Methods

Study population
The Korean Headache-Sleep Study was a nationwide, cross-sectional survey of headache and sleep in the Korean population in adults aged 19–69 years. The study design and methods were previously described in detail [29, 30]. Briefly, we adopted a 2-stage systematic random sampling method in all Korean territories except Jeju-do proportional to population distribution and sampled 2,695 individuals. Subjects were stratified according to age, gender, and occupation. To minimized interest bias, we informed candidates that the survey topic was “social health issue” rather than headache. Trained interviewers conducted structured interviews, using a questionnaire to diagnose headache and sleep disorders including RLS by door-to-door visit and face-to-face interview. The interview included questions on the symptoms of headache and RLS. All interviewers were employed by Gallup Korea and had previous social survey interviewing experience. The interviewers were not medical personnel. The study was undertaken from November 2011 to January 2012, and was approved by the institutional review board/ethics committee of Hallym University Sacred Heart Hospital and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from all participants.

Migraine assessment
We diagnosed migraine using a questionnaire (Additional file 1). The questionnaire established a headache profile, which was designed to comply with the second edition of the International Classification of Headache Disorders (ICHD-2). We investigated severity of headache based on effect of daily activity (mild, moderate and severe) and using visual analogue scale. Migraine was diagnosed based on the ICHD-2 criteria for migraine without aura (code 1.1) [31]. We did not attempt to separately diagnose migraine with aura and migraine without aura. As such, both were included in the present study. The questions used to diagnose migraine were previously found to have 75.0 % sensitivity and 88.2 % specificity, by comparing the diagnoses from the survey with doctors’ diagnoses obtained from an additional telephone interview. The validation process was described in one of author’s (MKC) previous article in detail [3].

Non-migraine headache assessment
If a participant responded positively to the question, ‘In the past year, have you had at least 1 headache lasting more than 1 min?’ and was not diagnosed as having migraine, she or he was diagnosed as having non-migraine headaches.

RLS assessment
A diagnosis of RLS was assigned if a participant responded positively to all 4 questions based on the IRLSSG criteria published in 2003: (1) ‘Do you have unpleasant sensations such as crawling or pain in your legs combined with an urge or need to move your legs?’ (2) ‘Do these feelings/symptoms occur mainly or only at rest?’ (3) ‘Does movement improve these unpleasant sensations?’ (4) ‘Are these symptoms worse in the evening or at night than in the morning?’ [14].

Covariate information assessment
We assessed the socioeconomic and demographic characteristics of participants by using a questionnaire. Sleep quality was assessed by the Pittsburgh Sleep Quality Index (PSQI) questionnaire considering total scores of >5 as ‘poor sleep quality’. The Korean translation of PSQI has previously been validated [32].

Statistical analysis
We compared the clinical characteristics of migraine participants with RLS and migraine participants without RLS using Student’s t-test for continuous variables and the chi-square test for categorical variables. For comparing RLS prevalence among non-headache control, non-migraine headache, and migraine participants, we used the chi-square test.
We used logistic regression analysis models to evaluate the association between migraine and RLS and calculated odds ratios (ORs) and their 95 % confidence intervals (CIs). We conducted multivariable logistic regression analysis adjusting for demographic variables and sleep quality. Specifically, the multivariable model controlled for gender, size of residential area, educational level, and sleep quality (PSQI score >5).
For assessment of the association between migraine and RLS according to age, we divided our participants into 5 age groups (19–29 years, 30–39 years, 40–49 years, 50–59 years and 60–69 years) and conducted logistic regression analysis for each age group.
As with most survey sampling designs, non-response resulted in data missing from several variables. The data reported are based on the available data. Sample sizes of some variables diverge from the total sample size of 2,695 because of non-responses for that particular variable. Imputation techniques were not employed to minimise non-response effects [33].
In all statistical analyses, the significance level was 0.05, unless otherwise specified. The results were analysed using the Statistical Package for the Social Sciences 22.0 (SPSS 22.0; IBM, Armonk, NY, USA).