To identify naturally occurring clusters of chronic migraine patients, and to examine relationships among several chronic migraine clinical features.

Agglomerative hierarchical clustering analysis wasconducted to identify natural cohorts in 87 chronic migraine patients using 13 migraine related clinical variables. Principle component analysis was used to condense the 13 clinical variables to those explaining the largest variation. Association analysis between these clinical features was conducted; Benjamini-Hochberg procedure was used to correct for multiple testing at false discovery rate of 5%.  

Two major clusters were identified. Higher levels of self-efficacy and exercise were found in Cluster 1, while higher levels of depression, somatic symptoms, anxiety, pain catastrophizing, poor sleep quality, post-traumatic stress disorder, and migraine-related disability were found in Cluster 2. Similarly, principle component analysis revealed one major pattern of clinical features (eigenvalue 3.8) characterized by depression, somatic symptoms, anxiety, pain catastrophizing, poor sleep quality, migraine-related disability, and negatively loaded by self-efficacy and exercise levels. Exercise levels were significantly associated with pain self-efficacy (Benjamini-Hochberg p= 0.002), while negatively correlated with depression (p= 0.001), somatic symptoms (p= 0.014), migraine-related disability (p= 0.017), poor sleep quality (p= 0.019), and anxiety (p= 0.032).