Effects of Sling Exercise for Neck Pain: A Systematic Review and Meta-Analysis

Abstract

Objective

The purpose of this review was to evaluate the effects of sling exercise on pain intensity, disability, and health-related quality of life in adults with neck pain.

Methods

The Cochrane Central Register of Controlled Trials, EMBASE, Physiotherapy Evidence Database (PEDro), and 6 other databases were searched from inception to July 2020. The reference lists of relevant articles to identify additional trials were also screened. Randomized controlled trials were included if they investigated the effects of sling suspension therapy in patients with neck pain, including mechanical neck disorders, cervicogenic headache, and neck disorders with radicular findings. Studies were required to be published in English or Chinese. The methodological quality and levels of evidence of studies were assessed using the PEDro scale and the Grading of Recommendations Assessment, Development and Evaluation approach, respectively. The random-effects model was used to perform meta-analyses.

Results

Eleven randomized controlled trials were included (n = 595). The mean total PEDro score was 4.64 (SD = 1.21) of 10, which indicated a fair methodological quality. The intervention groups showed significant improvements in pain intensity (SMD = −1.23; 95% CI = −1.88 to −0.58) immediately postintervention compared with the control groups. No significant effects were found for disability, cervical range of motion, and health-related quality of life. However, sensitivity analyses revealed significant short-term improvements in pain intensity, disability, and cervical range of motion and sustained effects on disability at intermediate-term follow-up.

Conclusion

Sling exercise appears to be beneficial for improvements in pain intensity (moderate- to low-level evidence) among patients with neck pain. However, no definitive conclusion could be made regarding the effect of sling exercise for neck pain due to methodological limitations and high heterogeneity in the included studies.

Impact

This review provides overall moderate- to very low-level evidence for health care professionals who may consider including sling exercise in the intervention program for patients with neck pain.

Disability, Exercise Therapy, Neck Pain, Systematic Review, Sling Exercise

Introduction

Neck pain is a prevalent condition and has a 1-year prevalence of 30% to 50% and a lifetime prevalence of 14.2% to 71%.^1–3 People with neck pain demonstrate various physical impairments, such as reduced range of motion (ROM),⁴ increased muscle fatigue,^5,6 decreased proprioception,⁷ and impaired motor control.⁸ They may also have difficulties with daily activities and be limited in work and social participation.⁹ Overall, a proportion of patients with neck pain recover within 1 year; however, 37.3% of people experience persistent neck pain and 22.8% have recurrent neck pain.^10–13 Therefore, it is essential to improve the efficiency of treatment for neck pain.

In a recently updated clinical practice guideline,¹⁴ exercise is recommended as an effective intervention for neck pain. Sling exercise is a specific form of exercise established by Meier to rehabilitate professional German athletes and later developed by Kirkesola to treat motor problems.¹⁵ Sling exercise is performed with a special sling system that suspends single or multiple body regions. The traditional sling exercise is used to eliminate the gravitational force for ease of performing ROM or strengthening exercises.¹⁶ Sling exercise has been further developed and commonly used in diagnosis and treatment for musculoskeletal disorders to regain normal movement patterns.¹⁷ Through different applications, the sling system can be used to identify the unstable regions or motor control deficits and to provide specific exercise training.¹⁷

Sling exercise utilizes various elements to maximize its therapeutic effect for motor control training, including rope level, vibration, and position.¹⁷ When sling exercise is performed, a single or multiple regions of the patient’s body are fully supported by the specially designed ropes to allow painless performance of exercise or to assist alleviating existing pain. The setup of the ropes allows the patient to use his or her own body weight as resistance for training and frees up the therapist’s hands for manual facilitation. When sling exercise is performed, controlled vibration can be applied to selected body regions by manually tapping the ropes or applying a vibration apparatus. Although the exact mechanisms of vibration are still not fully understood, previous studies have found that performing exercises on a sling with vibration significantly increased muscle activations compared with no vibration.^18–21 The sling system allows simultaneous training of different muscles through open- and closed-kinetic-chain exercises in various positions.^22,23 The level of difficulty of closed-kinetic-chain exercises can be easily increased or modified by adjusting the length of the used ropes and/or moving the location of the sling suspension points.

Sling exercise is a common form of exercise. There is growing popularity of sling exercise to treat musculoskeletal disorders.¹⁷ Previous studies have also demonstrated potential benefits of sling exercise on pain intensity,^24–26 disability,^24,25,27 ROM,^28,29 and health-related quality of life (HRQoL)^30,31 in patients with neck pain or low back pain. A systematic review of 9 randomized controlled trials (RCTs) concluded that sling exercise was more effective than thermomagnetic therapy at reducing pain and improving function in patients with chronic low back pain.¹⁵ However, to our knowledge, no systematic review and meta-analysis has been conducted to critically evaluate the effects of sling exercise for neck pain. Therefore, the aim of this review was to assess the effects of sling exercise in adults experiencing neck pain.

Methods

Protocol and Registration

This systematic review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA Statement^32,33 and registered on the International Prospective Register of Systematic Reviews (registration no. CRD 42020157592). The Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0, was followed.³⁴

Study Selection

Types of Studies

Only published RCTs were included. Quasi-RCTs, crossover trials, cluster trials, and controlled clinical trials were excluded. Studies published in English or Chinese were included.

Types of Participants

Participants included in the review were adults (aged 18 years or older) who had neck pain for any duration. Neck pain was categorized as mechanical neck disorders, such as whiplash associated disorders, myofascial neck pain, and neck pain associated with degenerative changes; cervicogenic headache; and neck disorders with radicular findings. Studies were excluded if they investigated neck pain with definite or possible long tract signs (eg, myelopathies); neck pain caused by other pathological entities; headache not of cervical origin; coexisting headache when neck pain was not dominant or when neck movements or sustained neck postures did not provoke headache; and “mixed” headaches.

Types of Interventions

Studies were included if they used sling exercise in the treatment of neck pain. Comparison groups accepted in this review were no intervention, placebo, another intervention, and sling exercise plus another intervention.

Types of Outcome Measures

The primary outcome was pain intensity measured by self-report scales (eg, visual analog scale [VAS], numeric rating scale). The secondary outcomes were disability measured by questionnaires (eg, Neck Disability Index [NDI] or other neck pain–specific functional questionnaires), cervical ROM, and HRQoL assessed by questionnaires. The duration of follow-up was categorized as short term (immediately postintervention) and intermediate term (at least 1 month after completion of the intervention).

Data Sources and Searches

The following databases were searched from their inception to July 2020: PubMed (via NLM), EMBASE (via Elsevier), The Cochrane Central Register of Controlled Trials (via Wiley Online Library), Cumulative Index to Nursing and Allied Health Literature (via EBSCO), ProQuest, Physiotherapy Evidence Database (PEDro), Wanfang Data, China National Knowledge Infrastructure, and Airiti Library. The last search was conducted on July 24, 2020. We used a combination of relevant MeSH terms and free text words according to the recommendations given by the Cochrane Back and Neck Review Group,³⁵ including search terms for anatomical terms (eg, cervical vertebrae), syndrome terms (eg, neck pain), treatment terms (eg, sling), and methodological terms (eg, RCTs). The search strategy is presented in Supplementary File 1. No publication date or language restrictions were applied. We also screened reference lists of all retrieved full-text articles for additional studies.

Once duplicates were removed, 2 reviewers independently screened the titles and abstracts (where available) of the studies identified in the literature search and selected studies based on the pre-specified eligibility criteria. Full-text articles were then retrieved and reviewed for eligibility. If there was any disagreement, a decision was made by discussion or consultation with the third reviewer when necessary. Authors were contacted if the full-text articles could not be accessed online.

Data Extraction

One reviewer extracted relevant data using the pre-piloted data extraction form, and the second reviewer cross-checked the data independently.³⁶ Any discrepancies were resolved by discussion or consultation with the third reviewer. We contacted study authors if more information was needed. The following data were extracted: first author’s name, year of publication, eligibility criteria, number of participants, sex, age, symptoms duration, interventions, time of outcome assessment, outcome measures, and results. Where data were presented graphically only, and we were unable to obtain original data from the authors, the means and SDs were estimated from the graphs. The graphs were printed out, and the scale and data points were carefully measured with a standard ruler.

Quality and Certainty of Evidence Assessment

Two reviewers independently assessed the methodological quality of included studies using the PEDro scale³⁷ and then verified the scoring with those published in PEDro when available. The PEDro scale is a valid and reliable checklist^37–39 consisting of 11 items (ie, eligibility criteria, random allocation, concealed allocation, baseline comparability, masked participants, masked therapists, masked assessors, adequate follow-up, intention-to-treat analysis, between-group comparison, and point estimates and variability); the eligibility criteria do not contribute to the total score.³⁷ Each article was rated with a score from 0 to 10, with a score below 4 considered to be of “poor” methodological quality, a score between 4 and 5 indicating “fair” quality, a score of 6 to 8 considered to be of “good” quality, and a score of 9 to 10 indicating “excellent” quality.⁴⁰ Any disagreements were resolved by discussion between 2 reviewers. The Grading of Recommendations Assessment, Development and Evaluation framework⁴¹ was used to guide assessment of the certainty of the findings from quantitative evidence in the review.

Data Synthesis and Statistical Analysis

Data were analyzed with SPSS 22.0 software (IBM Corp, Armonk, NY, USA). Kappa statistics and percentage agreement were calculated to evaluate agreements between 2 reviewers on study selection and quality assessment, with a kappa value of 0.8 or higher indicating excellent agreement.⁴² Meta-analyses were conducted using Review Manager (RevMan) version 5.3 for Windows (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark) for each outcome of interest if the outcome was assessed in at least 2 studies using the same or similar measures. The mean difference, the SD of the mean difference, and the number of participants in each group were entered into RevMan and analyzed. The mean difference was calculated by subtracting the baseline mean from the follow-up mean. The SD of the mean difference was imputed by applying the equations published in the Cochrane Handbook for Systematic Reviews of Interventions³⁴ and using a conservative correlation coefficient of r = 0.5 as suggested by Follmann et al.⁴³

The random-effects model was used to perform meta-analyses. The effect size was expressed as the weighted mean difference and 95% CI if all included studies measured the outcome on the same scale or as standardized mean differences with 95% CIs when different scales were used to measure the same concept. If sufficient data were available, all outcomes were analyzed in separate time frames: short term (immediately postintervention) and intermediate term (1 month or more postintervention). Heterogeneity among the included studies was assessed using the I² statistic, with high heterogeneity determined as an I² > 50%.³⁴ Statistical significance was set at a P < .05. Forest plots were produced to show the outcomes between the intervention and control groups. Sensitivity analyses were carried out to determine the robustness of results by removing studies that presented results in graphical form without numeric data, which may have had an impact on the accuracy and precision of effect estimate. Where possible, funnel plots were used to assess for publication bias.³⁴

Results

Selection of Studies

The search strategy retrieved a total yield of 11,020 studies. Duplicates (3119 studies) were removed. After screening titles and abstracts, the full texts of 20 studies were assessed for eligibility.^{26,27,29–31,44–58} Eleven studies^{26,27,29–31,47–50,56,57} met the inclusion criteria. No further studies suitable for inclusion were identified via hand searching. Figure 1 presents the selection process and includes the reasons for study exclusion.

Figure 1

Flow diagram of the review process.

The agreement between the 2 independent reviewers for study selection was excellent for both titles/abstracts (kappa = 1.00; percentage agreement = 99.98%) and full text (kappa = 1.00; percentage agreement = 100%). Tests for funnel plot asymmetry (publication bias) were not possible due to the small number of studies included in the meta-analyses.³⁴

Study Characteristics

Studies were published between 2007 and 2020, and there were a total of 595 participants across the 11 included studies.^{26,27,29–31,47–50,56,57} Eight of the 11 studies were written in Chinese.^{26,27,31,48–50,56,57} The other 3 studies were written in English.^29,30,47 Follow-up time points ranged from immediate to 12 months postintervention. The characteristics of included studies are detailed in Table 1.

Table 1

Summary of Included Studies^a

Study	Inclusion Criteria	Participants			Pain Duration	Interventions	Outcome Measures	Time Points	Results
		No.	Age (y)	Sex (Men/Women)
Vikne et al, 200729	Experienced traffic accident 6–12 mo previously Presence of grade 1 or 2 WAD symptoms	171 (NSET: 40, NSET+: 44, TP: 42, TP+: 45)	18–60	57/114	NR	CGb: traditional exercise performed for 4 mo and home training begun after 3 wk; all training stopped at 4 mo CG+: in addition to intervention in CG, home exercises continued after 4 mo of intervention IGb: 24 sessions of sling exercise performed within 4 mo and home training with sling performed after 3 wk IG+: similar to IG, except home exercises continued after 4 mo of intervention	VAS, modified version of Roland-Morris Disability Questionnaire, sick leave, HSCL, ROM, neck stability, cervicocephalic kinesthetic sensibility	Preintervention Postintervention (4 mo) follow-up (12 mo)	Because of dropouts, only 171 participants were included for data analysis VAS: no significant difference between groups (0.07–0.82) Other outcome measures: no significant improvement between groups
Yu et al, 201131	Nonspecific neck pain inpatients and outpatients in Worker Hospital of China’s 19th Metallurgy Group Ltd. Co.	78 (2 groups of 39)	18–60	CG: 20/19 IG: 18/21	NR	CG: IFC IG: SET training Both groups received intervention, 20 min/time, 1 time/d, 5 d/wk, for 20 d	NNP, NDI, SF-36	Preintervention (10 d) Postintervention (20 d)	All outcome measures improved significantly after treatment between groups (P < .01) NNPc IG: pre = 39.33 (11.54), post = 5.93 (12.54) CG: pre = 37.59 (8.87), post = 14.66 (14.31) NDIc IG: pre = 19.23 (10.64), post = 4.97 (5.86) CG: pre = 18.66 (9.12), post = 12.21 (10.83) SF-36c IG: pre = 462.80 (114.43), post = 614.13 (85.20) CG: pre = 484.72 (79.41), post = 606.66 (83.60)
Yang et al, 201427	Fulfill 3/4 criteria: aggravation with testing for spinal stenosis, symptom relief with distraction, aggravation at ULTT, and unilateral cervical rotation of <60° Age 20–65 y Symptoms for <3 mo First diagnosed as cervical radiculopathy Voluntarily participate and sign consent form	60 (2 groups of 30)	20–65	N/A	NR	CG: cervical traction (20 min), manual therapy (20 min), and IFC (20 min); 1 time/d, 5 d/wk, for 2 wk IG: additional Neurac training for 20–30 min, 3 times/wk	NPRS, NDI	Preintervention (wk 1) Postintervention (wk 2)	Both groups showed significant improvement in NPRS and NDI (P < .05) at 1 wk and 2 wk after treatment, with IG showing greater improvement than CG (P < .05) NPRSc IG: pre = 6.65 (1.19), week 1 = 4.97 (1.12), week 2 = 4.03 (1.31) CG: pre = 6.36 (1.12), week 1 = 5.52 (0.80), week 2 = 4.52 (0.92) NDIc IG: pre = 19.43 (2.80), week 1 = 12.93 (3.17), week 2 = 9.30 (4.14) CG: pre = 18.2 (3.02), week 1 = 13.72 (2.71), week 2 = 9.75 (3.29)
Chen, 201526	Neck pain and discomfort Neck tightness and trigger point No cervical surgery in 3 mo Spinal instability and stenosis	40 (2 groups of 20)	NR	CG: 3/17 IG: 8/12	NR	CG: Maitland mobilization in each direction, 30 min/time IG: mobilization and SET training (neck movement in each direction for 5 s, 5 repetitions/set for 3 sets, rest for 30 s between sets), 15 min/time Both groups received intervention 3 times/wk for 3 wk	VAS, clinical efficacy	Preintervention Postintervention (3 wk)	Both groups showed significant improvement in VAS (P < .05) IG showed greater improvement than CG (P < .05) Clinical efficacy revealed significantly greater improvement in IG than in CG (P < .05) VASc IG: pre = 6.5 (1.21), post = 3.1 (0.45) CG: pre = 6.4 (1.12), post = 4.6 (0.53)
Liu et al, 201550	Cervical spondylosis met manifestation and image criteria posted in second cervical pathology conference Age 18–60 y No infection, tumor, or rheumatoid arthritis	42 (2 groups of 21)	18–60	17/25	3 to 56 mo	CG: IFC, 20 min/time, 1 time/d, daily for 4 wk IG: additional SET training (neck movement in each direction for 3 s, 5 repetitions/set for 6 sets, rest for 30 s between sets), 15 min/time, 1 time/d, 2 d/wk	VAS, NPQ	Preintervention Postintervention (4 wk) Follow-up (1 mo)	IG showed significantly lower score than CG in VAS and NPQ after 1 mo of therapy (P < .05) VAS showed significant improvement in both groups after 1-mo follow-up VASc IG: pre = 4.61 (0.65), post = 2.78 (0.59), follow-up = 2.81 (0.27) CG: pre = 4.62 (0.71), post = 3.64 (0.48), follow-up = 4.59 (0.92) NPQc IG: pre = 38.56 (14.63), post = 13.57 (8.92), follow-up = 13.91 (8.14) CG: pre = 38.29 (15.76), post = 24.69 (9.74), follow-up = 38.64 (13.58)
Yun et al, 201530	NDI score >5 Symptoms for >3 mo	20 (2 groups of 10)	IG: 38.10 (12.25)c CG: 35.20 (10.08)c	CG: 2/8 IG: 2/8	NR	CG: traditional physical therapy, 30 min/time 3 times/wk for 4 wk IG: additional Neurac training, 30 min/session	VAS, NDI, Biorescue Test, questionnaire for fatigue symptoms standardized by Industrial Fatigue Research Group of Japan Society of Industrial Hygiene, SF-36	Preintervention Postintervention (4 wk)	VAS, NDI, fatigue, and SF-36 showed significant improvements in IG compared with CG after intervention VASc IG: pre = 4.95 (1.55), post = 2.56 (1.32) CG: pre = 4.91 (1.26), post = 4.85 (1.49) NDIc IG: pre = 12.50 (4.55), post = 7.10 (4.18) CG: pre = 12.00 (4.47), post = 11.60 (3.95)
Oh and Yu, 201647	Straight neck syndrome	20 (2 groups of 10)	20s	N/A	NR	CG: stretching exercises for sternocleidomastoid and upper trapezius muscles IG: sling stabilization exercise (cervical flexion-extension and lateral bending exercises, 15 repetitions/set for 3 sets, rest for 30 s between sets) Both groups received intervention, 60 min/time 3 times/wk for 6 wk	CVA, CRA, cervical ROM in every direction	Preintervention Postintervention (6 wk)	IG showed significant changes in CVA, CRA, and ROM in every direction except extension after 6 wk of treatment (P < .05)
Peng et al, 201648	Neck pain VAS score >3 and NPQ score >25% Previous neck pain in 30 d Neck pain in latest 7 d	32 (2 groups of 16)	IG: 22.36 (1.95)c CG: 23.64 (1.69)c	0/32	NR	CG: routine neck-shoulder protection health education and regular neck-shoulder training, daily for 6 wk IG: SET exercises for neck extension and flexion, rotation, lateral flexion, and scapula elevation and protraction; 20–30 min/d 3 d/wk for 6 wk	VAS, NPQ, EMG of SCM, UT, and SC	Preintervention Postintervention (6 wk)	IG showed significant differences in VAS, NPQ, and EMG after 6 wk of therapy VASc IG: pre = 5.89 (1.18), post = 3.89 (0.82) CG: pre = 6.12 (0.80), post = 5.95 (0.70) NPQc IG: pre = 34.96 (6.90), post = 22.03 (4.5) CG: pre = 33.97 (5.41), post = 30.95 (5.49)
Zhang and Yang, 201649	Car accidents Presence of neck pain and dysfunction Symptoms >1 mo and <2 mo	22 (2 groups of 11)	21–47	CG: 9/2 IG: 8/3	CG: 39.5 (8.2) dc IG: 39.1 (8.3) dc	CG: acupuncture and modality therapy IG: additional sling exercises Both groups received intervention, 4–6 sessions/d 6 times/wk for 6 wk	VAS, NDI, FABQ	Preintervention Postintervention (6 wk)	Both groups showed significant improvements after treatment (P < .05) IG showed significant improvements in VAS, NDI, and FABQ compared with CG (P < .05) VASc IG: pre = 6.61 (1.12), post = 1.01 (0.81) CG: pre = 6.81 (1.23), post = 3.12 (1.15) NDIc IG: pre = 28.81 (7.62), post = 3.12 (1.16) CG: pre = 29.41 (8.12), post = 15.13 (4.35) FABQc IG: pre = 75.23 (6.75), post = 15.31 (6.85) CG: pre = 74.34 (6.12), post = 35.21 (5.48)
Rao and Zhong, 201756	Presenting with symptoms related to chronic neck pain such as discomfort, ROM limitation, and pain for >6 mo Received treatment first time Cervical image showed curvature change, instability, stenosis, or osteophyte Signed consent form	46 (2 groups of 23)	NR	N/A	NR	CG: postural education in sitting IG: maintaining neutral position during SET in supine position Both groups received intervention, 20 min/session, 3 times/d	EMG, pain scale	Preintervention Postintervention	IG showed significant improvements in both EMG and pain scale (P < .05) compared with CG after intervention Painc IG: pre = 6.15 (1.25), post = 2.1 (0.65) CG: pre = 6.31 (1.17), post = 5.31 (1.05)
Zhao and Song, 201957	Neck pain present >3 mo Age 21–52 y VAS score 3–7	64 (2 groups of 32)	21–52	CG: 17/15 IG: 14/18	CG: 4.1 (0.6) moc IG: 4.2 (0.8) moc	CG: manual therapy IG: use of sling to train open chain, closed chain, and muscle coordination in every direction Both groups received intervention 3 times/wk for 2 wk	ROM, clinical effects, NPQ	Preintervention Postintervention (2 wk) 2 mo after intervention	IG showed significant improvements in cervical ROM (flexion, extension, and side bending) compared with CG after treatment (P < .05) There was no significant difference between 2 groups in NPQ scores after treatment; however, NPQ scores were significantly different between groups at 2-mo follow-up (P < .05)

^aCG = control group; CG+ = control group + additional home exercises; CRA = cranial rotation angle; CVA = craniovertebral angle; EMG = electromyography; FABQ = Fear Avoidance Beliefs Questionnaire; HSCL = Hopkins Symptom Checklist for Psychological Distress; IFC = interferential current therapy; IG = intervention group; IG+ = intervention group + additional home exercises; N/A = not available; NDI = Neck Disability Index; NNP = nonspecific neck pain meter; NPQ = Northwick Park Pain Questionnaire; NPRS = numeric pain rating scale; NR = not reported; NSET = new sling exercise therapy; NSET+ = new sling exercise therapy + additional home exercises; pre = preintervention; post = postintervention; ROM = range of motion; SC = splenius cervicis; SCM = sternocleidomastoid; SET = sling exercise therapy; SF-36 = Medical Outcomes Study 36-Item Health Survey Questionnaire; TP = traditional physical therapy; TP+ = traditional physical therapy + additional home exercises; ULTT = Upper Limb Tension Tests; UT = upper trapezius; VAS = visual analog scale; WAD = whiplash-associated disorders.

^bData used in meta-analysis.

^cMean (SD).

Participant Characteristics

Seven studies included participants with neck pain,^{26,30,31,48,49,56,57} 1 study included participants with whiplash-associated disorder symptoms,²⁹ 1 study included participants with cervical radiculopathy,²⁷ 1 study included participants with cervical spondylosis,⁵⁰ and 1 study included participants with straight neck syndrome.⁴⁷ Only 3 studies^49,50,57 reported the duration of symptoms, which ranged from 39.1 days to 56 months. The age of the participants ranged from 18 to 65 years. Six of the 11 studies included more female participants than male participants.^{26,29,30,48,50,57} The characteristics of participants are presented in Table 1.

Outcome Measures

Various instruments were used to assess the intervention effects. Pain intensity was measured using the VAS (8/11)^{26,27,29,30,48–50,56} and the nonspecific neck pain meter (1/11).³¹ The most commonly used assessment tools for measuring disability due to neck pain among the included studies were the NDI (4/11)^27,30,31,49 and the Northwick Park Pain Questionnaire (3/11).^48,50,57 Three studies evaluated cervical ROM using a standard goniometer^47,57 and a Cervical Measurement System.²⁹ Two studies^30,31 used the Medical Outcomes Study 36-Item Health Survey Questionnaire to assess HRQoL. Outcome measures are presented in Table 1.

Intervention

All studies compared sling exercise with another intervention, mainly traditional physical therapy (eg, manipulation, mobilization, massage, electrotherapy, cervical traction, acupuncture, stretching exercises, ROM exercises, and postural training).^{26,27,29–31,47–50,56,57} Intervention duration varied from 2 weeks to 4 months, and intervention frequency ranged from twice per week to 3 times per day. The types of sling exercise varied among the studies. Two studies did not state the details of the type of sling exercise provided.^27,31 Vikne et al provided 10 graded specific exercises in a ceiling-mounted sling.²⁹ Chen et al and Liu et al provided sling exercise therapy, which included active cervical ROM exercises with participants lying in supine with knee flexion, slings supporting around the thorax and pelvis, and head resting on a cushion.^26,50 Yun et al provided Neurac training with cervical movements using a sling in the supine and prone positions.³⁰ Oh et al provided sling stabilization exercises, which comprised cervical flexion-extension and lateral bending exercises with the sling equipment.⁴⁷ Peng et al provided sling exercises, which included cervical flexion, extension, side bending, rotation, and shoulder elevation and protraction in the supine, side-lying, and kneeling positions.⁴⁸ Zhang et al provided cervical nodding exercises and resistance exercises with a sling.⁴⁹ Rao et al provided cervical, thoracic, and pelvic exercises using a sling to maintain the neutral spine position in supine.⁵⁶ The intervention provided by Zhao et al included open-kinetic-chain exercises, stretching, and closed-kinetic-chain exercises with a sling.⁵⁷  Table 1 summarizes detailed information on the interventions.

Quality and Certainty of Evidence Assessment

The mean total PEDro score was 4.64 (SD = 1.21; range = 3 to 6) of 10, which indicated an overall fair quality of included studies (Tab. 2). All included studies met the item of point estimate and variability reported. None of the included studies met the criteria of therapist masking, participant masking, and intention-to-treat analysis. The agreement between the 2 reviewers for assessment of study quality was excellent (kappa = 1.00; percentage agreement = 100%). Assessment details of the levels of evidence with the Grading of Recommendations Assessment, Development and Evaluation approach are presented in Supplementary File 2 for each reported outcome.

Table 2

Quality Analysis of the Included Studies Using the PEDro Scoring System^a

PEDro Scale	Vikne et al, 200729	Yu et al, 201131	Yang et al, 201427	Chen, 201526	Liu et al, 201550	Yun et al, 201530	Oh and Yu, 201647	Peng et al, 201648	Zhang and Yang, 201649	Rao and Zhong, 201756	Zhao and Song, 201957
Eligibility criteria	Y	N	Y	N	Y	Y	N	Y	Y	Y	Y
Random allocation	Y	N	Y	N	Y	Y	Y	Y	Y	Y	Y
Concealed allocation	Y	N	N	N	N	N	N	N	N	N	N
Groups similar at baseline	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Participant masking	N	N	N	N	N	N	N	N	N	N	N
Therapist masking	N	N	N	N	N	N	N	N	N	N	N
Assessor masking	Y	N	Y	N	Y	N	N	N	Y	N	N
<15% dropouts	Y	N	Y	Y	Y	N	N	N	Y	N	Y
Intention-to-treat analysis	N	N	N	N	N	N	N	N	N	N	N
Between-group difference reported	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y
Point estimate and variability reported	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
Total score	6	3	6	4	6	4	3	4	6	4	5

^aN = no; PEDro = Physiotherapy Evidence Database; Y = yes.

Treatment Effects

Primary Outcome Measure: Pain Intensity

Nine studies^{26,27,29–31,48–50,56} were included to estimate the effect of sling exercise for pain intensity. The pooled results showed significant differences in pain reduction between groups in the short term (P < .001). The standardized mean difference was −1.23 (95% CI = −1.88 to −0.58) immediately postintervention (Fig. 2A), and the standardized mean difference was −1.04 (95% CI = −3.39 to 1.311) (Fig. 3A) at intermediate-term follow-up.

Figure 2

Short-term effects. (A) Meta-analysis of short-term changes in pain intensity in interventions vs controls. (B) Meta-analysis of short-term changes in disability scores in interventions vs controls. (C) Meta-analysis of short-term changes in cervical range of motion in interventions vs controls. (D) Meta-analysis of short-term changes in health-related quality of life in interventions vs controls. IV = inverse variance; NDI = Neck Disability Index; NNP = nonspecific neck pain meter; NPQ = Northwick Park Pain Questionnaire; Std. = standardized; VAS = visual analog scale.

Figure 3

Intermediate-term effects. (A) Meta-analysis of intermediate-term changes in pain intensity in interventions vs controls. (B) Meta-analysis of intermediate-term changes in disability scores in interventions vs controls. IV = inverse variance; NPQ = Northwick Park Pain Questionnaire; Std. = standardized; VAS = visual analog scale.

Secondary Outcome Measures: Disability

Disability was assessed in 8 studies^{27,29–31,48,49,57,59} using various questionnaires as outcome measures. The standardized mean differences were −0.45 (95% CI = −1.22 to 0.32) (Fig. 2B) and −12.41 (95% CI = −27.52 to 2.70) (Fig. 3B) in the short term and intermediate term, respectively.

Secondary Outcome Measures: Cervical Range of Motion

The pooled results of 3 studies^29,47,57 showed no significant difference between 2 groups in the improvement of cervical ROM in the short term. The standardized mean difference was −0.07 (95% CI = −0.40 to 0.26) (Fig. 2C).

Secondary Outcome Measures: Health-related Quality of Life

Two studies^30,31 assessed HRQoL (Fig. 2D). Pooling results across studies was not statistically significant in the short term. The standardized mean difference was −1.16 (95% CI = −2.96 to 0.65).

Sensitivity Analysis

The results in Vikne et al²⁹ were presented in graphs; hence, the figures used in meta-analyses were estimated from graphs. We conducted sensitivity analyses by excluding this study from the analyses, and the pooled results showed significant improvements in pain intensity (P = .001), disability (P = .03), and cervical ROM (P < .001) immediately postintervention compared with controls Supplemental File 3 and Supplemental Figures A, C, E. A sensitivity analysis of the intermediate-term effect on pain intensity, cervical ROM, and HRQoL was not possible due to insufficient data in the included trials. Following sensitivity analyses, the pooled standardized mean difference demonstrated significant improvements in disability scores at intermediate-term follow-up (P = .0010) (Suppl. File 3 and Suppl. Figures A, C, E).

Discussion

Moderate- to low-level evidence suggests that sling exercise when used alone^29,31,48,56 or combined with traditional physical therapy (eg, cervical traction, manual therapy, or electrotherapy)^26,27,30,50 may provide large beneficial effects on pain intensity immediately postintervention. However, the effect of pain relief did not last at intermediate-term follow-up.^29,50 Based on the results of sensitivity analyses, moderate- to low-level evidence suggests that there may be medium and large beneficial effects on disability with the use of sling exercise at short-term and intermediate-term follow-up assessments, respectively. Low-level evidence supports that there may be medium improvements in cervical ROM with the use of sling exercise at short-term follow-up. In addition, there may be little to no difference in HRQoL with the use of sling exercise.

Sling exercise may provide beneficial effects for neck pain for several reasons. Although the direct pathoanatomical cause of neck pain is unclear,^11,14 spinal instability has been considered as one of the important causes of neck pain.⁶⁰ Patients with chronic neck pain demonstrate motor control deficits in their deep stabilizing muscles, which is associated with impaired motor control of the cervical spine.⁸ The sling system may be used to identify the “weak link” in the kinetic chain of the cervical spine and to determine motor control deficits between deep stabilizing muscles and superficial mover muscles.¹⁷ The function of deep stabilizing muscles in the cervical spine region may also be measured and trained with the sling system.¹⁷ Oh et al reported that both sling stabilization exercise and stretching exercise are effective at improving the cervical ROM and alignment in patients with straight neck⁴⁷; the mechanism may be due to cocontraction of the masticatory and neck flexors muscles during neck stabilization exercise.⁶¹ A previous study investigating the effects of sling exercise for low back pain also showed that sling exercise resulted in a significantly higher activation of deep stabilizing muscles in the lumbar spine region (transversus abdominis) compared with traditional bridging exercise.⁶² In addition, static-dynamic cervicoscapulothoracic strengthening exercises, which is a major component of sling exercise, may improve pain intensity and function in patients with neck pain.⁶³ Sling exercise has been shown to achieve similar effects as McKenzie exercise on pain intensity, disability, muscle strength, and ROM in young adults with chronic neck pain.⁶⁴

Overall Completeness and Applicability of Evidence

Of the 11 studies included in this review, 8 studies adequately described the study population.^{26,27,29,30,48,49,56,57} Only 3 studies provided details about the symptoms duration.^49,50,57 Heterogeneous samples in terms of types of neck pain, symptom duration, age, sex, etc, may explain the inconsistency in study results and influence the applicability of evidence.

The majority of included studies did not report important intervention details. For example, Yu et al³¹ simply stated that sling exercise was used for the intervention group, 20 minutes per time, once per day, 5 d/wk, for 20 days. Similarly, Yang et al²⁷ reported that sling exercise was conducted based on a training manual. Vikne et al²⁹ explained 2 exercise examples for the sling exercise program comprising 10 graded exercises and did not report the number of exercise repetitions/sets/sessions, session duration, or exercise progression. The expertise of the person who provides the intervention may also affect the outcomes. However, only 1 study clearly explicated that physical therapists who received special training prior the intervention conducted the sling exercise program.²⁹ In addition, none of the included studies reported the number of adverse events and the extent to which sling exercise was delivered as planned. A checklist on reporting details of the intervention elements would be beneficial in future trials.⁶⁵

Five included studies provided the same control interventions with additional sling exercise to the intervention groups,^{26,27,30,49,50} and the remaining 6 studies offered the control groups alternative interventions.^{29,31,47,48,56,57} Using the same intervention provides a better understanding of additional effects of sling exercise to a treatment plan; however, the effect of sling exercise may be overestimated due to the interaction effect between sling exercise and the other intervention. This may explain different findings between our review and the previous systematic review,¹⁵ which examined the effects of sling exercise for low back pain. The review by Yue et al reported no clinical evidence to support beneficial effects of sling exercise compared with other forms of exercise.¹⁵

All included studies clearly reported the timing of outcome assessment. Of the 9 studies that reported pain intensity,^{26,27,29–31,48–50,56} 8 studies found that sling exercise has a positive effect on reducing pain intensity immediately postintervention.^{26,27,29–31,48,50,56} However, improvement in only 2 studies^30,56 reached the minimal clinically important difference of 20 mm on the VAS.⁶⁶ Of the 7 studies that reported disability,^{27,29–31,48,49,57} 4 studies used the NDI.^27,30,31,49 The NDI is the most widely used measure for neck-specific functions; however, the applicability of the NDI has been challenged in the recent literature. The NDI is not a unidimensional interval scale and exhibits an exceptionally large floor effect,^67,68 which can influence the effect estimate of sling exercise in the neck pain population. The development of a more appropriate measure of disability related to neck pain may be warranted for a future study. In addition, only 3 studies included intermediate-term follow-up assessments.^29,50,57 The sustained effects of sling exercise on pain intensity, disability, cervical ROM, and HRQoL need further investigation.

Quality Assessment

According to the PEDro results, the most frequent methodological flaws were the lack of masking of the participants and therapists, intention-to-treat analysis, and concealed allocation. None of the included studies used any masking method for the participants and therapists. However, it is difficult or even impossible to mask the participants and therapists during intervention involving exercise therapy. A pragmatic view of potential performance bias due to the lack of masking of the participants and therapists is necessary. None of the included studies used the method of intention-to-treat analysis to deal with the participants who did not receive the assigned intervention or follow-up assessments. Only 1 study satisfied the criteria of concealed allocation by using an external research institute for group assignment.²⁹ The lack of intention-to-treat analysis and adequate concealed allocation can result in attrition and selection bias, respectively. These issues can threaten internal validity of the studies and influence the inference on treatment effects.

Strengths and Limitations of the Review

This is the first systematic review, to our knowledge, to investigate the effects of sling exercise for neck pain by using a broad search strategy in both English and Chinese databases. Language bias was reduced during the study selection of the review process. We were also able to conduct sensitivity analyses to determine the robustness of the results.

Several limitations need to be acknowledged. All the included studies contain methodological limitations, which may influence the effect estimate of sling exercise for neck pain. The findings of this review need to be interpreted with caution because the included studies were limited to publications written in English or Chinese, which may have led to language bias. The sample sizes of the included studies in this review were relatively small, with only 4 studies^27,9,31,57 that enrolled more than 50 participants. The studies^30,47,49 with small sample sizes (<30 participants) could be underpowered to detect a subtle effect of sling exercise between the intervention and control groups. Given the limited number of studies available on this topic, subgroup analyses considering types of participants and duration of symptoms were not possible; hence, the effects of clinical heterogeneity on effect estimates remain unknown. Further limitations include performing the search strategy in the few databases with a small number of search terms and lack of searching trial registry platforms for ongoing studies. None of the included studies reported compliance of the intervention groups or monitored the impact of co-interventions provided to the control groups, which may increase the uncertainty of the effect estimate in this review.

Implications for Future Research

Due to the limitations in the included studies of this review, well-designed studies with larger sample sizes and long-term follow-up assessments are warranted to clarify the effect of sling exercise in people with neck pain. Future studies should clearly describe the eligibility criteria and use an adequate method for concealed allocation. Future studies should follow the participants for a longer period of time (≥6 months) and include assessment for impairments of body function and HRQoL.¹⁴ Researchers should adhere to the CONSORT guidelines or use the Template for Intervention Description and Replication checklist to improve the quality of reporting of RCT studies.

The findings of this systematic review provide overall a moderate- to very low-level of evidence of sling exercise for physical therapists or health care professionals in designing effective treatment plans for patients with neck pain. There is some evidence to suggest that sling exercise may be effective in reducing pain intensity (moderate- to low-level evidence), disability (moderate- to low-level evidence), and improving cervical ROM (low-level evidence) in adults with neck pain. However, no definitive conclusion could be made regarding the effect of sling exercise for neck pain due to methodological limitations and high heterogeneity in the included studies. Future research should aim to investigate the long-term effects of sling exercises on pain intensity, disability, and HRQoL.

Author Contributions

Concept/idea/research design: Y.-J. Tsai, Y.-L. Kuo

Writing: K.-Y. Lin, Y.-L. Kuo

Data collection: P.-Y. Hsu, C.-S. Tsai, Y.-L. Kuo

Data analysis: K.-Y. Lin, Y.-L. Kuo

Project management: Y.-L. Kuo

Consultation (including review of manuscript before submitting): K.-Y. Lin, Y.-J. Tsai

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Systematic Review Registration

This systematic review was registered on the PROSPERO International Prospective Register of Systematic Reviews (CRD 42020157592).

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.