Abstract
Background
Most migraine patients report neck pain as part of their migraine symptomatology, but it is unknown whether triggering neck pain would induce migraine attacks. Our aim was to assess the occurrence of headache and/or neck pain after an endurance test of the neck muscles among migraineurs and controls.
Methods
Sixty-five patients with migraine and 32 headache-free participants underwent a manual examination of the cervical spine by an assessor blinded towards the diagnosis and were sub-classified according to the appearance or absence of neck pain. Subsequently, the endurance of the neck flexors and extensors was tested three times, in a random order. The maximum sustained duration was recorded and the test was terminated when the subject was unable to maintain the position or reported pain. On the day after the assessment, participants were asked to report the potential occurrence of headache or neck symptoms.
Results
None of the controls reported headache after assessment, while migraine-like headache was reported by 42% of the patients with migraine (p < 0.001) after 15.8 h (SD: 10.0). Neck pain was more prevalent in migraineurs compared to controls (45% vs. 16%, p = 0.006). When considering the neck pain subtype, there were no differences among the three profiles regarding neck pain but participants with referred pain to the head reported a migraine attack more often (45%, p = 0.03).
Conclusion
Patients with migraine are more likely to report neck pain and migraine attacks following a neck muscle endurance test. Participants with neck pain referred to the head during manual examination had a greater prevalence of migraine attacks than those without or with only local pain.
Introduction
The association between migraine and musculoskeletal dysfunction has been repeatedly demonstrated (1–4). Compared to headache-free individuals, patients with migraine are more likely to report neck pain (5,6), altered cervical alignment and mobility (2,7–10), myofascial trigger points (2,8,11,12), and increased muscle sensitivity (13–15). Furthermore, the presence of neck pain is related to a greater disability level in this population (16,17), potentially contributing to migraine chronification (16,18). Neck pain is more prevalent than nausea, but not an official criterion for migraine diagnosis (6).
The temporal relation between these two conditions has been an ongoing debate. Neck pain and/or stiffness were for decades considered to be a premonitory symptom (19–21), since they are already among the most frequent complaints during the premonitory phase, affecting more than 30% of migraine patients (22). Many patients also report that the neck pain starts concomitant with the headache onset, accompanying most of the migraine attack (6,23–25). Neck pain is further described by almost 40% of the patients as a migraine trigger or an aggravating factor (26). Of note, migraine patients with neck pain are more likely to experience exercise-triggered attacks, including running and fitness training (27).
Neck pain can be understood as a peripheral manifestation of migraine (23). On the other hand, the cervical afferences may also act as triggers of the central mechanisms known to be involved in a migraine attack (15,23). A convergence among the trigeminal and the first three cervical nerves at the caudal part of the trigeminal nucleus in the brainstem has been suggested to allow nociceptive stimuli from the neck to activate the trigeminovascular system (28–31). A recent study also identified that a third of migraineurs perceive referred pain in the head after a manual palpation of the upper cervical spine (32).
Since distinguishing premonitory symptoms from migraine triggers through patient retrospective surveys may be challenging (21,33), experimental triggering studies are encouraged to explore the factors involved in the initiation of a migraine attack (21). Therefore, the aim of this study was to explore the occurrence of both migraine attack and neck pain provoked by an endurance neck test in migraineurs and controls, stratifying participants according to the presence of pain during manual palpation of the upper cervical spine. We hypothesised that patients with migraine are more likely to experience neck pain and that an endurance training of the neck may trigger a migraine attack. We hypothesised further that patients with pain during manual palpation of the upper cervical spine will have the greatest risk for both (32).
Material and methods
Sample
This experimental study was conducted at the University Medical Center Hamburg-Eppendorf (UKE), Germany. It was approved by the local ethic commission (process number: 18–350), followed the Helsinki statement (34) and signed consent was obtained from all participants before the experiment. Patients with migraine (n = 65) were screened in a tertiary headache clinic by neurologists with expertise in headaches and were diagnosed according to the International Classification of Headache Disorders (ICHD-3) 3rd edition (19). Adult males and females, with no other headache diagnosis (i.e. MOH, cervicogenic headache), with migraine for at least 2 years and with a minimum of six attacks per year were included. Headache-free participants (n = 32) were recruited from the community. Participants with one of the following conditions were excluded: i) any pathology of the cervical spine (e.g. disc disease), ii) history of trauma to the head or neck (e.g. whiplash associated disorder), iii) rheumatoid disease, iv) craniomandibular dysfunction, v) any other musculoskeletal, neurological, or psychiatric disease, vi) migraine attack on the day of the appointment.
Procedures
An assessor blinded to the participants’ diagnosis (GFC) carried out all assessment procedures. After the physician consultation, eligible participants were asked to lie in a prone position with the hands supporting the head. According to a previously described procedure (32,35), manual pressure was applied on the midline of C1, spinous processes of C2, and bilateral joints of C0/1 and C1/2. The palpation was performed with up to five oscillating movements, followed by sustained pressure (maximum 5 sec). Participants were asked about the presence of local neck pain or neck pain referred to the head, immediately after the manoeuvre applied to each segment.
Afterwards, the participants underwent a muscle endurance test for the neck flexors and extensors as detailed by previous studies (36,37). To test the extensors, participants were instructed to sustain a load (2 kg for women and 4 kg for men) applied around the head above the ears, in a prone position with legs straight and arms positioned at the sides. The subject then extended and raised the head just above the examination table, with the tip of the chin pointing against the floor. For the flexor muscles, the subject was positioned supine with legs straight and arms alongside the body. They were instructed to flex the upper cervical spine and perform “a small nod of the head and hold the chin against the chest” and raise the head above the examination table until exhaustion. All participants received standard verbal encouragement and feedback (36). The order of the tests was randomised and the time in seconds was recorded with a stopwatch, with three repetitions for each test. The test was terminated when the subject was not able to maintain the head position or decided to stop the trial, due to pain or fatigue.
Immediately after the physical tests, participants were asked to rate on a visual analogic scale (VAS, 0–100) their perceived level of neck fatigue and pain during the tests. They also received a set of questionnaires with inquiries regarding demographics, neck pain and headache features. Furthermore, the neck disability level, using the Neck Disability Index (NDI-G), was administered (38). In the 2 days after the assessment, the assessor contacted the participants and inquired regarding the presence of neck pain (onset, intensity and duration) and headache (onset, intensity, duration and migraine-like features) following the experiment.
Statistical analysis
After a pilot study, a sample size calculation was performed using G*Power 3.1.9.2 software. The number of participants was defined based on an effect size of 0.66 between the migraine and control groups in the outcome of extension endurance time, with alpha level of 5% and a power of 80%. This calculation resulted in a number of 30 participants per group. In order to implement the subclassification according to manual examination of the upper cervical spine, the number was increased to 65 migraineurs and 32 controls.
Relative frequencies, mean and standard deviations (SD) were calculated to demonstrate characteristics of variables. All continuous variables were normally distributed according to the Kolmogorov-Smirnov test (p > 0.05) and were compared using Student’s t test or analysis of variance (ANOVA). Categorical variables were compared using the chi-square or Fisher’s exact test. Relative risk to report migraine after the experiment was also calculated, using controls as reference. Data were analysed in SPSS version 26, at a significance level of 5%.
Results
Migraine patients and controls did not differ regarding age, sex or physical activity level (p > 0.05). None of the patients with migraine received GON blocks prior the assessment and 18% were under prophylactic migraine treatment. Patients with migraine presented a greater prevalence and greater severity of neck pain as well as higher scores of neck disability (Table 1). The average time for sustained neck flexion and extension was lower in the migraine group (flexion 00:40, SD 00:24 seconds; extension: 03:08, SD 2:19 minutes) compared to controls (flexion 01:00, SD 00:49 seconds, p = 0.001; extension 04:18, SD 2:48 minutes, p = 0.009) due to fatigue and local pain. Immediately after the endurance test, migraine patients reported a greater level of neck pain in comparison to controls (migraine group VAS: 39.0, SD: 25.7; control group VAS: 21.2, SD: 25.7; t = −3.2, p = 0.002). No differences regarding neck fatigue levels were verified between groups, immediately after the test (migraine group VAS: 56.3, SD: 23.3; control group VAS: 46.5, SD: 27.9; t = 1.8, p = 0.07).
Demographic characteristics according to the clinical diagnosis.
BMI: body mass index; VAS: visual analog scale; NDI: neck disability index; Sig: significance level.
Twenty two percent (n = 14) of migraineurs and 38% (n = 12) of the controls reported no pain (PF) during the manual examination of the upper cervical spine. Thirty seven percent (n = 24) of the migraineurs presented local neck pain (LP) and the majority (42%, n = 27) presented pain referred to the head (RP). Fifty percent (n = 16) of the controls had LP and 13% (n = 4) presented RP (χ2 = 8.5, p = 0.01). Migraine features such as frequency, duration and presence of a migraine attack in the previous 48 h of the assessment did not differ among the three neck profile subgroups (p > 0.05, Table 2).
Sample demographic characteristics according to the neck profile, including patients with migraine and headache-free controls.
Bonferroni post-hoc: *Neck pain free vs. local pain; **Neck pain free vs. referred pain; †Local neck pain vs referred pain; Sig: significance level.
None of the controls reported any headache in the 24 h after the assessment, compared to 42% of the migraineurs (χ2 = 18.40, p < 0.0001). Neck pain was also more prevalent in migraineurs compared to controls during the day following the assessment (45% vs. 16%, χ2 = 7.90, p = 0.006). When considering the neck pain subclassification, there were no differences among the three groups regarding neck pain after the assessment (PF: 23%, LP: 33%, RP: 48%, χ2 = 1.17, p = 0.12). However, patients with referred pain to the head during the manual examination reported an increased rate of migraine attacks after the evaluation appointment (PF: 23%, LP: 18%, RP: 45%, χ2 = 7.05, p = 0.03, Figure 1). The presence of migraine increased the risk of developing an attack after the neck endurance exercise by 27.5 (95% CI: 1.7–436.9, p = 0.02) times. Having migraine associated with referred pain to the head increased the risk of an attack by 34.2 (95% CI: 2.1–547.6, p = 0.01) times.

Percentage of participants with report of neck pain and/or migraine onset during the first 24 h after the assessment. *migraine attack: χ2 =18.40, p < 0.0001; neck pain: χ2 = 7.90, p = 0.006.
Migraineurs who reported neck or headache symptoms after the assessment had a mean onset time of both symptoms 15.8 h (SD: 10.0) after the assessment. Healthy controls reported neck pain 7.0 h (SD: 9.6) after completion of the assessment (t = −3.3, p = 0.001). The mean duration of the migraine attack reported by the migraine group was 12.5 h (SD: 16.1), and the mean migraine intensity on the VAS scale was 53.0 (SD: 41.0). The neck pain duration reported after the assessment did not differ among migraine (16.6, SD: 18.5 h) and control groups (14.8, SD: 11.7 h, t = 1.9, p = 0.05). The reported neck pain intensity was greater in migraineurs compared to controls (migraine VAS: 41.0, SD: 16.0; control VAS: 26.0, SD: 9.0; t = 3.6, p = 0.001).
Discussion
Our results show that the perceived fatigue level during the neck muscle endurance test did not differ between migraine patients and healthy controls, but migraine patients have a significant risk of triggering a headache. Patients who, following manual palpation of the neck before the endurance test, report a local pain that radiates to the head, exhibit an even greater incidence and risk to report a migraine attack in the following day. These migraine attacks started on average 15.8 h following the experiment.
It has previously been shown that fatiguing exercises can exacerbate pain through enhanced central facilitation, especially in patients with chronic pain conditions (39,40). In our study, similar fatigue levels were reported among migraine and control groups following the endurance test, although endurance duration differed with controls exhibiting a longer perseverance. During the endurance exercise and up to 15.8 h later, migraineurs experienced greater intensity of neck pain compared to controls, which is probably due to exercise effort and delayed onset muscle soreness. Exercise has often been used to induce delayed onset muscle soreness and mimic clinical musculoskeletal pain (39,41–44). The delayed onset muscle soreness initiates between 12 and 24 h after exercise, with a peak between 24 and 72 h (45). Therefore, the neck pain reported by migraineurs in this study is in accordance with the time interval above referenced. We note that although the premonitory phase could have started right after the exercise, patients developed a migraine headache after 15.8 h, mimicking the time delay of pharmacological (46–48) and carbon monoxide (49) provoking studies. One could argue that if the exercise is a trigger, the headache should start immediately afterwards. The neck muscle soreness (reported by the patients as neck pain) and the headache started after this time delay of 12–16 h, which is in accordance with the time frame for initiation of pain after all kinds of intense exercises (50) suggested by the American College of Sports Medicine (45). It is therefore likely that the neck pain and the following headache are linked both to each other and to the exercise performed 16 h earlier. We suggest that it is the neck soreness – and hence an increase in nociceptive input from C2–C5 that triggers the attacks. The question that arises in all trigger studies and has not been investigated so far is whether the individual patient needs to be below a certain threshold to be able to develop an attack (51), which may explain why a certain percentage do not experience attacks after provocation. How an increase in nociceptive input triggers an attack is not understood but we note that the opposite, a decrease in occipital input due to a pharmacological nerve block of the greater occipital nerve, is effective in certain headache types such as cluster headache (52–57), and trigeminal neuralgia (58) and is likely to be effective in migraine (59–64). The evidence for migraine is still less conclusive as one of the placebo-controlled studies was negative (65). However, from a clinical perspective, tenderness seems to be predictive for its clinical efficacy in migraine, whereas the extent of the anaesthetic effect had no influence on the clinical outcome (52). Data from uncontrolled studies suggest that the procedure may even be effective in medication overuse headache (52,66).
A recent study suggested that headache patients and controls differ in concentration levels of calcitonin gene-related peptide (CGRP) following high intensity aerobic exercise (67). CGRP is involved in migraine pathogenesis and plays a role in the central modulation of the nociceptive input, vasodilation of blood vessels and neurogenic inflammation (68). One could argue that a generalised higher level of CGRP following endurance exercise may have played a role in attack generation, but we have not measured CGRP levels in our cohort. It could theoretically explain, however, why some patients complain of sports as a headache trigger. Since the exercise in our cohort was not systematic but rather local, one could also argue that the continuous afferent nociceptive input from the neck induced a sensitisation of trigeminocervical neurons, activating the trigeminovascular system and thus initiating a migraine attack (8,11,28–30). The modulation of the trigeminocervical pathway through tailored and moderate-effort exercises can potentially be beneficial in patients with migraine, since it is already established that regular exercises are also related to reduction in excitability and increased inhibition in their respective sites in the brainstem (39,40). Furthermore, therapeutic and supervised exercises focused on the neck can be related to improvement of migraine frequency and intensity (69–71).
Although we suggest a causal relation of the neck endurance training and the migraine attack, we have not monitored all other possible triggers that may be related to the attack initiation. Given the rather large group, we still think our argumentation is reasonable. A recent study attempted to quantify the conditional probability of occurrence of the migraine attack, in order to provide a model to better study migraine triggers (72). The results suggested that the probability of having a migraine attack on any given day depends on the presence of an attack on the preceding day. In this situation, the probability of continuing an attack on the next day is 50.9% (95% CI: 47.2–54.3). When the preceding day is migraine-free, this probability drops to 8.5% (95% CI: 7.8–9.1) (72). In the current study, just 25% of the patients with migraine reported a migraine attack during the last 2 days. This indicates a much smaller incidence of migraine than the anticipated 42%, according to the described model (72). Furthermore, an increased incidence of migraine pain was observed in the patients who had pain referred to the head, although no differences regarding pain in the last 2 days or migraine frequency were observed among the three neck profile subgroups (32). We note that our study lacks a control group; that is, a migraine group who did not perform the neck endurance test. Another limitation includes expectancy from past experiences, since it is already established that the awareness of a specific trigger may contribute to expectancy mechanisms and therefore increase symptoms in other pain disorders (73). Although no suggestion was made by the blinded assessor, the perceived neck pain during the assessment could have contributed to symptoms reported by the migraine patients.
Despite these limitations, we demonstrate a functional relationship between induced neck pain and migraine. The lack of experimental triggering studies sustained for decades the dilemma whether neck pain is considered a trigger, a premonitory symptom or a part of the migraine attack (20–24,26,27,33,73). Our study strongly suggests that maximal endurance effort can act as a true migraine trigger. This offers new scientific opportunities to trigger attacks without medication. Clinically our findings should be considered during the prescription of therapeutic exercise for this population.
Key findings
Our results demonstrated that patients with migraine are more likely to report migraine attacks and neck pain following a maximum endurance test for the neck muscles. Participants with neck pain referred to the head during the manual examination of the upper cervical spine presented an even greater prevalence of migraine attacks compared to those without or with local pain. Endurance test for the neck muscles is easily accomplished and may be a good alternative to pharmacologically induced migraine attacks.
Footnotes
Acknowledgements
The authors thank Sima Daneshkhah and Annika Schwarz for help with the patients, and Waclaw M Adamczyk for fruitful discussions.
Authors’ contributions
GFC: Methodology design, data acquisition and analysis, drafting and writing of the manuscript. KL: Methodology design, data analysis, drafting and writing of the manuscript. TS: Methodology design, participants recruitment, review and edition of the manuscript. DB: Review and edition of the manuscript. AM: Conceptualisation of the study, methodology design, data analysis, interpretation, drafting and writing of the manuscript. All authors agree with the content of this manuscript.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the German Research Foundation, SFB936/A5 to AM, and by the FAPESP Foundation, 2018/12024-5 to GFC. The funding sources did not influence study conduct in any way.
