Harvard Medical School researchers report that higher day-to-day “trigger surprisal” scores were associated with migraine attacks over the next 12 and 24 hours. In this cohort, higher surprisal scores aligned with greater odds of a headache attack, even after accounting for recent headache history and fluctuations in daily experience.
Migraine management often revolves around trying to identify and control conditions that might trigger them. From food and drink and environmental exposures to physiological and psychological stressors, it is a crowded landscape of potential trigger suspects. Individuals attempting to match migraine attack timing to these suspected triggers rely on memory and a self-selected list of candidates rather than controlled experimental designs.
In the study, “Information-Theoretic Trigger Surprisal and Future Headache Activity,” published in JAMA Network Open, researchers conducted a cohort study to evaluate the association between surprise and future headache attacks.
Investigators treated the surprisal scores as individual measures of how atypical a given day’s experiences were and examined how those scores related to the next 12 and 24 hours of headache activity.
A total of 109 adults with migraine participated, all diagnosed using the International Classification of Headache Disorders, 3rd Edition criteria. Inclusion criteria required four to 14 headache days per month and ages 18 to 65 years.
Daily life during the study was captured through twice daily electronic diaries, one in the morning and one in the evening, over up to 28 days. Diary items were chosen to cover a wide range of behavioral, emotional, and environmental migraine triggers. Headaches occurred on 1,518 of 5,145 days with complete diary data.
Morning diaries focused on sleep duration and quality, nighttime awakenings, bedtime, wake time, late night meals, weather influences, and mood. Evening diaries again assessed mood and added questions on commonly reported food and drink triggers, environmental exposures, meal patterns including missed meals, weather influences, and daily stressors.
For the main exposure variable, investigators treated each diary response as part of an individual-specific pattern of experiences from which surprisal could be calculated.
Patterns of pain
Surprisal score analysis showed that when a person has a day which feels very different from their usual routine, migraine risk over the next 12 to 24 hours tends to increase.
If the previous day looked pretty typical, that jump in risk is steep as surprise goes up. If the previous day was already unusual, another unusual day adds much less extra risk, and the 12-hour window can even line up with lower risk in the model.
Over 24 hours, more surprise means more migraine risk, but the effect gets weaker when the previous day was already unusual.
The total score was associated with the odds of a future headache attack, rising by 86% at 12 hours and 115% at 24 hours, with a stronger association at the longer interval.
People differed a lot. Some were very sensitive to changes in their daily routine, while others showed almost no link between unusual days and migraines. For individuals with higher baseline headache risk, the predictive value was lower, suggesting that migraine activity might be related to factors other than daily behavioral or environmental changes.
How to interpret the findings
Surprisal may be useful for migraine self-management because it reflects how out-of-the-ordinary someone’s day feels, rather than focusing on specific “yes/no” triggers.
Researchers conclude that surprisal could work as an individualized measure for estimating short-term migraine risk and suggest that surprisal scores could eventually be built into apps to help forecast migraine risk in real time. Instead of avoiding specific triggers, approaches that support steadier routines or emotional regulation might help reduce surprisal and resulting attacks.
Study limitations include the smaller than planned sample size, which was reduced from an original target of 200 to 109 because of pandemic-related disruptions.
That reduction limited the ability to examine specific subgroups such as age bands or headache subtypes. It also left out some possible mitigating factors such as medication use.