Abstract
Objective. In a cohort of primary care patients with chronic neck pain, to determine whether specific neck stabilization exercises, in addition to general neck advice and exercise, provide better clinical outcome at 6 weeks than general neck advice and exercise alone.
Methods. This was a multicenter randomized controlled trial in 4 physical therapy departments. Seventy-four participants (mean age 51.3 yrs) were randomized to specific neck stabilization exercises with a general neck advice and exercise program (n = 37) or a general neck advice and exercise program alone (n = 37). They attended a 1-hour clinical examination, followed by a maximum of 4 treatment sessions. Assessments were undertaken at baseline, 6 weeks, and 6 months. The primary outcome was the Neck Pain and Disability Scale (NPDS). Analysis was by intention to treat.
Results. Seventy-one (96%) participants received their allocated intervention. There was 91% followup at 6 weeks and 92% followup at 6 months. The mean (SD) 6-week improvement (reduction) in NPDS score was 10.6 (20.2) for the specific exercise program and 9.3 (15.7) for the general exercise program. There were no significant between-group differences in the NPDS at either 6 weeks or 6 months. For secondary outcomes, participants in the specific exercise group were less likely to be taking pain medication at 6-week followup (p = 0.02). There were no other significant between-group differences.
Conclusion. Adding specific neck stabilization exercises to a general neck advice and exercise program did not provide better clinical outcome overall in the physical therapy treatment of chronic neck pain.
Neck pain is a common musculoskeletal problem and most people suffer from it at some point in their lives1,2. Neck disorders affect 13% of adults at any one time and up to 30% of men and 50% of women in a lifetime2–5. Of these, 14% to 19% may go on to develop chronic pain6,7. Precise diagnosis by clinical examination is problematic7, because signs and symptoms are frequently nonspecific, with poor reproducibility. One approach in primary care is to use “red flags” to identify potentially serious disease8; once these have been excluded, patients are classified as having “simple” or “nonspecific” neck pain.
In the UK, direct referral to physical therapy is popular, because of reduced wait times, lower costs, and potentially quicker recovery compared with referral to medical specialists in secondary care9. Various treatments are used in the conservative management of neck pain including electrotherapy, exercise, and manual therapy10,11. Systematic reviews have identified a paucity of high-quality trials in this field, and also the lack of conclusive evidence for any particular treatment12. Even in the field of exercise therapy, there does not seem to be any evidence to support one particular approach over another13.
Weakness or fatigue of the neck musculature has been linked to neck pain14–16, indicating a role for strengthening exercise, and several studies have explored the role of general exercises in the management of neck pain. In Finland, Ylinen, et al17 have shown that both strength and endurance exercises are superior to a control intervention in chronic neck pain in women, and Viljanen, et al18 demonstrated an improvement in range of motion, although no difference in pain, following either dynamic muscle training or relaxation, compared to control. Taimela, et al19 found some evidence favoring a multimodal treatment program over both home exercises and a control intervention.
In recent years there has been a resurgence in the use of specific spinal strengthening programs20. The theoretical basis of such programs is that pain causes inhibition of deep stabilizing muscles, creating imbalance around the spine and leaving it vulnerable to further strain and hence pain20. While the focus of these programs has been on low back pain, studies have also investigated the proposed stability role of the deep anterior neck muscles21–23. However, there has been limited investigation into the role of specific neck stabilization exercises in the management of neck pain11 in UK clinical practice. These muscles are difficult to assess because of their position. However, Jull24 proposed that the movement of head-neck flexion can be used to test the strength of the deep cervical flexor group. One randomized controlled trial has investigated the effects of manual therapy and specific cervical stability strengthening exercises on cervical headache, but found no difference between the active intervention groups25.
There is thus little evidence for the role of specific spinal stabilization exercises in relation to neck pain. We studied whether specific neck stabilization exercises, in addition to a general neck advice and exercise program, were better than a general neck advice and exercise program alone, in the treatment of chronic neck pain. We adopted a pragmatic approach, which addressed the comparative effectiveness of broad treatment packages — rather than individual treatment techniques or modalities — in everyday practice, where tight experimental control and blinding of the therapist and patient were not feasible26.
The primary objective was to compare at 6 weeks the effect of adding specific neck stabilization exercises to a program of posture correction techniques and neck range of movement exercises in patients with chronic neck pain. The secondary objective was a comparison of the effects of the treatment programs at 6 months.
MATERIALS AND METHODS
Study participants
This was a single-blind randomized controlled trial. Participants were recruited from 4 outpatient sites across one region of the UK (one acute hospital and 3 community hospital sites). The local research ethics committee approved the study. Eligible participants were aged 18 years and over with neck pain symptoms for longer than 3 months, and had been referred by their general practitioner to physical therapy with a new episode of neck pain. Exclusion criteria were “red flags,” for example, inflammatory conditions (e.g., rheumatoid arthritis), worsening neurological signs, severe bone disease (e.g., osteoporosis), or nonmechanical pain.
Participants recruited to the trial were identified by the trial coordinator from the non-urgent waiting lists at each center. Participants were sent a trial information letter and screened by telephone a few days later for eligibility and willingness to participate. Participants not wishing to enter the trial but still requiring physical therapy were offered a routine treatment appointment. Written informed consent was obtained before baseline data collection and participants were assigned a unique study number.
Randomization
A computer-generated randomization list with random permuted blocks (block size: 2, 4, or 6), stratifying for center, was devised by the Birmingham Clinical Trials Unit (BCTU). Participants were then allocated using telephone randomization at BCTU. Once contacted by the trial coordinator, BCTU allocated each participant to Group A or Group B.
Blinding
Only the treating clinicians knew which treatment group was A or B at each site, thereby ensuring that the trial coordinator (CG), who was responsible for data entry, and the statistician (JS) remained blind to patient allocation. A copy of the randomization sheet was attached to the therapy record sheet. Patients were aware that they had an equal chance of being allocated to one of the 2 interventions and that these were both exercise programs, but were unaware of the specific difference in content between the interventions, and thus were blind to the study hypothesis.
Interventions
Eleven musculoskeletal outpatient physical therapists were involved in treating the trial participants. They had a mean of 8 years’ musculoskeletal experience and were experienced physical therapists (i.e., all employed at Senior II or Senior I level). In the month prior to the beginning of the trial, they all attended a study session that presented the rationale for the study and outlined the trial protocol. To monitor the delivery of the interventions and check for any contamination, we performed an audit of treatment notes after 20 participants had been recruited. This was carried out by 2 assessors independent of the clinical centers (KD, JW) and showed good standardization of the study interventions. The results of this audit were conveyed to the treating therapists.
Posture correction techniques and active range of movement exercises were delivered in both arms of the trial. In the general exercise group, the active range of movement exercises prescribed were at the therapist’s discretion, but could include flexion, extension, side flexion, and rotation. Posture correction was taught in the context of functional and work activities. The general exercise group received an intervention based on this program alone. The specific exercise group were additionally taught a program of specific neck stabilization exercises. These included an isometric craniocervical flexion exercise, performed sitting, standing or lying, and an isometric craniocervical flexion exercise in an inclined sitting position with a head-lift off the supporting surface. The treating therapists were asked to aim for an isometric hold of up to 10 seconds, and up to 10 repetitions. Participants were also taught to use their deep flexor muscles during functional activities. The hold time, repetitions, and progression of the exercises were at the discretion of the treating therapist. Participants received a 1-hour clinical examination, and the study interventions were subsequently delivered over a maximum of four 30-minute treatment sessions within the first 6 weeks; the number of sessions up to this maximum was at the treating therapist’s discretion.
We asked participants in both groups to perform their exercises 5–10 times daily, and the interventions were reinforced with written exercise sheets; these were generated through the Physiotools computer program, which produces customized diagramatic illustrations of individual exercises. All treatments were recorded on case report forms. We validated these forms against therapy treatment records in a convenience sample of 20 participants.
At 6 weeks, the therapist could discharge the patient from treatment or continue with other treatment modalities if appropriate. The criterion for discharge was either if the participant reported being symptom-free, or if in the opinion of the treating therapist further treatment would not affect short- or longterm outcome. The trial protocol was designed to replicate clinical practice at all trial sites and therefore a minimal amount of therapy equipment was used. The therapists were given written patient exercise sheets to reinforce the treatment program.
Outcome measures
Outcome measurement was undertaken by the trial coordinator at baseline, and by postal self-completed questionnaires at 6 weeks and 6 months. The primary outcome measure was the Neck Pain and Disability Scale (NPDS)27,28. The NPDS is a 20-item scale covering pain and functional, social, and emotional aspects of neck pain. The scale uses a 0–5 visual analog scale (VAS) with the addition of incremental line points. We replaced the VAS with a 0–10 numerical rating scale to enhance data collection in the self-completed questionnaires. Prior to analysis, scores were rescaled to 0–100. Secondary outcome measures included the Northwick Park Neck Pain Questionnaire (NPQ)29, pain affect (11-point numerical scale of how distressing the patient feels the pain is), severity (11-point numerical scale) of patient-identified worst problem (elicited by the question: “Because of your neck, which one thing gives you the most problems?”), self-reported use of pain relief medications over the preceding 48 hours, patient’s global improvement (5-point numerical scale of how much the patient feels symptoms have improved from baseline), uptake of further treatment, and the Medical Outcomes Study Short Form-36 health status measure30. A 5-point Likert item (“I was able to do my exercise as often as I was told to”) was used to measure adherence to the exercise program (response categories: strongly disagree, disagree, not sure, agree, strongly agree). Similar items were used to measure the clarity of the information sheets (“The information sheets were clear and easy to follow”) and the perceived ease of performance of the exercise regimens (“I found the exercise easy to do”). The presence of chronic widespread pain, according to the American College of Rheumatology criteria31, was determined by a self-completed blank body manikin. The questionnaire package was successfully piloted on a sample of 20 patients prior to the trial.
Sample size
The sample size calculation was based on the NPDS. The banding system described in the validation study for the NPDS27 suggests that a change in scores between baseline and 6 weeks of 12 points is a clinically important change. To detect a 12-point difference in 0–6 week change on the NPDS between 2 groups — assuming a standard deviation of 16.527, a 2-tailed significance level of p ≤ 0.05, 80% power, and equal group sizes — a total sample of 62 patients was required. We aimed to recruit at least 70 participants to allow for loss to followup.
Statistical analysis
Analysis was undertaken on an intention-to-treat basis. Where values were missing, these were estimated by multiple imputation (with the assumption that values were missing at random).
Estimates of between-group differences on numerical outcomes were derived through analysis of covariance, including treatment center as blocking factor and controlling for sex, chronicity (natural log-transformed), and baseline values on the outcome measure. These control variables were selected a priori in terms of clinical or prognostic importance32. For categorical outcome data, multiple logistic regression was used, controlling for the same variables. For ordinal data, unadjusted between-group differences were analyzed by the Wilcoxon rank-sum test. For all analyses, p values and confidence intervals were adjusted for the number of imputed values33. A sensitivity analysis was performed on the primary outcome measure, the NPDS, using complete data only.
Statistical significance was set at p ≤ 0.05 (2-tailed). Analyses were carried out using SPSS version 14 and SOLAS 3.2. No interim analyses were undertaken.
RESULTS
Recruitment and followup
One hundred seventy-one consecutive patients with chronic neck pain were approached for possible recruitment, of whom 74 (mean age 51.3 yrs) were randomized (Figure 1). The most frequent reasons for patients not entering the trial (n = 97) were unable to contact (n = 30; 31%) and not wishing to be recruited (n = 27; 28%). There were no differences between those recruited and those not recruited in age (p = 0.84; t test) or sex (p = 0.17; chi-square test).
Trial profile — baseline to 6 months.
Baseline characteristics are shown in Table 1. Somewhat higher numbers of participants were not working and had received previous physical therapy in the general exercise group, whereas median chronicity was slightly higher in the specific exercise group. Otherwise, characteristics were similar across groups. The physical therapy clinical diagnoses were also similar, although a diagnosis of spondylosis was more common in the general exercise group. Overall, spondylosis (n = 23) was the most common diagnosis, followed by whiplash (n = 9), nonspecific neck pain (n = 8), and discogenic pain (n = 7). Seven participants were lost to followup at 6 weeks, and 6 participants at 6 months, giving 91% and 92% followup, respectively (Figure 1). For 2 participants, data were unavailable at both followup dates.
Baseline characteristics of participants. Lower scores are better on Neck Pain and Disability Scale, Northwick Park Neck Pain Questionnaire, pain affect, and rating of worst problem. Higher scores are better on anticipated improvement and SF-36 physical and mental component summaries.
One patient in the general exercise group and 2 patients in the specific exercise group were randomized and completed baseline measures, but subsequently either failed to receive any treatment or received fewer treatments than the therapist intended.
Interventions
Nearly all patients received their allocated intervention, 36 (97%) in the general exercise group, 35 (95%) in the specific exercise group. Participants in the general exercise group received a median (interquartile range) of 3 (3, 3) treatments within the trial, compared to 4 (3, 4) in the specific exercise group; in both groups, the range of treatments received was from 0 to 4.
Compliance with home exercises, as assessed by the treating therapist during the period of delivery of the intervention, was judged to be good to excellent for 84% of participants (on a 3-category scale of poor, good, excellent). The response to the Likert item measuring adherence (“I was able to do my exercise as often as I was told to”) differed somewhat between groups. In the general exercise group, more patients (27/33; 82%) agreed or strongly agreed with this statement than in the specific exercise group (20/32; 63%). Additionally, more patients (28/33; 85%) agreed or strongly agreed with the statement “I found the exercise easy to do” in the general exercise group than in the specific exercise group (19/32; 59%), while a similar number of patients agreed or strongly agreed with the statement “The information sheets were clear and easy to follow” in the specific exercise group (31/32; 97%) as in the general exercise group (28/33; 85%).
Outcome measures
Raw scores for all outcome measures are given in Table 2. For the raw scores on the primary outcome, the NPDS, the mean (SD) 6-week improvement (i.e., reduction) from baseline was 9.3 (15.7) points for the general exercise group and 10.6 (20.2) points for the specific exercise group. The mean (SD) 6-month improvement from baseline was 9.0 (20.2) points for the general exercise group and 14.7 (22.1) points for the specific exercise group. In both groups, therefore, the raw mean within-group change was not clinically important (< 12 points) at 6 weeks. At 6 months, however, mean within-group change was clinically important (≥ 12 points) for the specific exercise group. At 6 weeks, more participants in the general exercise group had achieved a clinically important change (17/37; 46%) than in the specific exercise group (14/37; 38%), although this difference was not significant (p = 0.48; chi-square test). At 6 months, the number achieving a clinically significant change was higher in the specific exercise group (22/37; 60%) than in the general exercise group (15/37; 41%), although again not significantly (p = 0.10; chi-square test).
Raw values of outcome measures at followup. Lower scores are better on Neck Pain and Disability Scale, Northwick Park Neck Pain Questionnaire, pain affect, and worst problem. Higher scores are better on SF-36 physical and mental component summaries and subjective improvement.
Figure 2 shows the scores on the NPDS adjusted for baseline values, sex, treatment center, and chronicity, and with missing values imputed. Table 3 shows the results of the primary intention-to-treat analysis, with missing values imputed. Between-group differences on the NPDS were statistically nonsignificant at both 6 weeks and 6 months, and were not clinically important (< 12 points). Table 3 also shows the sensitivity analysis on the complete data: estimated differences remained statistically nonsignificant and clinically unimportant.
Profile plot for mean scores on the Neck Pain and Disability Scale (NPDS) for the general exercise group and the specific exercise group, adjusted for baseline value, sex, treatment center, and chronicity.
Results on primary outcome measure: the Neck Pain and Disability Scale. All estimates adjusted for treatment center, sex, chronicity, and baseline value. Lower scores are better on the Neck Pain and Disability Scale.
Results of the analyses of the secondary outcomes are presented in Table 4. No between-group differences were found, other than for use of pain relief medications; at 6 weeks, participants in the specific exercise group were about 30% as likely as those in the general exercise group to be taking pain relief medications (p = 0.02). The magnitude of the treatment effects on the secondary outcomes was otherwise generally small.
Intention to treat analyses for secondary outcomes. Data are means and mean difference, except for Subjective improvement (medians and median difference) and Taking pain medications (frequencies and odds ratio). All estimates, except that for Subjective improvement, adjusted for treatment center, sex, chronicity and baseline value (odds ratio for Taking pain medications is adjusted, but raw cell frequencies are shown). Lower scores are better on Neck Pain and Disability Scale, Northwick Park Neck Pain Questionnaire, pain affect, worst problem, and anticipated improvement. Higher scores are better on SF-36 physical and mental component summaries and subjective improvement.
Cointerventions
More participants received additional treatment in the general than in the specific exercise group: 7 versus 2 at 6-week followup, and 10 versus 5 at 6-month followup. These differences were not statistically significant (p = 0.11 and p = 0.13, respectively; chi-square test). The additional treatment modalities used were similar across the 2 groups.
DISCUSSION
We report the findings from a randomized controlled trial in 5 UK physical therapy centers, investigating the effectiveness of adding a specific cervical strengthening exercise program to a general cervical exercise program in the treatment of chronic neck pain. Although the specific exercise group showed clinically significant improvement at 6-month followup, in terms of between-group comparisons we were unable to demonstrate additional benefit from adding specific neck exercises to a package of general neck exercises, except in relation to the use of pain relief medications at 6 weeks. Moreover, observed between-group treatment effects —with the exception of the use of pain relief medications at 6 weeks — were small, and unlikely to be clinically important. The sensitivity analyses on the primary outcome, based on participants with complete data, did not produce different conclusions.
The strengths of this study include achievement of the recruitment target, remote third-party randomization, and high followup (91% at 6 weeks, 92% at 6 months). The allocated treatment was received by 96% of the participants and was delivered in accord with the protocol. Outcomes were analyzed blind to treatment allocation and by intention to treat. We chose a primary outcome measure of neck-specific disability that is recognized as appropriate in this population34.
Use of broad eligibility criteria enabled participants with a wide range of nonspecific neck pain problems to be recruited. The size of the study does not, however, allow us to identify which subgroups of patients respond best to specific approaches.
The trial interventions were designed to reflect usual clinical practice, and were delivered over a maximum of 4 sessions within a 6-week period. The neck exercises were selected from a menu of exercises previously used in a protocol for a neck pain randomized controlled trial26. The selection of exercises was based on the subjective history and objective findings. However, there was no evaluation of the physiotherapists’ clinical decision-making in arriving at this choice. The number, repetition, and progression of exercises were recorded and were at the therapist’s discretion. Accommodating the natural variation in physical therapy treatment is one of the hallmarks of pragmatic studies, where allowing clinicians the flexibility to reflect clinical practice needs to be balanced against ensuring validity of the trial protocol34,35. As in other studies17–19, lack of effectiveness may be attributable to dose. However, in our study fewer treatments were given on average than were permitted by the protocol, suggesting that the treating therapists did not feel constrained in terms of the treatment they provided. While the therapists judged compliance with the exercises overall to be good, how frequently patients performed them was not directly measured. It is possible that one approach to exercising was easier to undertake than another. The comparative benefit of a new exercise approach may not be fully realized if participants follow the prescribed program less easily than an established one. While we cannot therefore link adherence with exercise to outcome, it may be that the lack of superiority of the specific exercise program is related to a lack of adherence. Such an interpretation is supported by the observation that fewer patients reported being able to do their exercises as often as they were told to in the specific exercise group than in the general exercise group. This suggests that patients in the specific exercise group may have performed too few exercises, and coupled with the greater reported ease of performance of the general exercise regimen, highlights the need to ensure that exercises are taught in such a way as to maximize adherence.
Although the difference in the numbers of patients in each group achieving a clinically important change at each timepoint was not significant, this finding may tentatively suggest that general and specific exercise regimens achieve their effects over differing time periods. This hypothesis could usefully be explored in future studies.
The intended sample size was achieved. However, variability of NPDS scores was in the event greater than the estimate used in the sample size calculation, which might suggest, in retrospect, that our study was underpowered. However, as the magnitude of the observed between-group effects on the NPDS was at best just over half that deemed clinically important, and secondary outcomes also predominantly showed very small effects, the clinical implications would be very unlikely to differ had the trial been larger.
Comparison of our study with others is difficult because few have examined the effectiveness of specific stabilization exercises, or have done so in relation to pain elsewhere in the spine20,36. Although it seems that alteration in specific muscle groups may occur with stability muscle training, studies have been limited by their small size. In a study of cervicogenic headache, Jull, et al25 evaluated manipulative therapy and a low-load exercise program used alone and in combination, compared with a control group. Participants (n = 200) were randomized to manipulative therapy, specific exercise therapy, combined therapy, and a no-treatment control group. At 12-month followup, the manipulative therapy and specific exercise groups both showed statistically significant improvement in headache frequency and intensity and in neck pain and disability, compared to the control group; there was no additive effect when the 2 treatments were combined. Effect sizes for these outcomes were deemed clinically relevant. It appears, therefore, that specific exercise can produce longterm reduction in the symptoms of cervicogenic headache and neck pain.
Our study was not designed to provide evidence for the effectiveness of general exercise, our comparator group, for nonspecific neck pain. However, we have previously shown that advice and exercise seems an attractive approach to the management of nonspecific neck disorders35.
We demonstrated no overall additional benefit from adding specific neck stabilization exercises to a package of general neck exercises. There would seem, therefore, to be no clear benefit of including specific neck exercises in the treatment of chronic neck pain generally. However, future studies may reveal subgroups of these patients for whom this approach is beneficial.
Acknowledgments
The Arthritis Research Campaign (KD), the North East Wales NHS Trust (CG), and our collaborators: the participating physiotherapists and therapy managers, and the Clinical Trials Unit, Birmingham University. We are grateful to Daniëlle van der Windt for advice on the manuscript.
Footnotes
-
Supported by the Physiotherapy Research Foundation and the Welsh Office.
- Accepted for publication September 23, 2008.
