We compared two treatment regimens for the treatment of patients with the combination of headache and neck pain in primary care. The two treatments being investigated were the GP treatment alone and in combination with manual therapy. The required number of patients could not be obtained within the timeframe that was foreseen for the entire project. Therefore the study was terminated prematurely.
The headache complaints evolved in both groups in positive sense such as a significant reductions in HIT-6 score and average headache intensity of the last four weeks. No significant differences between both treatment groups of follow-up measurements were found.
Due to the low number of recruited patients all our results should be interpreted cautiously. Despite this low number of participants, however, we report our results as the design of the study was published [
12] and registered in a trial database [
19]. By publishing the results of less successful trials, positive publication bias can be avoided and the results can be included in systematic reviews and meta-analysis. Moreover we propose to analyse causes of the limited number of included patients and formulate advice for further research in this domain in primary care.
Study design
The study design was that of a pragmatic trial. In pragmatic trials the effectiveness of interventions is studied like they present in daily practice. The subjects can have a certain heterogeneity, as long as the spectrum of the population to which the studied treatments might be applied is represented [
20,
21]. The choice for a pragmatic design was based on two considerations. First, the observation in daily practice that patients with the combination of the clinical signs and symptoms headache and neck pain are frequently referred for physiotherapy or manual therapy treatment. An observation which is confirmed in literature [
5,
9,
22]. Second, the description of beneficial effects of MT treatment for migraine, TTH and CEH [
6,
8,
9,
23]. This trial differs from previous trials because it investigates the additional value of MT as an adjuvant therapy to the usual care.
We included three headache types that are most frequently seen in primary care, and that are frequently accompanied with neck pain. This setup thus reflects daily practice, as patients from these three headache types can be exposed to the interventions studied. Taking these aspects in consideration, we believe the inclusion of more than one headache type is justified for our purposes. Unfortunately, the total number of participants remained far too low to allow subgroup analysis, with subgroups based on the headache diagnosis.
Group characteristics
The total sample of patients studied here is representative for the entire headache population with regard to age and sex distribution [
24,
25]. The long history of headache complaints, the high frequency, pain intensity and high HIT-6 score indicate that the patients in our sample experienced a great burden of their headache.
Our participating patients were recruited in various ways (GP's offices, advertisements and outpatient neurology clinics). Recruitment via advertisements and outpatient neurology clinics was added to increase the number of participants [
12]. Based on the exclusion and inclusion criteria and by means of a diagnostic questionnaire, complemented with the HIT-6, we have made an inventory of their headache characteristics and we presented a detailed description of their headache features. In all subjects a headache diagnosis was made based on responses to the questionnaire and using the revised IHS criteria for migraine and TTH [
15] and the CHISG criteria for CEH [
16]. Diagnoses were assessed independently, and in case of disagreement a consensus diagnosis was reached after discussing the case.
Treatment effects
We were unable to demonstrate differences in treatment effects between both treatment groups at the follow-up measurements (week 7, 12 and 26). Both groups evolved positively, despite different accents in the treatment plan. E.g. the UC group received more medication on prescription. If confirmed in studies with larger patients samples this could indicate a potentially reduced risk for side effects of medication (e.g. gastro-intestinal bleedings) in the UCMT than in the UC group.
Both treatment regimens appear to be equivalent. Nonetheless, further research with larger patients samples is needed to clarify potential differences between both groups.
For future trials we recommended to use a headache diary. This can provide more complete and day by day information allowing a more profound recording of outcome measures.
Treatment adherence and content
Of the original 18 patients in the UCMT group only 10 actually followed the suggested treatment. Of the remaining eight, five reported practical implications of MT treatment: lack of time to attend the therapeutic sessions regularly or transport difficulties. One patient was disappointed to be allocated in this treatment group, based on previous negative experiences with similar therapies. Two patients who entered the trial via the advertisements, were discouraged by their own physician to participate in the trial, without further specification of their arguments. We believe this is hard to avoid in the setup of a research protocol. As suggested by Vernon et al., maybe some kind of remuneration for participation could be useful to anticipate [
26].
Based on the reports of the treating therapists, which described the treatment that was delivered, we have strong indications that the suggested treatment strategy in the UCMT group (a combination of spinal mobilisations and low load exercises) was actually followed. The average number of treatments was 8.6. This lies within the range of the number of treatments in studies using the same therapeutic approach [
6,
7]. No previous training period with the therapists was held due to practical reasons and to avoid withdrawal of therapists participation because of the time investment. For future research a training period with all participating therapists can be useful to standardise the delivered treatment even more.
Regarding the medication intake of patients, we see that the UC group received more medication for which a prescription is required, whereas in the beginning they mainly used over-the-counter medication. Subjects in this group receive more NSAIDS and triptans than the UCMT group and more than at baseline. This indicates that the stepwise pharmaceutical approach as described in the proposed guideline is followed by the GP's. The GP's who actually participated by referring patients were probably positively biased, knew guidelines and understood their potential in improving patient care. No side or adverse effects were reported by the participants, but more pro-active monitoring of potential adverse effects in either treatment arms is recommended for future trials.
Patient enrolment
Sufficient patient enrolment is a key issue in every clinical trial. Recruitment difficulties are also reported by Vernon et al [
26]. Apparently chronic headache patients are difficult to recruit in primary care settings and difficult to maintain within the boundaries of a treatment protocol. Our disappointing enrolment is probably the result of multiple factors.
Only half of the contacted physicians actually referred patients for the study, and most of them referred only one patient. Some of the documented reasons for insufficient participation of physicians are lack of time, difficulties with complex enrolment procedures, insufficient knowledge and awareness of the trial, concerns about the impact of trial participation on the doctor-patient relationship or about the loss of professional autonomy [
27‐
29]. One recommends intrinsic motivation, frequent reminders and practically feasible study protocols to achieve a more successful participation of physicians.
We contacted all physicians personally (first by phone, followed by at least one visit). The research protocol was explained, the research question and its clinical relevance were discussed. All physicians recognised the frequently occurring and challenging combination of headache and neck pain. The clinical value of the research question was stressed as well as the importance of physician cooperation for clinical studies in primary care. These aspects are recommended to enhance the physicians motivation and to gain their commitment [
29,
30]. We provided additional study information on websites which contained also online enrolment procedures, making the referral procedure as easy as possible. Regularly reminder e-mails were sent to keep their attention with the study. Despite our efforts the number of referred and enrolled patients remained discouragingly low.
For future research we recommend the development of a far more extensive network of physicians, taking into account that only a limited number will actually participate and refer patients. For instance, Van der Wouden et al. found that only 30% of all contacted GP's actually participated in studies with an interventional setup [
31].
The use of software applications which are incorporated in the patients electronic health record should be considered. Embi et al. developed and used a Clinical Trial Alert System which increased the physicians participation and patient enrolment significantly [
28,
32].
In our opinion, the fear of physicians to lose professional autonomy was a major issue. Clinicians have treatment preferences, and randomised study protocols require them to take a step back from daily routine, or even to prescribe a treatment they do not fully support. This can result in selective participation and referral of patients. Only those patients who do not fit in the regular schema or in who the regular schema was not successful are considered for referral to participate in the study.
Not only clinicians, but also patients have treatment preferences. Current legislation demands that patients are a priori informed in detail about the study purposes and procedures, by means of a written informed consent. Patients who prefer one of the treatment strategies in the study protocol are potentially biased or risk to be de-motivated when they are not allocated to the group they wanted to be allocated to. This might be one of the explanations for a number of not returned follow-up questionnaires and dropouts.
Treatment preferences of both physicians and patients can thus lead to insufficient enrolment and dropouts. The number of dropouts was equivalent in both treatment groups, so these factors have played equally in both groups.
As a remedy, the use of observational studies should be considered in future studies. E.g. Pfeiffer et al. used a prospective cohort study to avoid conflicts with physicians routine and treatment preferences of both physicians and patients [
33].
If an observational study protocol results in a larger patient sample, the analysis of subgroups can be performed. The treatment that has been followed can be inserted in the analyses as independent variable.
Bell-Syer et al. recommend the use of wide inclusion criteria in the early screening of potential subjects [
30]. To avoid biased patients we excluded patients who received MT treatment for their headaches during the last 12 months or patients whose prescribed medication was changed during the last two months. In this patient population these can be too strict exclusion criteria. In future research this exclusion criteria can be formulated less strict, which will probably result in a greater enrolment.