Migraine and metabolic patterns in nationwide health records

In nationwide records, clinically significant migraine was much more common in women and was associated with several unexpected population-level differences, including lower glucose, HbA1c, obesity, and microalbuminuria, alongside slightly higher diastolic blood pressure and hemoglobin. These findings may point to metabolic or vascular pathways, but they do not mean low glucose causes migraine or that migraine protects against diabetes.

Study population

The retrospective database study examined clinically significant migraine in Israel using longitudinal electronic records from 2017 through 2022. Nearly 25,000 adults with migraine were characterized and compared with matched people without migraine using diagnoses, laboratory tests, medication use, and physiologic measurements.

Main observations

  • In 2022, recorded clinically significant migraine affected about 8.0% of women and 2.4% of men.
  • Prevalence was highest among women aged 40–59, approaching one in nine.
  • Migraine patients showed lower average glucose and HbA1c, lower recorded obesity, and lower microalbuminuria than matched controls.
  • Hemoglobin and diastolic blood pressure were slightly higher.
  • Most diagnoses came from neurologists, raising the possibility that migraine remains under-recognized in primary care.

How to interpret the metabolic signal

The data may reflect shared genetics, hormonal biology, health behavior, medication effects, selection into clinical care, or other factors. A difference between groups does not show which factor came first and cannot be translated into a treatment recommendation.

The value of the result is in defining a reproducible pattern that can guide mechanistic research—not in claiming that changing glucose or weight will prevent migraine.

Important limitations

Electronic records capture diagnosed and treated migraine, not every person with symptoms. Diagnosis practices and medication use can differ between groups. Residual confounding remains possible, and the study population may not generalize to every healthcare system.

Primary source