Neck-specific strengthening exercise compared with placebo sham ultrasound in patients with migraine: a randomized controlled trial

This two-armed, parallel-group randomized controlled trial was designed according to the CONSORT extension for nonpharmacologic guidelines [23] to verify the effectiveness of craniocervical exercises in migraine patients. The protocol study was published [24]. The trial was approved by the Research Ethics Committee of the Hospital das Clínicas, Faculty of Medicine of Ribeirão Preto, University of São Paulo (no. 6862/2016).and registered under the accession code RBR-8gfv5j on 28/11/2016 in the www.​ensaiosclinicos.​gov.​br/​rg/​RBR-8gfv5j/​. All participants provided written informed consent.

Men and women aged between 18 and 55 years, diagnosed only with migraine and at least three days of pain per month were included. The diagnosis of migraine was confirmed by a neurologist with over five years of experience, per the 3^rd edition of the International Headache Classification [25]. The volunteers were recruited from the local population after the study was released on social media (Facebook®, Instagram®, and the local university radio). The exclusion criteria comprised of: the presence of other concomitant headaches, including headache due to medication abuse, tension-type headache or trigeminal autonomic cephalalgias; a history of trauma to the neck or face; a history of a herniated disc or joint degeneration in the cervical region; and systemic diseases such as fibromyalgia and uncontrolled arterial hypertension; pregnancy; local anesthetic nerve block in the last three months and having performed physical therapy in the craniocervical region in the last year. The data collection and the interventions were carried out in the laboratory of the research group.

Included participants underwent baseline assessment, which evaluated demographic characteristics, primary and secondary outcomes. Then, the participants were randomly assigned to the two groups and the interventions started in both groups in up to seven days. After six weeks of treatment, an intermediate assessment of CCFT and the Global Perception of Change questionnaire was done. After eight weeks of treatment, the final assessment was carried out and all the questionnaires and physical tests were applied. Follow-up via telephone was performed one, two, and three months after the end of the treatment only for the primary outcomes [24]. All participants were advised to do not start any type of physical therapy during the study duration, including the follow-up period and, due to ethical issues, they were advised to maintain their pharmacological treatment as they are used to before the inclusion in our study.

The single therapist responsible for both interventions, CMSE and sham ultrasound plus home-stretching, was trained to maintain standardization and adherence to the proposed interventions protocols. The therapist training so that the proposed intervention protocols were strictly followed is a measure to avoid the occurrence of bias [26].

The intervention group performed an eight-week CMSE protocol consisting of specific exercises for cervical muscles [24, 27]. The first stage, which lasted for six weeks, focuses on the deep cervical muscles [27]. It was proposed to start with two sets of 10 repetitions with 10 s of endurance for deep flexor muscles of the cervical and two sets of 10 repetitions for the deep extensor muscles. Progression in the number of series, repetitions, and endurance during the six weeks occurred individually, according to the complaint or absence of pain and execution of the movement without compensation, such as retraction or elevation of the head and excessive use of superficial flexors. The second stage lasted for two weeks, and in addition to the deep cervical muscles, it involved the superficial muscles of the cervical region [27]. It consisted of three sets of 15 repetitions for both the flexor and cervical extensor muscles. All volunteers received instructions from a physiotherapist, who had five years of experience, in individual sessions once a week for approximately 20 min. In addition, the volunteers were instructed to perform the craniocervical exercises at home twice daily and cervical muscle stretches. During the weekly sessions with the therapist, home guidelines were reinforced but the volunteers’ adherence was not monitored.

The volunteers in the sham ultrasound group had individual sessions once a week for approximately 20 min, in which disconnected therapeutic ultrasound was performed bilaterally in the middle portion of the upper trapezius for eight weeks. Volunteers also received a guideline book containing only cervical muscle stretching exercises to be performed once daily (Additional file 1) Stretching exercises for the cervical flexors, extensors, tilters and rotators were oriented. For this group, home guidelines on cervical muscle stretches were not reinforced weekly during the sessions and the performance was not monitored”.

If any volunteers from either group interrupted the treatment and left the study, contact was made, and continuity was requested only in the assessments.

The primary outcomes were frequency (days with headaches per month) and intensity of headache by the numerical rating scale (NRS; 0–10) [28], collected using a paper headache diary. The gold standard for measuring data on the frequency and duration of headaches is self-report [29], and the use of headache diaries in randomized clinical trials is an important option to more accurately capture the effects of the interventions applied. in these outcomes [30]. The secondary outcomes were migraine and neck disability, collected using the Migraine Disability Assessment (MIDAS) [31, 32] and the Neck Disability Index (NDI) [33, 34], respectively; presence and severity of cutaneous allodynia verified by the 12-item Allodynia Symptom Checklist (ASC-12) [35]; presence of kinesiophobia detected by the questionnaire Tampa Scale for Kinesiophobia (TAMPA) [36]; functional changes over time related to migraine verified by the Patient-Specific Functional Scale (PSFS) [37]; and perceptions of improvement and satisfaction associated with an intervention, verified by the Global Perception of Change questionnaire [38]. It further included cervical range of motion, pressure pain thresholds, and performance of cervical muscles in CCFT, maximal isometric voluntary contraction (MIVC), and endurance with electromyographic records.

Cervical range of motion for flexion, extension, lateral flexion, and rotation were performed using the Cervical Range of Motion (CROM) (Performance Attainment Associates, Roseville, MN, USA) [39]. Pressure pain thresholds were assessed using a digital manual dynamometer (DDK-10 Kratos®) on the upper trapezius [7], sternocleidomastoid (SCM) [40], suboccipital, temporal [41] frontal [42] and levator scapulae muscles.

During the CCFT, MIVC, and endurance tests, muscle activity was recorded using surface electromyography. Wireless surface sensors (Trigno™ Wireless Systems, Delsys, Inc. Boston, MA, USA) were fixed with double-sided adhesive tape bilaterally over the SCM [43], anterior scalene [43], splenius capitis [44], and upper trapezius [45]. Data were processed by a customized MATLAB® routine employed to filter them at a frequency range of 20–450 Hz (4th order Butterworth), using a 0.375 s movable window with 50% overlap. The root mean square (RMS) from the CCFT records were normalized by their respective peak values during MIVC. For MIVC, RMS the mean RMS of each muscle was normalized to its peak value within the task. For the cervical muscle endurance test, the variables analyzed were the RMS slope and median frequency. To obtain them, initially, three windows were separated from the total acquisition window, 10% initial, 10% intermediate, and 10% final, thus creating a straight line of three points.

In the CCFT, volunteers remained in the supine position; a Stabilizer Pressure Biofeedback® device (Chatanooga, Hixson, TN, USA) was positioned at the back of the neck and, for each of the five stages, they were instructed to maintain the established pressure for 10 s without compensating for retraction or elevation of the head and excessive use of superficial flexors [46].

Muscle strength was measured. by a hand-held dynamometer (Lafayette Instrument Company®, model 2,201,163, Lafayette, IN, USA). To measure the strength of the cervical flexor muscles, the volunteers remained in the supine position, while for the extensor muscles, the volunteers remained in the prone position [47].

For the cervical flexor endurance test, the volunteers remained in the supine position and performed head and neck flexion without leaning their heads against the stretcher. For the cervical extensors, the volunteers remained in the prone position; the task was to keep the cervical muscles in a neutral position controlled by CROM and support a weight of 2 kg. The endurance test was interrupted in case of pain or fatigue, or inability to maintain the position [4]. Endurance time was measured in seconds and one repetition was performed.

The sample calculation was based on a between-group difference of 3.3 (standard deviation, SD = 3.1) days of headache per month [48] and 1.5 (SD = 1.4) in the NRS [49] considering a level of significance (α) of 0.025, a power of 80%, and a loss of 20% at follow-up. A sample size of 21 individuals from each group was recommended.

All volunteers were allocated in sequential order, determined by randomization using Excel®. The brown and opaque envelopes containing the group assignments were opened in front of each volunteer. A single therapist, blinded to the results obtained during evaluations, was responsible for all interventions in both groups. Similarly, the therapist responsible for data collection was blinded to the treatment allocation group. Both therapists were not blinded to the diagnoses because all the volunteers had migraine.

Mixed linear regression models were performed for the primary and secondary results of the quantitative variables to estimate comparisons between groups and within the group. “Time” and “group” were considered as fixed effects and the participants as a random effect (random interception model). The presence and intensity of neck pain was used as a covariate, based on the difference observed between the groups at baseline. For comparison, a post-test using orthogonal contrasts was used. A multinomial regression model with mixed effects was used for comparisons involving the global change perception questionnaire. The analyses were performed with the aid of SAS 9.4 and R 3.5.3. A power of 80% was considered, and a significance level (α) of 0.05 was adopted for all comparisons.

The minimum important difference (MID) and effect size (ES) were calculated based on the distribution methods to attribute clinical relevance to the differences found in both groups. The difference found was considered clinically relevant when it was greater than the MID and was combined with an ES > 0.4 [50].

Frequency of headache: There were no statistically significant differences in the group-by-time interaction for the frequency of headache (F = 0.44, df = 4; p = 0.782). There were also no within-group differences in headache frequency during the study period (p > 0.05) (Table 2) as well as clinically relevant differences. All data used to classify the clinical relevance of the results related to the primary outcomes can be found at Additional File 1. (Table 2 and Additional file 2).

Intensity of headache: There were no statistically significant differences in the group-by-time interaction for pain intensity (F = 1.51; df = 4; p = 0.202), although both groups showed significantly reduced intensity of headache in all comparison periods (p < 0.05) (Table 2). Clinically relevant within-group results were observed at some time points in the study for the intensity of headache. These results are described in Table 2 and Additional file 2.

Cervical range of motion: There were no significant differences in group-by-time interaction for any of the movements assessed during the study period (p > 0.05) (Table 3). Clinically relevant results were observed for some movements in both groups; these results are described in Table 3 and Additional file 3, in which all data used to classify the clinical relevance of the results related to the secondary outcomes can be found.

Pressure pain threshold: There was a significant improvement in the sensitivity of the frontal muscle (F = 4.30; df = 112; p = 0.040) in favor of the intervention group (Table 3). Within-group clinical relevance was observed for all muscles in the intervention group, except the SCM. No clinically relevant results were observed in the sham ultrasound group (Table 3 and Additional file 3).

Craniocervical flexion test: No significant differences were observed in the group-by-time interaction (F = 2.60; df = 70; p = 0.081) (Table 3). However, there was a significant reduction in the electromyographic activity of the upper trapezius muscle in the intervention group (p = 0.023) in the fourth stage (28 mmHg) of the test, and the anterior scalene muscle in the fourth (28 mmHg) (p = 0.030) and fifth (30 mmHg) (p = 0.010) stages of the CCFT. The graphical representation of these results can be viewed in the Additional file 4.

Maximum isometric voluntary contraction: No statistically significant differences were observed in the group-by-time interaction (p > 0.05) for clinical data on the MIVC (Table 3). A clinically relevant result was observed in the intervention group for normalized strength during the cervical extensor MIVC and, in both groups, for the peak time of flexors (Table 3 and Additional file 3). There were no differences between the groups over time (p > 0.05) in the electromyographic data during this test.

Cervical muscle endurance test: For clinical data, there was no significant difference in group-by-time interaction (p > 0.05) (Table 3) neither clinically relevant results (Additional file 3). The intervention group showed significantly increased median frequencies of the splenius capitis muscle during the endurance test of the cervical flexors (p = 0.014) and the anterior scalene during the endurance test of the flexors (p < 0.000) and cervical extensors (p = 0.045) when comparing the baseline and final assessments (Additional file 4).

Questionnaires: No significant results were observed in the group-by-time interaction (p > 0.05) for all applied questionnaires (Table 3). For the PSFS questionnaire, there was a significant within-group increase in the total score for both the intervention and sham ultrasound groups (p = 0.001 and p = 0.003, respectively), demonstrating an improvement in functionality throughout the treatment.

For the Global Perception of Change questionnaire, no significant group-by-time interaction (p > 0.05) was found. After six weeks of treatment, the mean of satisfaction in the intervention group was 4.47 (SD = 1.28), while that in the sham ultrasound group was 3.90 (SD = 1.33). At the end of eight weeks, the mean decreased in the intervention group to 3.47 (SD = 1.17) and in the sham ultrasound group to 3.50 (SD = 2.18) (Fig. 2A). The results also show that, for both groups, throughout the treatment, the improvement was categorized as “moderately improvement” to “improvement” (Fig. 2B).