Abstract
A recent randomized controlled trial tested the effectiveness of therapeutic exercise and manipulative therapy on 200 subjects with cervicogenic headache. Although treatments were efficacious, 25% of patients did not achieve a clinically acceptable outcome – 50% reduction in headache frequency. This study aimed to identify predictors from variables in subjects' demographics and headache history which might identify those who did or did not achieve a 50-79% or 80-100% reduction in headache immediately after the active treatments and 12 months postintervention. The results revealed no consistent pattern of predictors, although the absence of light-headedness indicated higher odds of achieving either a 50-79% [odds ratio (OR) = 5.45) or 80-100% (OR = 5.7) reduction in headache frequency in the long term. Headaches of at least moderate intensity, the patient's age and chronicity of headache did not mitigate against a successful outcome from physiotherapy intervention.
Introduction
The treatments for cervicogenic headache are many and varied, ranging from pharmaceutical agents, physical therapies, cognitive behavioural programmes to surgical procedures, of which the most common is radiofrequency neurotomies (1–10). However, little is known about the features of a cervicogenic headache that characterize responsiveness to a particular treatment approach, either conservative or surgical (11, 12).
A randomized clinical trial was recently conducted to examine the efficacy of the physical therapy treatments of manipulative therapy and therapeutic exercise used alone or in combination for the management of chronic cervicogenic headache against a control of no active treatment (13). The results indicated that headache frequency was significantly and fairly equally reduced in all intervention groups and the relief was maintained in the long term (12 months’ follow-up). However, not unexpectedly, all subjects were not equally responsive to treatment. As part of the trial, baseline data were collected on 29 variables pertaining to the subject and their headaches. This presented the opportunity to analyse these variables to determine if there was a pattern that might predict responsiveness to these conservative physical therapies. It is particularly important to be able to recognize at an early stage, features which might indicate that the patient is unlikely to gain a clinically relevant benefit from treatment; that is, the treatment is unable to achieve at least a 50% reduction in headache frequency (14). It was also questioned whether predictors would be the same for those who achieved a moderate or excellent result. Thus this study aimed to determine whether there were predictors which would indicate the likelihood of responsiveness to physiotherapy treatment in both the short (post-treatment) and long term (12 months’ follow-up).
Materials and methods
Subjects
Two hundred subjects (18–60 years of age) participated in the clinical trial and were recruited by referral from general medical practitioners or through advertising. The inclusion criteria were those documented by Sjaastad et al. (15) for cervicogenic headache: unilateral or unilateral dominant, side consistent headache, associated with neck pain and aggravated by neck postures or movement. No diagnostic anaesthetic blocks were performed but upper cervical joint tenderness (C0–3) had to be present as detected by manual palpation (16, 17). The frequency of headaches had to be at least one headache per week with a history between 2 months and 10 years. Exclusion criteria were symptoms suggestive of another cause, e.g. bilateral headaches (which typifies tension-type headache), features suggestive of migraine (18), any condition which might contraindicate manipulative therapy (19), or current involvement in third-party or workers compensation.
Measurements
The baseline data collected from all subjects in the study included demographic data (age and gender), length of history of cervicogenic headache, the presence of another headache type and any family history of headache. More specifically in relation to the characteristics of the cervicogenic headache, data were compiled on headache frequency (average number of headache days per week), average intensity [10 cm visual analogue scale (VAS)] (20) and the presence of any associated symptoms such as nausea, vomiting, light-headedness, dizziness, blurred vision or paresthesia. Subjects also nominated the site of onset of pain (neck or head), known provocative factors, known relieving factors and if they considered that their headache condition was stable, worsening or improving. Baseline medication intake was monitored in daily headache diaries during a 2-week pretreatment period. Over-the-counter medications were predominant, which were mainly analgesics and in some cases anti-inflammatory medications, taken only in short and low doses for pain relief. For the purposes of analysis, medication was converted to a defined daily dose (DDD) of analgesics using the Anatomical Therapeutic Chemical code (21). Three questionnaires were completed, the Northwick Park Neck Pain Questionnaire (22) relating to neck pain and disability, the McGill Pain Questionnaire (MPQ) (23) relating to headache symptoms from which the pain rating index (PRI) was calculated, and the Headache Specific Locus of Control Scale (HSLC) (24) which assessed the patients’ perceptions of the influence of self-control (Internal), powerful others [health professionals; (External)] and chance (Chance) on their headaches. Physical assessments of the cervical region included pain produced on active cervical movements (10 cm VAS; the average score for the three most painful movements was used in analysis), pain provoked on manual examination of the upper cervical joints (10 cm VAS; the highest pain score used in analysis), the assessor's rating of perceived joint motion (seven-point scale, 4 was a normal motion rating and 3–1 and 5–7 represented slight, moderate and marked hypermobility and hypomobility, respectively) (16). Performance was measured in a five-stage muscle test of cranio-cervical flexion (CCFT) (25). The stage achieved and the number of repetitions performed at that stage to a maximum of 10 repetitions was documented. A photographic measurement was taken of cervical postural angle, the forward head postural position (26, 27).
Procedure
The study was a prospective, multicentre RCT with unblinded treatment and blinded outcome assessment (13). Subjects were randomly allocated to the manipulative therapy (n = 51), exercise therapy (n = 52), combined exercise and manipulative therapy (n = 49) or control (n = 48) groups. The active treatment period was 6 weeks and outcomes were evaluated in the week immediately postintervention (week 7) and at 3, 6 and 12 months after the intervention period.
Data management and analysis
For the purposes of this study investigating predictors of response to treatment, only the data of subjects from three active intervention groups were considered (n = 152). The data from these groups were analysed collectively as results from the RCT indicated that they responded in a similar way in the short and long term (13). Five of these subjects were lost to the trial at the 12-month follow-up and data from another subject were incomplete, resulting in a total of 146 cases used in the multivariate analyses.
Responsiveness was judged as a reduction in headache frequency. For the purpose of analysis, participants with complete data were categorized in three groups: did not achieve a clinically acceptable result of a 50% reduction in headache frequency (n = 35), achieved a 50–79% reduction (n = 43) or achieved an 80–100% reduction (n = 63) in headache frequency. Predictors of responsiveness were analysed using multivariate logistic regression analyses for the data for both the short (postintervention) and long term (12-month follow-up). A cross-tabulation was made of responders and non responders at the postintervention (7-week) and 12-month time points to investigate the stability or not within the groups.
Results
The mean frequency of headaches for the subjects in the treatment groups was 3.5 ± 1.8 headaches per week. The other baseline characteristics of the subjects in the three postintervention (7 weeks) response categories are presented in Table 1. Immediately following the 6-week intervention period, 45% of subjects gained an 80–100% reduction in headache frequency, 30% gained a 50–79% reduction and 25% failed to achieve a 50% reduction in their headache frequency. The frequencies were 42%, 34% and 24% for the three outcomes, respectively, at 12 months post-treatment.
Baseline characteristics of groups based on reduction in headache frequency at the postintervention time point (7 weeks)
HSLC, Headache Specific Locus of Control Scale; NPQ, neck pain and disability scores; PRI, pain rating index; DDD, defined daily dose; VAS, visual analogue scale; CCFT, cranio-cervical flexion test.
Results of the multivariate logistic regression analyses are shown in Table 2. The variance explained in the models ranged from 0.37 to 0.41. In Table 2, an odds ratio (OR) greater than ‘1’ indicates that higher scores on the predictor variable had an increased odds of achieving a reduction in headache frequency.
Results of multivariate logistic regression analyses of predictors of a 50–79% and 80–100% reduction in headache frequency immediately postintervention (7 weeks) and at 12 months’ follow-up∗
Group status included as a covariate.
χ2 significant at P < 0.05.
HSLC, Headache Specific Locus of Control Scale; NPQ, neck pain and disability scores; PRI, pain rating index; CCFT, cranio-cervical flexion test.
The results indicated that at the immediate post-treatment time point, the only feature indicating that subjects had lower odds of achieving a 50–79% reduction was a higher internalizing score (HSLC). With respect to an 80–100% reduction in headache frequency, subjects had lower odds of achieving this outcome if the onset of the headache was in the head region, headache intensity scores were lower, neck pain and disability scores (NPQ) and MPQ (PRI) scores were higher, externalizing scores (HSLC) were lower and length of headache history was longer.
After 1 year, subjects had lower odds of achieving a 50–79% reduction if they experienced light-headedness, had lower externalizing scores (HSLC) or lower joint pain scores on manual palpation. They had lower odds of achieving an 80–100% reduction if they were males, experienced light-headedness or had higher MPQ (PRI) scores. In summary, each of the outcomes has a separate profile in terms of the predictor variables.
The results of a cross-tabulation of response groups at the postintervention and 12 month time points are presented in Table 3. This table shows that approximately 50% of subjects retained their response level at both follow-up time points. Of the 36 who did not achieve a 50% reduction in headache frequency at the postintervention time point, half had achieved either a 50–79% (n = 9) or an 80–100% (n = 9) reduction by the 12-month follow-up. A further 20 subjects who had achieved a 50–79% response postintervention achieved 80–100% reduction in headache by 12 months. Conversely, four and 12 subjects lost the respective 50–79% and 80–100% postintervention responses by the 12-month follow-up time point.
Cross-tabulation of responders and non-responders immediately postintervention and at 12 months
Discussion
The cervicogenic headache group in this study suffered, on average, second daily headaches (mean 3.5 headache days per week) of at least moderate intensity (5/10 on a VAS). Females were predominant in the population (70%), headaches were chronic (an average 6-year history), and similar to other studies, approximately 40% of the cohort suffered from another headache type in addition to their cervicogenic headache (28).
Results of this study showed that the predictors of responsiveness for a reduction in headache frequency for both those who received a 50–79% and 80–100% reduction were disparate in terms of both possible clinical significance and time frame (immediately after treatment and 1 year's follow-up). The outcomes for each group were characterized by different numbers and types of predictor variables when compared with those who achieved less than a 50% reduction in headache frequency. Additionally, the magnitude of the significant odds ratios in most cases was small. Thus the aim to document a consistent headache pattern that could predict which patients might or might not respond to physiotherapy management was unable to be fulfilled across time and category of responder. This result is not unique to this study. The expert panel convened by Shekelle and Coulter (12) was able to offer only very general guidelines for which patients might benefit from cervical manipulative therapy, following their systematic review of manipulation in the treatment of the cervical spine and headache. McDonald et al. (11) found that 70% of 28 patients with chronic neck pain whose pain was eliminated by diagnostic blocks responded to radiofrequency neurotomy. However, there were no differences between responders and non-responders to radiofrequency neurotomy on available historical data of age, gender, length of history of neck pain, segmental level, impending litigation or whether diagnosis was by comparative or placebo-controlled blocks.
Some of the variables which reached statistical significance were curious with respect to time frames of response. For example, if onset of headache was in the head rather than the neck, the odds were significantly less for achieving an 80–100% reduction of headache in the short term, but not in the long term. Onset of pain in the neck is one of the features that characterizes cervicogenic headache (15) and this was a good sign for responsiveness in the short term. Pain starting in the head has been documented in cervicogenic headache but is more characteristic of other headache types, such as migraine (29). The occurrence of a second headache, from a pain processing perspective, might influence where persons first perceived their pain, but observation of the frequencies of reports of a second headache revealed that there was no higher incidence of reports of a second headache in those whose pain started in the head compared with the neck. A longer history of headache was associated with lesser odds of achieving an 80–100% reduction immediately post-treatment, but not with odds of achieving this result in the long term.
Of the significant predictor variables, two in particular were present on more than one occasion. These were higher pain rating intensity (PRI) scores from the MPQ completed for headache symptoms and the presence of light-headedness accompanying headache. Higher PRI scores indicated lower odds of achieving an 80–100% response to physiotherapy treatment in the short and long term, although not a 50–79% response. This is seemingly at variance with the finding that the odds were greater for achieving an 80–100% response in the short term with a higher headache intensity scored on a VAS. The use of the MPQ may reflect the benefits of evaluating the multidimensionality of the pain experience in headache, rather than the single dimension measured with a VAS (23).
The presence of light-headedness was associated with lower odds of achieving either outcome in the long term (although not the short-term postintervention responsiveness). This was the only variable with clinically compelling odds ratios (OR > 5.0). Light-headedness is a well-recognized symptom of upper cervical joint dysfunction and is attributed to disordered cervical proprioception (30–32). It is possible that the presence of light-headedness might reflect greater dysfunction in the cervical joints, reflecting in turn more disordered proprioceptive input. In support of this contention, Treleaven et al. (32) found greater cervical kinaesthetic deficits in whiplash subjects with symptoms of light-headedness and dizziness. There are two possible reasons why subjects with light-headedness in association with their headaches did not respond as well to treatment. The magnitude of pathology in the upper cervical joints may have been beyond that which can be assisted by conservative treatments, or the treatments of manipulative therapy and the specific exercise programme may not have addressed this impairment satisfactorily in the long term. There are exercise programmes tailored for these symptoms, consisting of head relocation and of eye, head coupling exercises that have been shown to be effective (33) but were not incorporated in the exercise regime in the clinical trial.
Approximately 50% of subjects displayed similar treatment outcomes at both follow-up time points. Not unexpectedly, there was movement between response levels over the trial (Table 3), which also confounds the task of trying to predict responsiveness on the features assessed in this study. Half of the subjects (n = 18) who did not achieve clinically relevant beneficial responses immediately after the intervention achieved them in the long term. Conversely, 14.5% (n = 16) who did achieve clinically relevant responses immediately postintervention failed to gain long-term relief.
There is also some interest in variables that did not predict responsiveness. Sjaastad et al. (3) reviewed the therapies that could be used in the management of cervicogenic headache. They considered that conservative methods used in physiotherapy were probably only suitable for patients with minor but not major symptoms and any effects were likely to be time limited. The results of this study challenge these beliefs. Long-term relief was obtained from the interventions. The mean frequency of headaches in this cohort was 3.5 headache days per week, which suggests a substantial headache problem. The average headache intensity was at least moderate (5/10) and headache intensity did not mitigate against a successful outcome, although males with a higher PRI on the MPQ had lower odds of achieving an excellent outcome, though not moderate relief of headaches. Length of history of headaches did not affect outcomes and the cohort had an average his-tory of 6 years of headache. Likewise age was not a predictor of responsiveness. Thus the absence of substantial predictors of responsiveness at this time would support the suggestion that conservative therapies should be the first treatment of choice for cervicogenic headache before surgical procedures are considered (2).
Finally, it could also be questioned whether subject demographics and headache features, as measured in this study, are the most sensitive variables to predict responsiveness. Another direction for future investigation might be to measure and better understand the pathophysiology underlying the individual cervicogenic headache patient's pain syndrome. There is growing interest in the role of central sensitization of nervous system in various headache types (34) as well as the role of abnormal central pain processing (35). Signs suggestive of both central sensitization and abnormal pain processing have been identified in patients with chronic neck pain (36–38) and these signs have been shown to be predictive of a poorer prognosis in whiplash-injured patients (39). Little is known of their prognostic potential in cervicogenic headache.
In conclusion, this study has determined that no strong, consistent pattern emerged from features of a subject's demographics and headache history that indicate their responsiveness to physiotherapy management in the immediate post-treatment period or in the long term. The presence of light-headedness with headache in a patient's headache syndrome warrants attention. However, it seems that headaches of at least moderate intensity, the patient's age and the chronicity of headache do not mitigate against a successful outcome from physiotherapy intervention. The results highlight the need for further prospective studies which should investigate a wider range of historical and pathophysiological features of cervicogenic headache to better identify responders and non-responders to conservative management.
Footnotes
Acknowledgements
Supported by grants from The National Health and Medical Research Council (NH & MRC grant no. 971139), The Physiotherapy Research Foundation, The University of Queensland Foundation, The St Vincent's Foundation and the Centre of National Research on Disability and Rehabilitation Medicine.