The Role of Diet and Nutrition in Migraine Triggers and Treatment

A Systematic Literature Review

Nada Ahmad Hindiyeh, MD; Niushen Zhang, MD; Mallory Farrar, PharmD; Pixy Banerjee, MPharm; Louise Lombard, MNutr; Sheena K. Aurora, MD


Headache. 2020;60(7):1300-1316. 

In This Article


Search Strategy

The systematic review was performed using an a priori protocol. The format of this review was based on the Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA).[17] A literature search was conducted in MEDLINE and EMBASE in the OVID platform to identify studies assessing the effect of diet on migraine. The search strategies combined free-text and controlled vocabulary terms for the disease and outcomes of interest. The search terms are outlined in the Appendix. The literature search was limited to articles in the English language, involving adult human subjects (≥18 years old) and published from January 1, 2000 to March 5, 2019. Inclusion and exclusion criteria were determined as depicted in Table 1.

Study Selection

Eligibility Criteria. RCTs, non-RCTs, prospective or retrospective observational studies, systematic literature reviews and surveys, in which diet, food, and nutrition were assessed as patterns, triggers or an intervention, in people with migraine aged ≥18 years were included (Table 1).

Study Selection. The studies were selected based on a 2-level screening process. Level 1 screening entailed a broad screening based on the titles and abstracts of the citations retrieved. The full text of all citations passing level 1 screening was retrieved for full-text article screening in level 2 and screened for final eligibility for the review. One reviewer completed level 1 screening (MF or PB), while a second reviewer conducted a quality check of a random sample of abstracts. Level 1 screening was verified by 2 expert reviewers (NAH and NZ). Level 2 screening was conducted by 2 independent reviewers (MF and PB) and discrepancies were resolved by consensus within the team or by involving a third team member (LL, NAH, or NZ). All screenings were recorded using prior developed eligibility criteria as described above. Bibliographic searches of systematic literature reviews were conducted to identify relevant studies.

Data Extraction

Data were manually extracted from the reports by 2 researchers (MF and PB) and independent reviewers (LL, NAH, NZ or SA) further verified the extractions. A standardized data extraction form was used to extract the data from each included study and studies were categorized as diet patterns, diet-related triggers, and dietary interventions. The articles were searched for the type of diet, type of study, sample size, migraine type, results of outcome measures as reported by the articles for dietary patterns and triggers (correlation and association measures, prevalence, number or percentage of triggers), and the effect of interventions (effect on intensity, frequency or duration of headache or migraine attacks, pain or other medication use, number of migraine days).

Quality Assessment of Included Studies

All included RCT studies were assessed for quality by the National Institutes for Health and Care Excellence (NICE) checklist for RCTs.[18] The items of the NICE checklist were rated as "yes", "no", "unclear", or "N/A". A "yes" response for an item indicates that the design/conduct of the study minimizes the risk of bias for that item. A "no" response denotes high risk of bias for an item. An "unclear" response to a question may arise when the item is not reported or not clearly reported. "N/A" was used when an RCT cannot give an answer of "yes" regardless of how well it has been designed. The risk of bias was determined according to the responses for each item. The disagreement between researchers was resolved by consensus. The Modified Downs and Black checklist[19,20] was used to assess the quality of the observational studies. The checklist was modified for the scoring of item 27 that refers to the power of the study. Instead of rating according to an available range of study powers (0-5), the rating was performed based on whether or not the study performed power calculation (1 or 0). Therefore, the highest possible score for the checklist was 28 (instead of 32), with higher scores indicating higher study quality.