Using Perceived Stress to Predict Episodic Migraine Attacks

man rubbing eyes under his glasses
man rubbing eyes under his glasses
Future predictive models may need to include other migraine risk factors to enhance predictive accuracy.

According to results from a study published in Headache, among individuals diagnosed with episodic migraine (EM), stressful events can predict, with good accuracy, an individual’s risk of a headache even before the onset of prodromal symptoms. The predictive model could potentially allow for the prescribing of preemptive treatment in order to prevent an attack.

The Headache Prediction Study (HAPRED), conducted by a multicenter team of investigators, evaluated the efficacy of a forecasting model in development by assembling a cohort of 95 patients with EM with or without aura who contributed 4626 days of electronic diary data to the study.

A total of 1613 headache episodes were recorded over 4195 days (after accounting for 431 days of missing data). “The prediction model does require daily participation by the patient in the form of completing diaries,” study investigator Timothy T. Houle, PhD, of the department of anesthesia, critical care, and pain medicine at Massachusetts General Hospital in Boston, told Neurology Advisor. “The diary entries are not too cumbersome and only take about 5 minutes to complete. The good news is that the more diary entries completed by the participant, the more accurate the predictions.”

The diaries captured the occurrence, intensity, and duration of headaches and abortive medications used. Reported stress levels (using the Daily Stress Inventory, a 58-item scale that helps to subjectively identify stressors experienced during the day)2 were generally low to moderate, which corresponded with moderate levels of headache disability.

Each of the 4 different models tested predicted similar levels of future migraine events across variable headache and stress patterns, showing good accuracy. Only moderate rates of resolution were achieved, however. The probability of prediction corresponded directly to each patient’s general risk of a headache, such that an individual with a 1 in 4 risk of a migraine attack was able to predict only 25% of future attacks.

A forecasting goal of 100% is achievable, according to Dr Houle. “To refine the predictions, we will need to tap into other forms of migraine risk. For example, many studies have discovered that weather is at least modestly associated with headache activity, so an enhanced model to consider the weather conditions near the patient might be able improve prediction,” he explained.

Related Articles

The predictive model is in the early stages of development and expansion to include domains other than stress will be evaluated in future studies. “We think the model might be made to be clinically useful in the near future,” Dr Houle said. “We are actively recruiting for a validation study to ensure that the predictions are as accurate as previously found, and that they are useful for a range of headache conditions like tension-type headache.”


  1. Houle TT, Turner DT, Golding AN, et al. Forecasting individual headache attacks using perceived stress: development of a multivariable prediction model for persons with episodic migraine. Headache. 2017;57(7):1041-1050.
  2. Brantley PJ, Waggoner CD, Jones GN, Rappaport NB. A Daily Stress Inventory: development, reliability, and validity. J Behav Med. 1987;10(1):61-74.