Background: To date, optimal strategies for the management of patients with cervical radiculopathy remain elusive. Preliminary evidence suggests that a multimodal treatment program consisting of manual therapy, exercise, and cervical traction may result in positive outcomes for patients with cervical radiculopathy. However, limited evidence exists to support the use of mechanical cervical traction in patients with cervical radiculopathy.
Objective: The purpose of this study was to examine the effects of manual therapy and exercise, with or without the addition of cervical traction, on pain, function, and disability in patients with cervical radiculopathy.
Design: This study was a multicenter randomized clinical trial.
Setting: The study was conducted in orthopedic physical therapy clinics.
Patients: Patients diagnosed with cervical radiculopathy (N=81) were randomly assigned to 1 of 2 groups: a group that received manual therapy, exercise, and intermittent cervical traction (MTEXTraction group) and a group that received manual therapy, exercise, and sham intermittent cervical traction (MTEX group).
Intervention: Patients were treated, on average, 2 times per week for an average of 4.2 weeks.
Measurements: Outcome measurements were collected at baseline and at 2 weeks and 4 weeks using the Numeric Pain Rating Scale (NPRS), the Patient-Specific Functional Scale (PSFS), and the Neck Disability Index (NDI).
Results: There were no significant differences between the groups for any of the primary or secondary outcome measures at 2 weeks or 4 weeks. The effect size between groups for each of the primary outcomes was small (NDI=1.5, 95% confidence interval [CI]=−6.8 to 3.8; PSFS=0.29, 95% CI=−1.8 to 1.2; and NPRS=0.52, 95% CI=−1.8 to 1.2).
Limitations: The use of a nonvalidated clinical prediction rule to diagnose cervical radiculopathy and the lack of a control group without treatment were limitations of this study.
Conclusions: The results suggest that the addition of mechanical cervical traction to a multimodal treatment program of manual therapy and exercise yields no significant additional benefit to pain, function, or disability in patients with cervical radiculopathy.
The annual incidence of cervical radiculopathy (CR) has been reported to be 83 cases per 100,000 people in the population, with an increased prevalence noted in the fifth decade of life.1 This disorder is most commonly associated with a cervical disk derangement or other space-occupying lesion, resulting in nerve root inflammation, impingement, or both.1,2 Common signs and symptoms of CR include upper-extremity pain, paresthesia or numbness, weakness, or a combination of these signs and symptoms. Patients also may have scapular pain,3,4 headaches,5 and neck pain.6 Patients with both neck and upper-extremity symptoms have been reported to have greater functional limitation and disability than patients with neck pain alone.7
Diagnostic imaging (magnetic resonance imaging) and electrophysiological tests (nerve conduction velocity, electromyography) are commonly used to confirm a diagnosis of CR.8–11 Using nerve conduction velocity and electromyographic data as a gold standard, a clinical prediction rule (CPR) was derived to identify the presence of CR using a limited subset of variables from the clinical examination.12 The CPR for identifying CR includes the Spurling test, the distraction test, the Upper-Limb Tension Test 1 (ULLT1) (median nerve bias), and ipsilateral cervical rotation of less than 60 degrees. The CPR exhibited a specificity of 94% (positive likelihood ratio=6.1, 95% confidence interval [CI]=2.0 to 18.6) when 3 of 4 criteria were satisfied.
Physical therapy interventions often used for the management of CR include cervical traction, postural education, exercise, and manual therapy applied to the cervical spine and thoracic spine.13 Studies indicate that some combination of these interventions may result in improved outcomes for patients with CR.14–23 Previous controlled clinical trials investigating the treatment of patients with CR have not used the CPR as an inclusion criteria.14,15,17,23,24 To date, only 2 case series18,21 and a cohort study22 have examined standardized treatment programs in patients diagnosed with CR, using the previously defined CPR. The prospective cohort study identified predictor variables that can identify which patients with CR are likely to have short-term successful outcomes.22 A multimodal approach to management including manual therapy, cervical traction, and deep neck flexor strengthening was identified as the set of predictors; however, the study design does not allow for identification of a cause-and-effect relationship. Moreover, the treatment protocol in that study was not standardized. A randomized clinical trial is needed to compare the effectiveness of standardized treatment approaches in a homogenous sample of patients with CR.
The clinical use of intermittent cervical traction for CR is common, but its effectiveness has been examined in only one clinical trial.17 Joghataei et al17 found that exercise and intermittent cervical traction were superior to exercise and ultrasound in improving grip strength (force-generating capacity) following 5 visits in patients with C7 radiculopathy. However, the lack of a measure of pain or disability limits application of these results. There remains a paucity of quality outcome studies investigating commonly used interventions in a homogenous population of patients with CR. Thus, the purpose of this study was to examine the effects of manual therapy and exercise, with or without the addition of intermittent cervical traction, in patients with CR, as identified by the previously described CPR.
Materials and Method
A multicenter randomized clinical trial involving orthopedic physical therapy clinics in Virginia, Georgia, Alabama, and West Virginia (N=7 clinics) was conducted between October 2006 and December 2007. A total of 10 physical therapists (9 male, 1 female) with an average of 7 years (range=0.5–12) of experience treating patients with spinal conditions participated in data collection. In order maximize standardization, all clinicians were given on-site training by the primary investigator (I.A.Y.) and provided with an instruction manual and video on all examination, treatment, and data collection procedures.
Our original sample size estimate for data analysis was 80 subjects. Because the outcome measures used in this study have not been used in previous clinical trials for this patient population, an accurate power analysis based on effect size could not be calculated. With an estimated small effect size (f=0.25), a sample size of 80 would have given the study a power of 94%.
Consecutive patients with reports of unilateral upper-extremity pain, paresthesia, or numbness, with or without neck pain, were screened by a physical therapist for study eligibility. Of the patients screened for participation (N=121), 40 were excluded or refused to participate for variety of reasons. A flow diagram of patient recruitment and retention is presented in Figure 1. Patients who satisfied the eligibility criteria (Tab. 1) were invited to participate in the study. All enrolled patients (n=81) provided informed consent for participation in the study. Following consent, each patient underwent a standardized history and physical examination, as well as collection of data for all outcome measures.
The physical examination included the items in the CPR, repetitive motion testing (cervical protraction and retraction),25 deep tendon reflexes (biceps, brachioradialis, triceps), myotomal assessment (C5–C8, T1), and grip strength bilaterally. Primary outcome measures were the Numeric Pain Rating Scale (NPRS),26,27 the Neck Disability Index (NDI),28,29 and the Patient-Specific Functional Scale (PSFS).29,30 Secondary outcome measures were the Fear-Avoidance Beliefs Questionnaire (FABQ),31,32 a pain diagram,33 the Global Rating of Change Scale (GROC),34 patient satisfaction,35 and grip strength.36,37 Each outcome measure and its psychometric properties are described in the Appendix. Data for the outcome measures were collected at baseline and at 2-week and 4-week follow-ups.
After the examination, patients were randomly assigned to 1 of 2 treatment groups: a group that received manual therapy, exercise, and intermittent cervical traction (MTEXTraction group) and a group that received manual therapy, exercise, and sham intermittent cervical traction (MTEX group). In order to decrease the potential effect of the clinic on treatment outcomes, concealed randomization, stratified by clinic, was used to place patients into treatment groups. Numbered, sequential, sealed envelopes containing group allocation for each clinic were opened by the evaluating therapist after the baseline examination. Support staff, who were unaware of group assignment, administered all patient self-report measures and grip strength testing as instructed by the therapist.
Patients were treated for an average of 7 visits (SD=2.08), over an average of 4.2 weeks, with a standardized treatment protocol. Treatments were performed sequentially to include postural education, manual therapy, and exercise and ended with traction or sham traction. All patients received a home exercise program on their first visit, including one or more of the available exercises used in the standardized treatment protocol. The home exercise program was updated, as needed, on each visit by the physical therapist.
On the initial treatment visit, patients were educated on importance of correct postural alignment of the spine during sitting and standing activities. Posture was addressed on subsequent visits only if the physical therapist deemed it necessary.
Manual therapy was defined as either high-velocity, low-amplitude thrust manipulation or nonthrust manipulation. Initial treatment included manipulation procedures directed at the upper- and mid-thoracic spines of spinal segments identified as hypomobile during segmental mobility testing.38 Thrust manipulation of the thoracic spine could include techniques in a prone, supine, or sitting position based on therapist preference. Nonthrust manipulation included posterior-anterior (P-A) glides in the prone position. Therapists were required to perform at least one technique targeting the upper thoracic spine and one technique targeting the mid thoracic spine during each visit. Following treatment directed at the thoracic spine, at least one set (30 seconds or 15–20 repetitions) of a nonthrust manipulation was directed at each desired level of the cervical spine. The cervical spine techniques could include retractions, rotations, lateral glides in the ULTT1 position, and P-A glides. The therapists chose the techniques based on patient response and centralization or reduction of symptoms.
After completing the manual therapy procedures, the therapist instructed the patient on specific exercises to complement the manual procedures performed. Exercises included cervical retraction, cervical extension, deep cervical flexor strengthening, and scapular strengthening. At least one exercise was used during each treatment visit. All manual therapy and exercise procedures are described in the eAppendix.
Traction and sham traction.
After exercise, patients received either mechanical intermittent cervical traction or sham traction for 15 minutes according to their random assignment. Each patient was positioned supine, with the cervical spine placed at an angle of approximately 15 degrees of flexion. The traction force was started at 9.1 kg (20 lb) or 10% of the patient's body weight (whichever was less) and increased approximately 0.91 to 2.27 kg (2–5 lb) every visit, depending on centralization or reduction of symptoms. The maximum force used was 15.91 kg (35 lb). The on/off cycle was set at 50/10. The sham traction protocol included the identical setup; however, only 2.27 kg (5 lb) or less of force was applied. All other traction parameters were the same as for the group that received intermittent cervical traction.
A separate repeated-measures, mixed-model analysis was performed for each of the primary and secondary outcomes, with alpha set at .05. Treatment group (MTEX versus MTEXTraction) was the between-patient factor, and time (baseline, 2-week follow-up, 4-week follow-up) was defined as the repeated factor. The primary and secondary outcomes were used as the dependent variables. To allow for correlations within participants and of participants within clinics, we modeled patient and clinic as random effects without interactions. The main hypothesis of interest was the group × time interaction. Linear contrasts were constructed to determine the between-group differences at each time point. The main effects of the interventions were obtained by constructing linear contrasts to compare the mean change in outcome from baseline to each time point. The effect size was calculated from the between-group differences in change score from baseline to the 4-week follow-up in all of the primary outcome measures. Analyses followed intention-to-treat principles. All analyses were performed using SAS statistical software (JMP version 8.0*).
Role of the Funding Source
This study was funded by a grant from the Saunders Group.
Patients (N=121) were screened for eligibility, and 81 patients were eligible and agreed to participate (Fig. 1). Twelve patients (n=6 in each group) were lost to follow-up between baseline (pretreatment) measures and the 4-week follow-up. Baseline demographics and data for outcome measures are listed in Table 2.
No significant interaction or main effects of group were found for the primary or secondary outcome measures (Tab. 3). There was a significant main effect (P<.05) of time for the NPRS, PSFS, NDI, and body diagram, indicating there were significant improvements in pain, function, disability, and symptom distribution regardless of group assignment (MTEX versus MTEXTraction) from baseline to the 4-week follow-up. The adjusted effect size from the mixed-models analysis for each of the primary outcomes was small (NDI=1.5, 95% confidence interval [CI]=−6.8 to 3.8; PSFS=0.29, 95% CI=−1.8 to 1.2; and NPRS=0.52, 95% CI=−1.8 to 1.2).
This randomized clinical trial investigated the effects of a multimodal treatment approach including manual therapy and exercise, with and without the addition of intermittent cervical traction, in patients with CR. The results indicate that the addition of supine intermittent cervical traction yielded no additional benefit to a program of manual therapy and exercise. Regardless of group assignment (MTEX versus MTEXTraction), patients with CR experienced significant improvements in both primary and secondary outcomes following 4 weeks of standardized physical therapy intervention.
Although there were no significant differences between groups with any of the outcome measures, the precision of the point estimates of the treatment effects must be considered. At the 2-week follow-up, the lower boundary of the 95% CI for the NDI was −7.0 (Tab. 3). This value meets the threshold for meaningful clinically important change of the NDI (7.0). Furthermore, at the 4-week follow-up, the lower boundary of the 95% CI for the NPRS was −1.8 (Tab. 3). This value exceeds the threshold for meaningful clinically important change of the NPRS (1.3) adopted for this study. Thus, we cannot confidently exclude a treatment effect for these variables at these specific time points.
Although statistically significant changes over time were found in both groups with all of the primary outcome measures, the threshold for minimum clinically important change was surpassed with the NPRS (n=47 [67%]) and the PSFS (n=44 [64%]) for those patients who completed the 4-week follow-up. A total of 2 points of change on the PSFS has been found to exceed the threshold for minimal clinically important change in patients with CR.29 A change of 1.3 points on the NPRS recently was found to meet the threshold for minimal clinically important change in patients with neck pain.27 As no study has identified a minimal clinically important change value in patients with CR, this change score (1.3 points) on the NPRS was adopted for this study. Of the patients who completed the 4-week follow-up, only 32 (46%) surpassed the minimal clinically important change of at least 7 points on the NDI.29 A recent study27 suggests that the minimal clinically important change on the NDI may be more than twice as high as the original reported threshold of 7 points in patients with mechanical neck pain. With these inconsistencies regarding the appropriate threshold for clinically important difference, perhaps the responsiveness to change of the NDI may not be sufficient in this patient population. As the NDI is a commonly used self-report measure in patients with all neck-related disorders, future studies with larger sample sizes should investigate to detect change in patient status in conjunction with the NPRS, PSFS, and GROC in patients with CR.
The present study used a CPR to identify the presence of CR.12 The CPR has a sensitivity of 0.39 (95% CI=0.16 to 0.61), a specificity of 0.99 (95% CI=0.97 to 1.00), and a positive likelihood ratio of 30.3 (95% CI=1.7 to 538.2) when all 4 test items are positive. The CPR has a sensitivity of 0.24 (95% CI=0.05 to 0.43), a specificity of 0.94 (95% CI=0.88 to 1.00), and a positive likelihood ratio of 6.1 (95% CI=2.0 to 18.6)] when 3 of 4 tests are positive. We used 3 of 4 criteria that are positive for eligibility despite other studies using 4 of 4 criteria, due to the narrower CI and the lower-bound estimate for 3 of 4 criteria. To date, the CPR used in the present study has not been validated.
The protocol for the intermittent cervical traction may have been the reason a treatment effect was not identified. Although a multitude of traction parameters are used in the clinical setting, there is no convincing evidence to suggest which parameters are most effective in the management of CR. Cleland et al21 used an on/off cycle of 30/10 and a traction angle of approximately 25 degrees, increasing force by 0.45 to 0.91 kg (1–2 lb) per visit, whereas Waldrop et al18 used an on/off cycle of 20/10 and a 15- to 24-degree angle of traction. Each of these case studies started with a traction force of 8.18 kg (18 lb) and monitored the centralization and reduction of symptoms to determine progression of force. Furthermore, both studies performed traction for 15 minutes and used a minimum traction force during the off cycle.
In the clinical trial by Joghataei et al,17 a 13.64-kg (30-lb) traction force at a 24-degree angle of pull was used for a period of 20 minutes, with an on/off cycle of 7/5. In the present study, we used a longer duration of pull (on/off cycle of 50/10), a 15-degree flexion angle, and no traction force during the off cycle. In this study, the average traction force was 11.64 kg (SD=2.8, range=9.09–14.09) (25.6 lb, SD=2.8, range=20–31) for the MTEXTraction group and an average of 1.65 kg (SD=0.70, range=0.90–4.52) (3.5 lb, SD=1.1, range=2.0–5.0) for the MTEX group. Interestingly, Zybergold and Piper24 found no significant difference in pain reduction among groups of patients with CR who received static traction, intermittent traction, manual traction, and treatment without traction. Possibly, more-aggressive traction protocols (more force or greater frequency) may have had a greater effect on the patient sample in the present study. Moreover, we are unable to determine whether the sham traction force of no greater than 2.3 kg (5 lb) had a treatment effect on the patients in this study. Although a control group receiving a “subtherapeutic” traction force has its limitations, we feel this was the best control choice to address the setup, subsequent force production, and treatment time involved with this modality. In this study, there appeared to be no relationship between the amount of traction force used and perceived recovery (Fig. 2).
The manual therapy procedures used in this study were a combination of thrust and nonthrust manipulation techniques designed to centralize and reduce the cervical and upper-extremity symptoms. In order to simulate clinical practice, the therapist was allowed to select individual techniques based on centralization or reduction of symptoms and the patient's response to treatment. If a manual therapy procedure centralized or reduced the patient's symptoms, this procedure continued to be used until there was no further benefit. Conversely, if a manual procedure worsened or peripheralized the patient's symptoms, this procedure was abandoned and another technique was selected. The procedures are modifications of techniques first described by McKenzie,25 Maitland,38 Greenman,39 and Vicenzino et al.40 An average of 2 manual procedures were performed on both the thoracic and cervical spines during each visit. Supine thoracic thrust manipulation, cervical retraction nonthrust manipulation, and cervical retraction exercise were the most commonly used procedures in the study (Fig. 3). Although thoracic manipulation procedures have been shown to have a significant short-term treatment effect on patients with mechanical neck pain,41,42 these techniques have not been studied in patients with CR. Restoration of normal biomechanics to the thoracic spine may have a role in lowering mechanical stresses and improving distribution of joint forces in the cervical spine.41,43,44 Manipulations directed at the cervical spine were not performed in this study, as supporting evidence is sparse in patients with CR45 and considerable attention has been devoted to the risk of serious complications.46–48
The exercises used in this study included strengthening of the scapulothoracic and deep neck flexors, as well as cervical retraction and extension exercises. Scapular strengthening and deep neck flexor exercises have provided some benefit in previous studies.21,22 Cervical retraction is thought to improve resting neck posture, relieve neck pain or radicular or referred pain,25 and possibly decompress neural elements in patients with CR.49 An average of 2 exercises per visit were used in the present study.
This clinical trial supports previous randomized clinical trials demonstrating effective conservative management of CR17,23,24 and cervicobrachial pain14,15,24 Prior to the present study, only one randomized clinical trial isolated the effect of intermittent cervical traction, finding that exercise and intermittent cervical traction were superior to exercise (cervical isometrics) and ultrasound on the outcome of grip strength after 5 visits in patients with C7 radiculopathy.17 However, there were no significant differences between groups at 10 visits (discharge from physical therapy).17
We acknowledge several limitations of this study. First, we used a CPR to identify the presence of cervical radiculopathy that has yet to be validated, which may imply less-than-optimal diagnostic accuracy of this condition. Second, we are unsure of how effective the blinding was during the course of treatment, as the patients were not asked whether they could identify which group they were in at the 4-week follow-up. If the patients thought they were receiving the sham treatment, this may have had an influence on their outcome. Third, the lack of a strictly recorded, dose-specific home exercise program maintained during the course of treatment was a limitation. Fourth, without a control group (a group not receiving treatment), we are unsure whether there was a spontaneous resolution of symptoms over the course of this 4-week treatment.
The addition of mechanical intermittent traction does not appear to improve outcomes for patients with CR who are already receiving manual therapy and exercise. Although traction provided no additional benefit in this study, subsequent investigations examining traction at different dosages may be of interest in this patient population. The effect of CR can be disabling, and continued research in the areas of diagnosis and treatment of this condition is of paramount importance.
Mr Young, Dr Michener, Dr Cleland, and Dr Aguilera provided concept/idea/research design. Mr Young, Dr Michener, Dr Cleland, and Dr Snyder provided writing. Mr Young, Dr Michener, Dr Aguilera, and Dr Snyder provided data analysis. Mr Young and Dr Michener provided project management and fund procurement. Dr Michener, Dr Cleland, Dr Aguilera, and Dr Snyder provided consultation (including review of manuscript before submission).
The authors thank Advance Rehabilitation and Fredericksburg Orthopaedics for their support of this study; physical therapists Chris Brown, Dan Walker, Jon Lamb, and Richard Linkonis for their patient recruiting and treatment efforts; Amee Seitz for her help with data analysis; and Jennifer Chastain for her help with study/data management. A final thanks to Robin Saunders for her support of this study.
The study was approved by the Rocky Mountain University of Health Professions Internal Review Board.
Platform presentations of this research were given at the Combined Section Meetings of the American Physical Therapy Association; February 6–9, 2008; Nashville, Tennessee; and February 9–12, 2009; Las Vegas, Nevada.
This study was funded by a grant from the Saunders Group.
↵* SAS Institute Inc, PO Box 8000, Cary, NC 27513.
- Received September 13, 2008.
- Accepted March 25, 2009.
- American Physical Therapy Association