We showed that performance on validated tests of WM, SRT and CRT worsens during the headache and postdromal phase of a migraine, but not on a non-migraine headache day, in comparison to a patients interictal baseline. This is the first study to utilise home-based computerised cognitive tests to examine the effect of cognition during the phases of a migraine, and a non-migrainous headache day.

These findings are broadly in keeping with several previous studies that have variably reported transient declines in processing speed, working memory and immediate and sustained attention during the headache phase of a migraine [7, 9]. It provides further validation of patient reports of cognitive symptoms during the migraine postdromal phase [2, 3].

Possible explanations for altered cognition in migraine have been summarised previously [13]. Previous observations include increased functional connectivity and fMRI activation in the temporal lobe in patients with migraine, and deficits in task-related suppression during acute pain, irrespective of pain catastrophizing or pain intensity [14,15,16]. Variations in grey matter density have also been reported in patients with migraine [17].

Our study provides several insights into contributory mechanisms to cognitive symptoms in migraine. Firstly, our observation of an impact on test scores on a migraine, but not on a non-migraine headache day, is significant. One hypothesis and possible confounder is that cognitive symptoms and test results relate to ‘distraction by pain’, our observation that there was no significant difference in pain severity between migraine and headache days, but that only migraine days were associated with significant deviations on testing, suggests that distraction by pain is not an adequate explanation for cognitive symptoms. Given deficits in task-related suppression during acute pain seen in fMRI studies however [15], pain remains a significant possible confounder in the design of cognitive study.

Secondly, the findings of our study describe a dynamic process impacting markers of cognition in migraine, that is not completely explained by static changes in grey matter density. In Mathur et al.’s fMRI study, patients with migraine underwent cognitive testing during and in the absence of painful stimuli [16]. Evidence of abnormally blunted cognitive task-related deactivation of the left dorsolateral prefrontal cortex and left dorsal anterior midcingulate cortex suggested alterations of cognitive processing in migraine, which were further modulated by migraine frequency [16]. Furthermore, functional studies have also shown increased activation of cortical areas related to executive function during a migraine attack [18]. Taken together, this suggests maladaptive functional connectivity of altered pain-cognition networks in migraine that are associated with attack frequency, which may explain the association of migraine with cognitive symptoms.

There are several limitations to this study. The lack of observed change during the defined prodromal period may relate to the timing of the test, which was defined retrospectively from the headache phase rather than prospectively from reported symptoms. Secondly, due to the study design and sample size, the generalisability to chronic migraine, and association with disease frequency and duration were not investigated, however the study design, utilising paired samples, controls for variation in education and background. The study population included patients who were on a preventative agent for migraine. As they were all on a stable regimen, this was felt to not be responsible for the observed variation in cognition, and an accurate representation of the lived experience of a patient with migraine. Finally, the effect of other variables such as altered sleep during a migraine was not examined. Given the observation that poor sleep quality may precede a migraine attack [19, 20] and the impact of sleep on cognitive performance [21], this is a significant confounder that requires further study.