Associations Between Depression/Anxiety and Headache Frequency in Migraineurs

A Cross-Sectional Study

Hsuan-Te Chu, MD; Chih-Sung Liang, MD; Jiunn-Tay Lee, MD; Ta-Chuan Yeh, MD; Meei-Shyuan Lee, PhD; Yueh-Feng Sung, MD, PhD; Fu-Chi Yang, MD, PhD


Headache. 2018;58(3):407-415. 

In This Article



This cross-sectional study enrolled 623 subjects that attended an outpatient headache clinic in the Department of Neurology at Tri-Service General Hospital (TSGH) between June 2014 and May 2016. The current study expanded the dataset on which we previously published two articles on migraineurs in association with restless legs syndrome[16] and sleep quality.[4] Patients suffering from migraines, both with and without auras, were enrolled after providing informed consent according to the criteria of the International Headache Society.[17] The study protocol received approval from the Institutional Review Board of TSGH. After excluding patients with medication overuse headache (MOH), those with insufficient data to determine migraine frequency, and those with missing information on clinical characteristics, a total of 588 subjects were finally included for the analysis. Among the patients that participated in this study, 65 subjects suffered from chronic migraines (≥15 headache days/month [hd/months]), 99 had high-frequency migraines (9-14 hd/months), 85 experienced medium frequency migraines (5-8 hd/months), and 160 reported low-frequency migraines (1-4 hd/months). In addition, 179 nonmigraine volunteers were selected with the following criteria: no family history of migraines and no previous diagnosis of other primary or secondary headache disorders, except for infrequent episodic tension-type headaches (<6 days/year). Subjects who completed the screening questionnaire were interviewed by a board-certified neurologist and headache specialist (F-CY) to confirm the diagnosis of migraines and to exclude emotion-related headaches, as determined according to the International Classification of Headache Disorders, 3rd edition (beta version).[17] The sample size determination was calculated according to a previous study,[18] which investigated the correlation between Beck's Depression Inventory (BDI)[19] scores and migraine frequency in 179 women. The mean scores were 15.1 (SD = 12.4), 18.7 (SD = 11.3), and 21.2 (SD = 11.1) in the 53 controls, 37 episodic migraineurs, and 89 chronic migraineurs, respectively. With a pooled standard deviation of 19.1, we obtained an effect size (Cohen's f) of 0.138. Given the alpha of 0.05 and power of 0.80, a minimum sample size of 510 was required. We conducted a power analysis for an F test (analysis of variance).


Patients were interviewed with a structured questionnaire packet containing the BDI,[19] the Hospital Anxiety and Depression Subscales (HADS),[20] a restless legs syndrome (RLS) screening questionnaire from the RLS Foundation, the Pittsburgh Sleep Quality Index (PSQI),[21] and the Migraine Disability Assessment questionnaire (MIDAS). Patients who scored ≥18 on the BDI (maximum score = 63) were classified as depressed. The HADS is a 14-item scale, with seven items related to anxiety and seven items related to depression. Each item is rated on a four-point scale (0, not at all; 1, sometimes; 2, often; and 3, all the time), giving a maximum subscale score of 21. The PSQI estimates sleep quality over the prior month. A PSQI final score of 6 was considered indicative of poor sleep. Participants who answered "yes" to at least 6/11 RLS symptoms screening questionnaire items were considered to have a high probability of RLS. The MIDAS, a five-item questionnaire, evaluates disability over the previous 3 months. We use two tools, BDI and HADS, for the depression severity assessment. We also expected to detect the critical point of depression early, and help with early diagnosis, particularly to prevent possible suicidal behavior.

Data Analysis

Data are reported as means ± standard deviations (SDs) of continuous variables or as numbers and proportions of categorical variables. Patients with migraines were classified into four ordinal migraine frequency groups (described above). Linear trends in the distributions of subject characteristics were assessed across the migraine and nonmigraine control group using the Cochran-Armitage chi-square test for categorical variables or a linear contrast of univariate linear regression for continuous variables. To investigate the effect of migraine frequency on BDI and HADS scores, we performed tests of linear contrast in multivariable linear regression analyses with adjustments for gender, age, body mass index (BMI), education, employment, smoking status, alcohol consumption, coffee consumption, and PSQI score. This analysis was stratified for migraines with versus without auras. Finally, we carried out multivariable linear regression analyses to explore factors associated with BDI and HADS scores. Data analysis was conducted using SPSS 22 (IBM SPSS, Armonk, NY: IBM Corp). The statistical significance was set as alpha = 0.05.