Most commonly affected anatomical areas
This study showed that the broadest anatomic areas of the head are the most commonly affected sites during migraine attacks in both EM and CM groups – i.e., the frontal, temporal, and parietal regions.
The temporal and frontal involvement in migraine pain have already been reported by Kelman as the second (58.3%) and third (55.9%) most prevalently affected regions one-to-two thirds of the time (3). Kelman’s paper was the first to focus on migraine pain location in a sizable sample – 1,283 subjects (350 CM cases). However, this report showed the orbital region (67.1%) as the most prevalent area.
In a study of 200 EM subjects conducted in a headache center, the most prevalent areas were the frontal (43.8%) and temporal (42.4%) sites (20) as well. The orbital and occipital sites gain more importance in a population study in Hungary, where the most common sites were temporal (45%), orbital (45%), occipital (40%), frontal (28%), vertex (22%), and generalized (18%) (21). Therefore, it is possible that the study design has a role in these different pain location patterns.
Pain drawings were used to explore differences between EM (n = 48), CM (n = 30) and cervicogenic headache (n = 36) (6). The most commonly affected sites in the EM group were right frontal, right temporal, and left orbital (81.25% for each location). For the CM group, the sites were left frontal and left temporal (83.33%) followed by right frontal and right temporal (80%). These results are quite similar to our findings, probably due to the use of direct recording of pain location. However, we notably found smaller proportions for these areas (frontal – CM, 65% and EM, 73%; temporal – CM, 73% and EM, 67%). The differences in the recording method might have contributed to this discrepancy. The limitation to no more than 3 points per image could have been responsible for some underestimation in our findings in relation to the use of pain drawings. We decided on this limit with the purpose of better characterizing the most important areas that are affected in migraine attacks. Therefore, it is possible that subjects have selected the most affected sites and/or the ones presenting the most severe pain.
Finally, recording pain location during a migraine attack might possibly lead to different results. Olesen studied 750 patients during a migraine attack in order to extract more precise data. The most common pain locations were right hemicrania (n = 168), whole head (n = 147), left hemicrania (n = 134), and bifrontal (n = 98) (2).
Neck pain, migraine and chronification
In our sample, only the posterior cervical region showed a different prevalence between the two groups, with more patients being affected in the CM group (CM 33%, EM 21%).
The intriguing involvement of the neck – a structure outside the area supplied by the trigeminal nerve – in migraine pain has been long recognized (22). When compared to our findings, most studies report a higher NP prevalence in migraine. A study conducted in a headache clinic showed a 70.5% NP prevalence (20). In a population study, NP prevalence was 76.2% for those with pure migraine, 89.3% for those with coexisting tension-type headache (TTH), and 83.3% for those with episodic migraine with or without episodic TTH (23). As discussed earlier, our recording method and the 3-point limitation per image might have been responsible for this discrepancy.
Interestingly, the aforementioned population study (23) showed that the frequency of migraine attacks is correlated with the number of days with NP (r = 0.33, p < 0.001). The largest study dedicated to the topic reported a higher frequency of NP with migraine chronification (3) as well as a higher frequency of occipital and diffuse pain (3, 24). In a prospective study of a sample selected in a headache clinic and the general community, NP prevalence varied according to headache pain intensity: mild (42.8%), moderate (61.1%), and severe (72.6%) (24). Again, NP prevalence directly correlated with headache frequency (r = 0.32). Lastly, the recent use of pain drawings (6) showed a marginal evidence for higher pain extent in the posterior region of the head in the CM group when compared with the EM group (p = 0.07). Therefore, the association of frequent migraine attacks and/or CM with the presence of NP has been a recurrent finding in different studies.
Other neck pain-associated factors in migraineurs
In addition to group classification, NP was also associated with the presence of at least one cardiovascular risk factor; longer-term migraine; more diffuse, frequent, and intense attacks; the presence of mechanical and tactile allodynia; and the presence of medication overuse and prodromal nuchal rigidity. The multivariate analysis adjusted for migraine duration, number of affected anatomical regions, tactile allodynia, and medication overuse. An increase in the odds of NP would be expected with the presence of the two latter factors and/or more diffuse pain. The absence of factors considered to be associated with NP using unadjusted analysis in the final model does not mean that they exert no effect on the odds of NP. It simply means that only these four terms were included in the simplest model that was best fitted to our sample. This is particularly important in the case of high correlated covariates during the model selection process.
The presence of more diffuse pain was the most important NP-associated factor. NP could simply represent a preferred location in the pain spreading process seen in those with higher chronification risk. Data from the Chronic Migraine Epidemiology and Outcomes Study (CaMEO) were used to investigate the association of the presence of non-cephalic pain in eight body regions with the EM-to-CM progression and CM persistence over a 3-month period (25). At baseline, the CM group showed 1.09–1.29 more non-cephalic pain locations than the EM group. At 3 months, each additional location exerted some effect on CM odds independently of other covariates (demographics, depression/anxiety, allodynia, BMI, and baseline acute headache treatment).
A latent class analysis of CaMEO data provided interesting results (26). Firstly, the conditional probability of self-reported NP was more common in the latent class characterized by having the most comorbidities (82%) and in the two classes in which pain was a distinguishing feature – respiratory/pain (79%) and pain (76%). This is consistent with our findings of an association of cardiovascular comorbidity with NP (in the unadjusted analysis only). Secondly, these 3 classes had a higher CM prevalence, higher MIDAS scores, and more allodynia and medication overuse. The association of NP and CM has been previously discussed. We did not find an association of headache intensity/severity with NP, probably because our definition of NP was based on patients indicating the sites affected by pain rather than self-report. An association of allodynia in EM with attacks of higher intensity (p = 0.016) has been reported (27).
Our final model explicitly showed that the interaction between allodynia and medication overuse was more important than each individual factor. The large Migraine in America Symptoms and Treatment (MAST) study investigated allodynia predictors and included 15,133 subjects (28). The fully adjusted model showed that medication overuse was independently (other variables included sociodemographic characteristics, headache frequency, migraine severity, pain intensity, depression and anxiety) and significantly associated with increased odds of having allodynia (OR 1.23, CI 1.09–1.38).
Although we did not explore the sensitivity to pressure, several studies demonstrated lower pressure-pain thresholds (PPT) in migraineurs when compared with controls including the cervical and distant extra-trigeminal areas (29–31). There is an anterior-to-posterior crescent gradient of PPT in the scalp (29) of migraineurs (and healthy controls) mimicking the sites discussed above as the most frequently affected by migraine pain.
Finally, migraine years showed marginal evidence (p = 0.078) of association with NP. The largest migraine pain location study showed that a trend towards pain in the neck and the eyes might exist for longer-term migraine (3). The same author found that 50 year-and-older patients might have attacks without the typical migraine characteristics (photophobia, phonophobia, nausea, vomiting, throbbing pain) (32). The presence of NP was also more common (p = 0.004) in the 30–49 (42.5%) and > 50 (41.1%) age groups than in the 16–29 age group (30.7%).
Possible mechanisms of neck pain in migraine
The involvement of extratrigeminal areas – such as the neck – in migraine could be explained by the convergence of trigeminal and cervical inputs (33) and/or may represent the clinical manifestation of the peripheral sensitization mechanisms found in chronic pain conditions (34). Calhoun et al. have explored the role of NP in migraineurs through a series of studies, and they found that: (1) NP is prevalent in migraine; (2) its presence on the day preceding migraine is associated with treatment resistance; and (3) it is a predictor of disability independent of migraine frequency and severity (35). They raised the possibility that NP in migraineurs represents hyperalgesia or allodynia.
A mechanical process may be present. Women with CM have greater neck extensors (splenius capitis and upper trapezius) activity compared to healthy controls, which suggests that motor adaptations may play a role in the chronification process (36).
Migraine pain location has been reported to potentially change to the region of an injury (“migraine remapping”), and the authors state that this could impact the evaluation of facial and neck pain (37).
Migraine and the neck in clinical practice
A neck pain complaint presented by a patient could represent some disorder other than migraine, with a special concern for the cervicogenic headache (CH) in clinical practice. A detailed clinical history and a careful physical examination are generally enough to differentiate between these two types of headache. A side-locked pain and a posterior-to-anterior radiation pattern favor the diagnosis of CH (1). Some provocative maneuvers can be used to assess the neck of patients without pain during evaluation, and the distraction test could be used otherwise (38, 39). However, care must be taken not to base the diagnosis solely on these maneuvers. The eliciting of local pain and referred pain to the head upon upper cervical spine palpation are more common in migraineurs (88.8% and 46.6%) than in healthy controls (49.3% and 15.5%) (40). It seems reasonable to assume that NP in migraineurs is mostly a manifestation of their migraine. Up to 77% of patients who believe they suffer from a “cervical pain syndrome” receive a final diagnosis of migraine after an evaluation in a tertiary headache center (41).
For some migraineurs, it may be harder to understand that their pain is due to migraine. Less than 8 years of formal education (OR = 1.72-20; p = 0.004), absence of migraine aura (OR = 1.37-20; p = 0.015), and pain in the back of the head at attack onset (OR = 14.2–167; p < 0.0001) are predictors of migraine that is self-diagnosed as a different NP syndrome (41). The latter is also associated with the initiation of a late acute treatment (p < 0.01) in face of greater pain intensity (p < 0.001) (35). The authors speculate that migraineurs may fail to relate NP to their migraine attack, which potentially leads to suboptimal treatments. This could, in turn, increase the chances of chronification. NP has been considered as an important predictor of disability (42), and higher Neck Disability Index scores may be found in CM than EM (p < 0.001), indicating a role of NP in the overall burden of chronification (43).
Other studies focused on the prevalence of cervical musculoskeletal dysfunction in the migraine group (44). This is a debated topic with divergent results from different studies (45), and it is beyond the scope of this text.
Strengths and limitations
To the best of our knowledge, this is the largest study that used an explicit non-verbal recording method to locate migraine pain. The diagnosis and evaluation by a headache specialist further support our findings, as well as balanced groups of individuals with EM and CM.
Some limitations must be considered, namely: (1) The cross-sectional design allows us to establish associations of some explanatory variables with NP not as a causal relationship; (2) we did not stipulate a migraine-free period before the interview, and memory bias may have interfered with our results; (3) we did not use a standardized instrument to measure allodynia; (4) as discussed earlier, the 6-point limitation may cause an underestimation of the sites affected by migraine pain; (5) our results are mainly based on patient reports and medical records. Future studies should consider a prospective record of pain location. Also, recording whether the neck and other subregions respond differently to established migraine treatments would be interesting.