ESWT	As single session of ESWT	20	20	PROM of the wrist measured using a digital goniometer
Lee et al51	Adults with stroke	ESWT	A single session of ESWT	10	10	PROM of the ankle joint measured using a goniometer
Wang et al52	Children with CP	ESWT	1 ESWT session per wk for 3 mo.	34	33	PROM of the ankle joint measured using a goniometer
Gonkova et al53	Children with CP	ESWT	A single session of ESWT	25	25	PROM of ankle joint
Moon et al54	Adults with stroke	ESWT	3 sessions of ESWT, 1 session/wk for 3 wk	30	30	PROM of the ankle measured using a goniometer
Vidal et al55	Adults with CP	ESWT	Group 1 received ESWT in the spastic muscle, group 2 received radial ESWT in the spastic muscle and in the antagonistic muscle. 3 sessions, 1-wk intervals.	Group 1=14 musclesGroup 2=13 muscles	13	PROM of lower limbs measured using a goniometer
BTX
Love et al56	Children with CP	Botox	1 session of Botox into gastrocsoleus and where clinically indicated also into tibialis posterior	12	12	PROM of ankle joints measured using a goniometer
Hawamdeh et al57	Children with CP	Botox	3 successive Botox injections at intervals of 3-4 mo	40	40	PROM of ankle DF measured using a protractor goniometer
Rameckers et al58	Children with congenital spastic hemiplegia	Botox	1 session of Botox injections	10	10	PROM of wrist and elbow extension measured with a Mie goniometer
Meythaler et al59	Adults with stroke	Botox	Botox with therapy or placebo injections with therapy. 12-wk intervention.	21	21	PROM of elbow and wrist joint measured monthly using a goniometer
Tedroff et al60	Children with CP	Botox	Two Botox injections at 6-mo intervals	6	9	PROM of multiple joints measured using a goniometer
Koman et al61	Children with CP	Botox	Botox injections at baseline and at wk 4	56	58	PROM of ankle joint measured using a goniometer
Schasfoort et al62	Children with CP	Botox	Control group received 12 wk of conventional rehabilitation, intervention group received 12 wk of rehabilitation plus Botox injections	41	24	PROM of multiple joints measured using a Lafayette goniometer
El-Etribi et al63	Children with CP	Botox	Botox administered after baseline measurements	20	20	Ankle joint PROM measured using goniometer
Physical activity
Horsley et al64	Adults with stroke	Upper limb training	Active repetitive motor training by using the SMART Arm device for up to 1 h/d, 5 d/wk for 5 wk	25	25	PROM of multiple joints measured using a digital goniometer and a torque-controlled measure
Scholtes et al65	Children with CP	Resistance training	12-wk program of functional PRE training, 3 times/wk for 60 min	24	25	PROM of the multiple joints measured using a goniometer
Schmid et al66	Adults with stroke	Yoga	Therapeutic yoga sessions were delivered in group sessions for 1 h 2 times/wk for 8 wk	37	10	PROM of hamstrings muscles measured using a goniometer
Rydwik et al67	Adults with stroke	Exercise program	Exercise program including active and passive range of motion of the ankle with a portable device (Stimulo), 3 times/wk for 30 min, over a 6-wk period	9	9	PROM of ankle joint measured using a goniometer
Baik et al68	Children with CP	Horseback riding	Therapeutic horseback riding 60 min/d, 2 d/wk for 12 wk. Daily program consisted of 10 min of warm-up, 40 min of workout, and 10 min of cooldown.	8	8	PROM of hip joint measured using a goniometer
Lorentzen et al69	Adults with CP	Treadmill training	30-min daily uphill gait training for 6 wk on a treadmill	12	11	Passive stiffness of the ankle joint quantified using a stationary and hand-held dynamometer.The hand-held dynamometer also to assess the PROM of the ankle joint.
Kirk et al70	Adults with CP	Resistance training	Resistance training, 3 times/wk for 12 wk	12	11	Passive stiffness of ankle plantar flexors measured using a stationary dynamometer
An and Won71	Adults with stroke	MWM and WBE	30 min of MWM or WBE 3 times/wk for 5 wk	MWM 12WBE 8	10	PROM of the ankle joint using a isokinetic dynamometer
Teixeira-Machado and DeSantana72	Children with CP	Dance	24 one-h sessions twice a wk for 3 m	13	14	PROM of multiple joint measured using a goniometer
Hemachitara et al73	Children with CP	Horse riding	1 session of horse riding using a horse riding simulator	12	12	PROM of hip abduction measured using a goniometer
Robot-assisted rehabilitation
Mirbagheri et al74	Adults with SCI	Robotic-assisted step training	Three 1-h robotic-assisted step training sessions/wk for 4 wk	23	23	Intrinsic ankle stiffness measured as using torque/unit change in ankle position
Waldman et al75	Adults with stroke	Stretch and active movements	A portable rehabilitation robot with controlled passive stretching and active movement training capabilities. 18 sessions, 3 times/wk for 6 wk	12	12	Ankle DF PROM measured using the robotic device
Mirbagheri et al76	Adults with SCI	Robot-assisted locomotor training LOKOMAT	LOKOMAT training 3 d/wk for 4 wk	23	28	Intrinsic dynamic stiffness of the ankle joint
Franceschini et al77	Adults with stroke	Upper limb rehabilitation	Upper limb robot-assisted rehabilitation; 30 sessions, 5 d/wk for 6 wk	25	23	PROM of shoulder and elbow joint
Sale et al78	Adults with stroke	Robot-assisted therapy	Thirty 45-min sessions, 5 d/wk for 6 wk, using the robotic system that supported arm movements	26	27	PROM of the shoulder and elbow joint
Other
Rayegani et al79	Adults with SCI	Passive cycling	Motorized cycle that passively moved legs for 20 min, 3 times/wk for 2 mo	35	29	PROM of multiple joints measured using a goniometer
Xu et al80	Adults with stroke	MT combined with neuromuscular electrical stimulation	MT group received 30 min of MT training.Control group performed the same training but with nonreflecting side of the mirror.MT+NMES group combined MT with 30 min NMES.	MT 23MT+NMES 23	23	PROM of ankle joint DF assessed using a goniometer
Lorentzen et al81	Adults with TBI	Neural tension technique	1 session of neural tension technique treatment	10	10	Passive knee stiffness measured using the Neurokinetics RA1 Rigidity Analyzer
Mathew et al82	Children with CP	Antispastic medication	Participants received A (placebo), B (0.5/1.0mg diazepam), or C (1.0/2.0mg diazepam) for 15-20 d	60	60	PROM of ankle joint measured using a goniometer
Velasco et al83	Children with CP	Physical therapy based on head movements and serious games	10 sessions of gaming using the ENLAZA interface	5	5	Cervical PROM
Wayne et al84	Adults with stroke	Acupuncture	Traditional Chinese acupuncture, twice a wk for 10 wk	16	17	PROM of each major upper extremity joint
Cheng et al85	Children with CP	Whole body vibration	8-wk whole body vibration intervention	16	16	PROM of knee joint measured using an electrogoniometer
Fosdahl et al86	Children with CP	Stretching and PRE	16 wk of 3 weekly sessions of stretching and resistance training	17	20	Passive popliteal angle registered as maximum passive extension of the knee measured using a goniometer
Takeuchi et al87	Adults with cerebrovascular disease	HI-LPNR and stretching	Participants were randomized to 1 session of HI-LPNR, stretching, a combination, or a control group	HI-LPNR 10 stretching 10combination 10	10	PROM of ankle DF and passive resistive joint torque of ankle DF
Ghannadi et al88		Dry needling	1 session of dry needling	12	12	PROM of dorsiflexors measured using a goniometer

Abbreviations: AFO, ankle-foot orthosis; CP, cerebral palsy; DF, dorsiflexion; ESWT, extracorporeal shock wave therapy; FES, functional electrical stimulation; HI-LPNR, high-intensity pulse irradiation with linear polarized near-infrared rays; MT, mirror therapy; MWM, mobilization with movement; NMES, neuromuscular electrical stimulation; PRE, progressive resistance exercise; SCI, spinal cord injury; TA, tibialis anterior; TBI, traumatic brain injury; WBE, weight-bearing exercise.

Evidence quality

Table 2 summarizes the quality assessments performed based on the Downs and Black checklist. Data are presented as the subtotal scores, the total score, and the quality ranking of all trials. Furthermore, the average score for the different intervention types are presented. For detailed scoring of each individual article, we refer to the supplemental table S2 (available online only at http://www.archives-pmr.org/).

Table 2

Risk of bias in the included trials assessed using the Downs and Black checklist

Study	Reporting	External Validity	Internal Validity: Bias	Internal Validity: Confounding	Power	Total	Percentage	Quality
Stretch
Fox et al19	10	3	5	5	3	26	79	High
Maas et al20	11	3	6	6	3	29	88	High
Copley et al21	10	3	4	5	1	23	70	Moderate
DeMeyer et al22	10	3	5	5	3	26	79	High
Beckerman et al23	7	3	3	5	3	21	64	Moderate
Harvey et al24	11	3	6	6	5	31	94	High
Kerem et al25	10	0	4	3	3	20	61	Moderate
Harvey et al26	10	3	6	3	4	26	79	High
Theis et al27	8	1	5	3	2	19	58	Low
Harvey et al28	10	2	5	6	3	26	79	High
Cheng et al29	10	0	3	4	3	20	61	Moderate
Lannin et al30	9	2	5	5	3	24	73	Moderate
Basaran et al31	10	1	5	5	3	24	73	Moderate
Moseley32	9	1	4	5	2	21	64	Moderate
Pradines et al33	10	1	5	5	3	24	73	Moderate
Lee et al34	9	0	3	5	3	20	61	Moderate
Harvey et al35	11	2	5	4	3	25	76	High
Hill36	6	1	3	4	3	17	52	Low
Lannin et al37	9	0	6	5	3	23	70	Moderate
Smedes et al38	10	2	3	2	2	19	58	Low
Horsley et al39	11	2	6	6	3	28	85	High
An and Jo40	9	1	3	5	3	21	64	Moderate
Averages	10	2	5	5	3	23	71	Moderate
Electrical stimulation
Pool et al41	9	0	3	3	3	18	55	Low
Pool et al42	9	1	4	6	3	23	70	Moderate
Sabut et al43	10	3	3	5	4	25	76	High
Bakaniene et al44	9	0	4	2	2	17	52	Low
Malhotra et al45	9	2	5	5	5	26	79	High
Nakipoglu Yuzer et al46	9	0	4	4	3	20	61	Moderate
Leung et al47	10	1	5	5	3	24	73	Moderate
Sabut et al48	9	3	5	4	3	24	73	Moderate
Beaulieu et al49	10	0	6	5	2	23	70	Moderate
Averages	9	1	4	4	3	22	67	Moderate
Shockwave therapy
Manganiotti and Amelio50	11	2	5	3	3	24	73	Moderate
Lee et al51	10	3	6	6	2	27	82	High
Wang et al52	11	3	4	3	5	26	79	High
Gonkova et al53	6	1	4	1	4	16	48	Low
Moon et al54	10	0	4	4	4	22	67	Moderate
Vidal et al55	5	0	4	3	3	15	45	Low
Averages	9	2	5	3	4	22	66	Moderate
Botox
Love et al56	10	3	4	5	4	26	79	High
Hawamdeh et al57	10	2	4	5	4	25	76	High
Rameckers et al58	9	0	4	5	2	20	61	Moderate
Meythaler et al59	8	0	6	4	4	22	67	Moderate
Tedroff et al60	11	1	5	4	2	23	70	Moderate
Koman et al61	6	0	4	3	5	18	55	Low
Schasfoort et al62	10	1	5	2	5	23	70	Moderate
El-Etribi et al63	8	0	2	3	3	16	48	Low
Averages	9	1	4	4	4	22	66	Moderate
Physical activity
Horsley et al64	10	3	6	6	4	29	88	High
Scholtes et al65	11	3	5	5	5	29	88	High
Schmid et al66	9	0	2	4	5	20	61	Moderate
Rydwik et al67	9	0	5	5	2	21	64	Moderate
Baik et al68	8	0	2	0	2	12	36	Low
Lorentzen et al69	10	0	6	5	3	24	73	Moderate
Kirk et al70	9	0	5	3	4	21	64	Moderate
An and Won71	8	0	3	3	2	16	48	Low
Teixeira-Machado and DeSantana72	10	0	3	5	3	21	64	Moderate
Hemachitara et al73	11	0	5	5	3	24	73	Moderate
Averages	10	1	4	4	3	22	66	Moderate
Robot-assisted rehabilitation
Mirbagheri et al74	10	2	5	4	4	25	76	High
Waldman et al75	10	3	5	5	3	26	79	High
Mirbagheri et al76	5	0	3	3	4	15	45	Low
Franceschini et al77	11	1	4	6	4	26	79	High
Sale et al78	10	0	5	5	4	24	73	Moderate
Averages	9	1	4	5	4	23	70	Moderate
Other interventions
Rayegani et al79	10	3	2	5	5	25	76	High
Xu et al80	9	3	5	6	4	27	82	High
Lorentzen et al81	11	1	6	5	2	25	76	High
Mathew et al82	7	3	6	3	5	24	73	Moderate
Velasco et al83	8	0	4	4	1	17	52	Low
Wayne et al84	9	0	5	5	3	22	67	Moderate
Cheng et al85	9	0	3	3	3	18	55	Low
Fosdahl et al86	11	1	4	6	3	25	76	High
Takeuchi et al87	8	0	3	4	2	17	52	Low
Ghannadi et al88	9	0	6	6	3	24	73	Moderate

For stretch interventions, 8 trials were of high quality, 11 trials of moderate quality, and 3 trials of low quality. For electrical stimulation interventions, 2 trials were of high quality, 5 trials of moderate quality, and 2 trials of low quality. For shockwave interventions, 2 trials were of high quality, 2 trials of moderate quality, and 2 trials of low quality. For BTX interventions, 2 trials were of high quality, 4 trials of moderate quality, and 2 trials of low quality. For physical activity interventions, 2 trials were of high quality, 6 trials of moderate quality, and 2 trials of low quality. For robot-assisted interventions, 3 trials were of high quality, 1 trial of moderate quality, and 1 trial of low quality (table 3).

Table 3

Assessment of outcome measures

Study	Blinded Assessor	Objective Outcome Measure
Stretch
Fox et al19	Yes	No
Maas et al20	Yes	Yes
Copley et al21	Yes	No
DeMeyer et al22	No	Yes
Beckerman et al23	Unable to determine	No
Harvey et al24	Yes	Yes
Kerem et al25	No	No
Harvey et al26	Yes	Yes
Theis et al27	No	Yes
Harvey et al28	Yes	Yes
Cheng et al29	No	No
Lannin et al30	Yes	No
Basaran et al31	Yes	No
Moseley32	No	Yes
Pradines et al33	Yes	No
Lee et al34	No	No
Harvey et al35	Yes	Yes
Hill36	Yes	No
Lannin et al37	Yes	Yes
Smedes et al38	No	No
Horsley et al39	Yes	Yes
An and Jo40	Unable to determine	No
Electrical stimulation
Pool et al41	No	No
Pool et al42	No	No
Sabut et al43	No	No
Bakaniene et al44	No	No
Malhotra et al45	Yes	Yes
Nakipoglu Yuzer et al46	Unable to determine	No
Leung et al47	Yes	No
Sabut et al48	Yes	No
Beaulieu et al49	Yes	No
Shockwave therapy
Manganiotti and Amelio50	No	No
Lee et al51	Yes	No
Wang et al52	Unable to determine	No
Gonkova et al53	Yes	Unable to determine
Moon et al54	No	No
Vidal et al55	Yes	No
Botox
Love et al56	No	No
Hawamdeh et al57	No	No
Rameckers et al58	Yes	No
Meythaler et al59	Yes	No
Tedroff et al60	Yes	No
Koman et al61	Yes	Unable to determine
Schasfoort et al62	Yes	No
El-Etribi et al63	No	No
Physical activity
Horsley et al64	Yes	Yes
Scholtes et al65	Yes	No
Schmid et al66	No	No
Rydwik et al67	Yes	No
Baik et al68	No	No
Lorentzen et al69	Yes	Yes
Kirk et al70	No	Yes
An and Won71	No	No
Teixeira-Machado and DeSantana72	Yes	No
Hemachitara et al73	Yes	No
Robot-assisted rehabilitation
Mirbagheri et al74	No	Yes
Waldman et al75	Unable to determine	Yes
Mirbagheri et al76	No	Yes
Franceschini et al77	Yes	No
Sale et al78	Yes	No
Other interventions
Rayegani et al79	No	No
Xu et al80	Yes	No
Lorentzen et al81	Yes	Yes
Mathew et al82	Yes	No
Velasco et al83	Unable to determine	No
Wayne et al84	Yes	No
Cheng et al85	No	No
Fosdahl et al86	Yes	No
Takeuchi et al87	No	No
Ghannadi et al88	Yes	No

NOTE. The information in this table corresponds to the results of questions 15 and 20 in the Downs and Black checklist.

Table 3 depicts the results of question numbers 15 and 20 of the Downs and Black checklist. Question number 15 concerns assessor blinding; question number 20 concerns whether joint mobility was measured objectively. The assessor was blinded in 39 trials and not blinded in 25 trials. We were unable to determine whether the assessor was blinded in 6 trials. We rated the primary outcome measure as objective in 18 trials and not objective in 50 trials. In 2 trials, we were unable to determine if the primary outcome measure was measured objectively. In 19 trials, joint mobility was measured using neither assessor blinding nor an objective measure. In 4 of the trials where we were unable to determine the use of assessor blinding, joint mobility was measured using a nonobjective measure.

Effect of stretch on joint mobility (fig 2, fig 3)

Short-term effect is defined as effects measured up to 1 week after the end of the intervention. Of the 22 trials investigating the short-term effect of stretch on joint mobility,19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40 we were able to obtain pre/post (±SD) measurements of PROM from 19 studies.19, 20, 21, 22,24, 25, 26,28, 29, 30, 31, 32, 33, 34, 35,37, 38, 39, 40 Three of these trials,, compared 2 types of stretch interventions with a control situation. For these trials, we combined the experimental groups into 1 single group. The short-term effect of stretch intervention on joint mobility was investigated by pooling data from 17 RCTs with available data. Stretch had a pooled effect of 3° (95% CI, 1-4°; prediction interval (PI)=−2 to 7°; I=66%; P<.001). To explore differences in types of stretch, we explored the use of subgroup analysis. Here, we divided RCT studies in a casting/splinting subgroup and a stretching subgroup (including passive stretching protocols, self-stretching protocols, etc) (see fig 3). The effect of casting/splinting was 2° (95% CI, 0-5°) and the effect of stretching was 3° (95% CI, 1-5°).

Fig 3

Forest plot with subgroups showing the mean difference with 95% CI for short-term effects of stretch on joint mobility. Stretching includes interventions such as passive stretching and self-stretch protocols.

Effect of shockwave therapy on joint mobility (fig 4)

Of the 6 included trials investigating the effect of shockwave therapy on joint mobility,50, 51, 52, 53, 54, 55 we were able to obtain pre/post (±SD) measurements of PROM from 5 studies.50, 51, 52, 53, 54, However, only 1 of these studies was an RCT. The single RCT study had a short-term effect of 2° (95% CI, −5 to 10°; P=.56).

Effect of physical activity on joint mobility (fig 5)

Of the 10 trials investigating the effect of physical activity,64, 65, 66, 67, 68, 69, 70, 71, 72, 73 we obtained pre/post (±SD) measurements of PROM from 9 studies.64, 65, 66, 67, 68, 69, 70, 71, 72 The short-term effect of physical activity on joint mobility was investigated by pooling data from 7 RCTs with available data. Physical activity had a pooled effect of 3° (95% CI, −2 to 8°; PI=−15 to 20°; I2=87%; P=.28).

Effect of BTX on joint mobility (fig 6)

Of the 8 included trials investigating the effect of BTX on joint mobility,58, 59, 60, 61, 62, 63, 64, 65 we were able to obtain pre/post (±SD) measurements of PROM from 6 studies.58, 59, 60, 61, 62, 63 The short-term effect of BTX on joint mobility was investigated by pooling data from 5 RCTs with available data. BTX had a pooled effect of 4° (95% CI, −1 to 8°; PI=−13 to 20°; I=85%; P=.13).

Effect of electrical stimulation on joint mobility (fig 7)

Of the 9 included trials investigating the effect of electrical stimulation on joint mobility,56, 57, 58, 59, 60, 61, 62, 63 we were able to obtain pre/post (±SD) measurements of PROM from 8 studies.41, 42, 43, 44, 45, 46, 47, The short-term effect of electrical stimulation on joint mobility was investigated by pooling data from 5 RCTs with available data. Electrical stimulation had a pooled effect of 3° (95% CI, −1 to 6°; PI=−8 to 13°; I2=78%; P=.11).

Effect of robot-assisted rehabilitation on joint mobility (fig 8)

Of the 5 included trials investigating the effect of robot-assisted rehabilitation on joint mobility,74, 75, 76, 77, 78 we were able to obtain pre/post (±SD) measurements of PROM from 3 studies.,, The short-term effect of robot-assisted rehabilitation on joint mobility was investigated by pooling data from 5 RCTs with available data. Robot-assisted rehabilitation had a pooled effect of 1 (95% CI, −0 to 2; PI=−8 to 9; I2=73%; P=.03).

Effect of other interventions on joint mobility

Of the 70 included trials, 10 were not of the abovementioned intervention types. Rayegani et al found significant improvements in hip and ankle PROM after a 2-month passive cycling intervention in individuals with spinal cord injury. Xu et al investigated the effect of 4 weeks of mirror therapy or mirror therapy plus neuromuscular electrical stimulation. Compared with a control group, they found a significant effect of both interventions on ankle dorsiflexion PROM. Mathew et al investigated the effect of the antispasticity drug diazepam in children with cerebral palsy. After 15-20 days of intervention, they found a significant increase in PROM in the group receiving a large dose of diazepam but not in groups receiving placebo treatment or low-dose treatment. Wayne et al investigated the effect of up to 20 sessions of traditional Chinese acupuncture in adults with chronic hemiparesis after stroke. After treatment, they found significant increases in some but not all PROM measures in the acupuncture group compared with the control group. Ghannadi et al investigated the effect of dry needling in adults with stroke and found significant improvements of dorsiflexion PROM after treatment compared with the control group. Trials investigating the effect of neural tension technique, serious games, whole body vibration, stretch combined with resistance training, and high-intensity pulse irradiation with near-infrared rays found no significant effects on joint mobility.

Sensitivity analysis

Table 4 depicts the results of the sensitivity analyses. In the sensitivity analyses, we examined the effect of randomization.

Table 4

Sensitivity analyses

Variables	Intervention Type
	Stretch	Shockwave Therapy	Physical Activity	Botox	Electrical Stimulation	Robot-Assisted Rehabilitation
Pooled effect	5° (2 to 7°) n=19	12° (4 to 21°) n=4	3° (−1 to 7°) n=9	2° (−2 to 7°) n=6	3° (1 to 6°) n=8	1° (0 to 2°) n=3
Randomization (studies with adequate sequence allocation)	3° (1 to 4°) n=17	2° (−5 to 10°) n=1	3° (−2 to 8°) n=7	4° (−1 to 8°) n=5	3° (−1 to 6°) n=5	1° (0 to 2°) n=3

NOTE. Pooled effect with all studies included in the analysis and with only randomized trials included. Results are presented as mean difference/standardized mean difference (95% CI). n=no. of studies included in analysis.

Discussion

In this systematic review, we aimed to determine whether the existing literature supports that nonsurgical treatment options can reduce muscle contractures in individuals with neurologic disorders. Through our systematic search, we found 70 trials (57 RCTs) eligible for inclusion; 22 trials (19 RCTs) on stretch interventions, 6 trials (2 RCTs) on shockwave interventions, 8 trials (7 RCTs) on BTX interventions, 9 trials (5 RCTs) on electrical stimulation interventions, 10 trials (8 RCTs) on physical activity interventions and 5 trials (5 RCTs) on robot-assisted interventions. Additionally, there were 10 single trials on other intervention types. Through meta-analysis and quality assessment, we did not find convincing evidence supporting the use of any nonsurgical treatment option.

Similarly to Harvey et al, we do not consider a treatment effect of <5° PROM as clinically important. From the only available RCT on shockwave therapy, we found a nonsignificant effect of 2°. By including the 4 available nonrandomized studies, there was a significant effect of 12° (CI, 4-21°) (see fig 4 and table 4). Based on the Downs and Black checklist, 1 trial was of low quality, 2 were of medium quality, and 2 were of high quality. Perhaps more importantly, 2 of the 5 trials used neither assessor blinding nor an objective measure of joint mobility, thus introducing a large possibility of bias. The trial reporting the largest short-term effect (30°) did not use assessor blinding, an objective measure of joint mobility, or randomization. Four of the 5 trials measured PROM of the ankle joint, 1 measured PROM of the wrist. Four studies used a single session of shockwave therapy, and 1 study used 3 sessions of shockwave therapy. Because of limited data, we were not able to investigate the long-term effect of shockwave treatment through meta-analysis. However, 4 trials did in fact investigate possible sustained effects at follow-up intervals.,,, Gonkova et al found an immediate significant effect of 14° after shockwave treatment; after 4 weeks the effect was 11° and still significant compared with before treatment. Moon et al found a significant 30° effect of the shockwave intervention; at the 4-week follow-up the effect was 20°, and at the 12-week follow-up the effect was 10°. They found significant differences between baseline and measurements immediately after and 4 weeks after intervention. They did not find a statistical difference between baseline and 12-week follow-up measurements. Manganotti et al found an immediate nonsignificant effect of 3°; at the 4-week follow-up this difference was 4° and still nonsignificant compared with baseline. Lee et al found an immediate nonsignificant difference in joint mobility of 2.33° between the control group and the shockwave group; at the 4-week follow-up this difference was 3.55° and still nonsignificant. Because all indications concerning the effect of shockwave therapy are based on only a few trials of limited quality, we encourage cautious interpretations of the results.

Fig 4

Forest plot showing the mean difference with 95% CI for short-term effects of shock wave therapy on joint mobility.

From RCTs on stretch and robot-assisted rehabilitation interventions, we found small, clinically nonimportant effects on joint mobility. The estimated effect of stretch interventions was 3° PROM (CI, 1-4°). This finding is roughly consistent with that of the most recent systematic Cochrane review on the effect of stretch interventions on joint mobility in individuals with neurologic disorders by Harvey et al. Harvey found no short-term effect of stretch (mean difference=2° (95% CI, 0-3°). The estimated effect of robot-assisted rehabilitation interventions was 1° PROM (95% CI, 0-2°). We did not find significant effects from RCTs on physical activity (P=.27), electrical stimulation (P=.11), or BTX interventions (P=.13) on joint mobility.

An important finding of this review was the lack of objective measures of muscle contractures found in many trials. Only 18 of the 70 included trials used objective measures of muscle contractures such as passive stiffness or torque-controlled goniometric measurements; most of these were trials investigating the effect of stretch. The remaining 52 trials measured PROM using primarily standard, non–torque-controlled goniometric measurements. Furthermore, these nonobjective measures were used in 23 trials without convincing use of assessor blinding, thus introducing a large possibility of bias. In future research in this field, we strongly advocate the use of objective, instrumented measures such as passive stiffness (eg, measured using the portable stiffness assessment device) or torque-controlled goniometric measurements.

Study limitations

As with all systematic review studies, there is a possibility of retrieval bias—the fact that potentially eligible trials might have been missed. To minimize retrieval bias we chose to use a broad search string, which we tested by its ability to identify already known eligible trials. This strategy resulted in a large amount of identified trials, but we hope that it minimized the amount of missed trials. We are aware of the fact that the inclusion of nonrandomized studies introduces a possibility for bias. To address this issue we based conclusions primarily on meta-analyses performed on RCTs only and performed sensitivity analyses investigating the effect of randomization. In the data extraction process, the reviewers doing the data extraction used subjective judgment to determine if the intervention was administered to treat muscle contractures. We acknowledge that doing this without objective and clear criteria is problematic but believe that this was the best possible solution. In the meta-analyses, we combined data from studies on different joints using absolute PROM measures. Although range of motion does differ between joints, we decided to maintain the use of an absolute outcome measure to ensure easy transferability and interpretation in a clinical setting. In all included trials, the severity of contractures at baseline may affect the effect of the intervention. Unfortunately, we were not able to quantify the severity of contractures at baseline because the included trials used different measurement tools, investigated different joints, etc. Similarly, past treatment history is likely to influence the effect of the intervention. Because only a very limited number of studies included information on treatment history, we were not able to include this information. This is therefore a limitation to the study. A possibility of bias is also introduced because 2 of the authors (J.L., J.B.N.) of this review were also authors of included trials. We addressed this possibility of bias by not letting authors extract data from trials in which they had been involved. Despite the fact that all trials were screened by 2 authors and arbitrated by a third review author in case of unsolvable disagreement, we acknowledge the possibility of selection bias in systematic reviews such as this.

Conclusions

The central findings of this systematic review are that effective, nonsurgical treatment of muscle contractures is yet to be convincingly achieved and that there is a need for the use of objective measures of muscle contractures. Future research in this field should focus on the use of an objective measure of muscle contractures, thereby increasing the validity of the trials. We believe that the implementation of such objective measures would advance the continued search for effective, nonsurgical treatment of muscle contractures in individuals with neurologic disorders.

Supplier

Review Manager (RevMan) [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014.