BALTIMORE—Multiple dimensions of sleep may temporally precede acute risk of migraine attack among patients with episodic migraine, according to a study presented at the 32nd Annual Meeting of the Associated Professional Sleep Societies. While some aspects of sleep appear to protect against next-day headache, “these novel pilot data support the hypothesis that self-reported fragmented sleep is associated with higher risk of migraine two days later,” said Suzanne M. Bertisch, MD, MPH, and colleagues. Dr. Bertisch is an Assistant Professor of Medicine at Harvard Medical School and a Director of Behavioral Sleep Medicine at Brigham and Women’s Hospital in Boston.
Suzanne Bertisch, MD, MPH
Retrospective studies have indicated that nearly half of patients with migraine identify too little or too much sleep as a migraine trigger. One prospective study reported a higher incidence of headache, both migraine and tension type, after two consecutive nights of four or fewer hours of sleep, as measured by self-report. This study accounted for daily stress but not other potential triggers of migraine. “To date, there have been no prospective studies on the association between objectively assessed sleep parameters and migraine incidence while accounting for other potential triggers of migraine,” Dr. Bertisch said.
A Cohort Study of Migraine Triggers
She and her colleagues assessed the independent contribution of sleep characteristics as temporal precedents of migraine. “We were particularly interested … in sleep duration, fragmentation, and self-reported quality.” To examine these factors, they developed a cohort study of migraine triggers that they conducted from March 2016 to August 2017.
The researchers enrolled 101 adults with episodic migraine from the greater Boston area. Inclusion criteria included at least two migraines per month but fewer than 15 headache days per month, a history of migraine for at least three years, and fulfillment of ICHD-3 criteria for episodic migraine. Exclusion criteria included untreated obstructive sleep apnea, pregnancy, and current opioid use.
Data were collected for six weeks. Study participants were prompted to complete morning and evening diaries that recorded information on sleep, including the pattern, fragmentation, and sleep quality; physical activity and daily mood; medications; and headache characteristics. Patients also wore wrist actigraphs for the duration of the six-week study. Each patient had about 40 days of diary and actigraphy data.
Patients reported the onset and duration of headaches, associated symptoms, whether their pain was a headache or a migraine, maximum pain intensity, and any abortive medications used. Data on daily covariates such as alcohol and caffeine consumption, self-reported physical activity, menstrual cycle, stress, and mood prior to bedtime also were collected.
Dr. Bertisch and colleagues used self-matched case–crossover analyses. “Each person served as [his or her] own control. This approach accounts for time-invariant confounders, including sex, genetics, and usual migraine frequency. We used a conditional logistic regression model that was self-matched by day of the week, because there might be an influence of weekend versus weekday sleep patterns, as well as migraine, and we adjusted for time-dependent covariates, including daily alcohol and caffeine use.”