Abstract
Objective
This longitudinal observational study investigated the role of prodromal symptoms in predicting headache onset, using electronic headache diaries to assess symptom occurrence and predictive utility in individuals with migraine.
Background
Prodromal symptoms are early indicators of migraine that occur 2–48 h before the headache phase. Whereas commonly reported symptoms include fatigue, stiff neck, and sensitivity to light and sound, few studies have assessed their predictive abilities for headache onset. Understanding the relationship and reliability of these symptoms is important for early migraine intervention strategies.
Methods
This is a secondary analysis of a longitudinal observational headache diary study. Participants with episodic migraine (N = 43) were recruited from Einstein College of Medicine between 2017 and 2021 and completed a mobile headache diary three times daily over 3 months. They reported the presence of 11 prodromal symptoms and headache anticipation during headache‐free intervals. Associations between prodromal symptoms, anticipation, and subsequent headache onset were examined across multiple time lags. Headache anticipation was evaluated as a moderator of these relationships.
Results
Across 7061 headache‐free observations, prodromal symptoms were reported in 52.7% of cases (n = 3721). Phonophobia doubled the odds of head pain onset 6 h before head pain. Nausea, photophobia, and difficulty thinking/concentrating were associated with higher odds of head pain within 24 h of symptom reporting. Blurred vision was associated with higher odds of future head pain at all time points evaluated. Thirst and dizziness were associated with higher odds of head pain 18–36 h before head pain onset. All prodromal symptoms were associated with increased odds of headache anticipation. Headache anticipation itself was associated with higher odds of headache onset within the ~6, 12, and 36 h prior to headache onset (~6 h odds ratio [OR] = 3.35, 95% confidence interval [CI] [2.57, 4.36]). Anticipation weakened the predictive value of several symptoms on headache occurrence, including feeling tired/weary (OR = 0.50, 95% CI [0.37, 0.68]); difficulty thinking/concentrating (0.37, 95% CI [0.24, 0.58]); thirst (OR = 0.42, 95% CI [0.31, 0.58]); blurred vision (OR = 0.63, 95% CI [0.42, 0.93]); photophobia (OR = 0.15, 95% CI [0.07, 0.33]); phonophobia (OR = 0.19, 95% CI [0.12, 0.28]); nausea (OR = 0.33, 95% CI [0.17, 0.65]); and dizziness (OR = 0.17, 95% CI [0.08, 0.35]).
Conclusion
Cardinal (phonophobia, photophobia, and nausea) symptoms of migraine were all associated with higher odds of future head pain 24 h before head pain onset. Other neurologic and homeostatic symptoms also emerged as predictors for head pain. However, predictive performance was modest; therefore, these symptoms should not be considered stand‐alone migraine predictors. Headache prediction was weaker when individuals anticipated a headache, suggesting that anticipation may influence symptom interpretation.
Keywords: ecological momentary assessment, headache, migraine, predictive, premonitory, prodrome
Plain Language Summary
People with migraine often notice early warning signs before head pain, like feeling tired or being sensitive to light and sound, but it is unclear if these symptoms can be used to predict a headache. In this study, adults with migraine used a smartphone diary three times a day for 3 months to record their symptoms. The findings showed that symptoms of nausea, light sensitivity, sound sensitivity, difficulty concentrating, and blurred vision were the best predictors of a headache within the next 24 h, which suggests that recognizing these signs may help people manage their migraine condition better and treat attacks sooner.
Plain Language Summary
People with migraine often notice early warning signs before head pain, like feeling tired or being sensitive to light and sound, but it is unclear if these symptoms can be used to predict a headache. In this study, adults with migraine used a smartphone diary three times a day for 3 months to record their symptoms. The findings showed that symptoms of nausea, light sensitivity, sound sensitivity, difficulty concentrating, and blurred vision were the best predictors of a headache within the next 24 h, which suggests that recognizing these signs may help people manage their migraine condition better and treat attacks sooner.
Abbreviations
- CI
confidence interval
- GEE
generalized estimating equations
- ICHD‐3
International Classification of Headache Disorders, 3rd edition
- NPV
negative predictive value
- OR
odd ratio
- PPV
positive predictive value
INTRODUCTION
Migraine is among the leading causes of disability worldwide, 1 with an estimate of over 1 billion people worldwide experiencing migraine. 2 Although migraine is primarily known for its episodes of head pain, associated symptoms like nausea and sensitivity to light and sound are also diagnostic criteria. Additionally, neurologic symptoms are increasingly being identified as key components of migraine, occurring either before the head pain onset (known as premonitory or prodrome symptoms) and/or after head pain (known as postdrome symptoms). 3
Prodromal migraine symptoms precede and forewarn migraine attacks, typically within 2–48 h. 4 Prodromal symptoms include: homeostatic or hormonal changes (i.e., thirst, yawning, food cravings, sleep problems), sensory and migranous sensitivities (i.e., photophobia, phonophobia, stiff neck), mood and cognitive symptoms (i.e., difficulties concentrating, mood changes), and cranial autonomic symptoms (i.e., abnormal taste, nausea). 5 , 6 , 7 , 8 To date, studies are inconsistent regarding what symptoms define the prodrome, the time frame in which they occur, and how long the prodrome phase lasts. 8 , 9 , 10
The reported prevalence of prodromal symptoms in patients with migraine has varied widely across studies, with estimates ranging from 30% to 88%. 6 , 8 , 10 , 11 , 12 , 13 This variation is likely due to differing methodologies used to identify prodromal symptoms, which can involve both subjective self‐reports and objective clinical measures. 5 Additionally, the ability to accurately identify prodromal symptoms likely varies across time frame, symptom type, and individual. For example, Giffin et al. 10 found that migraine attacks were correctly predicted less than one‐fifth of the time when prodromal symptoms occurred approximately 18 h before head pain, whereas more than two‐thirds of the time when prodromal symptoms occurred approximately 6 h before the attack. They identified yawning, increased emotionality, difficulty reading, and trouble speaking as the most reliable indicators of migraine. 10 Conversely, Quintela et al. 8 found that most migraine attacks were preceded by prodromal symptoms within 24 h of head pain onset, with concentration difficulties being the most accurate predictor. Understanding when the prodrome phase begins and which symptoms most accurately predict onset are essential for optimizing early migraine treatment.
Despite the growing interest in prodromal symptoms, it remains unclear how accurately individuals can predict the onset of their migraine attacks, which specific symptoms reliably predict migraine onset, and the time frame in which they are present. The current observational study utilizes an electronic headache diary to allow individuals to record their migraine symptoms over 3 months, three times per day, enabling the examination of prodrome symptom frequency, duration, and predictive value. Although the sample size was modest, the ecological momentary assessment (EMA) design yielded a large number of observations, enhancing the study's statistical power.
It was hypothesized that:
Prodromal symptoms would be associated with headache anticipation.
During head pain‐free entries, the odds of head pain phase onset in the next ~6 to 36 h would be significantly higher when headache was anticipated.
During head pain‐free entries, the odds of head pain phase onset in the next ~6 to 36 h would be significantly higher when prodrome symptoms were reported.
Exploratory aim:
Explore whether headache anticipation moderates the relation between prodromal symptoms and subsequent headache onset.
METHOD
Study design
This study is a secondary analysis of a 3‐month longitudinal observational diary study (3 times/day), which served as the baseline assessment for a randomized control trial with a goal to evaluate a clinical decision tool to support patient migraine management (NCT03706794). This study was approved by the Albert Einstein College of Medicine (New York, NY) Institutional Review Board (2015‐5743).
Participants
In the parent study, participants were identified through the electronic health records of a large urban healthcare system and online advertising between October 2017 and July 2021. Participant inclusion and exclusion criteria were guided by the American Headache Society Behavioral Clinical Trials Guidelines. 14 Inclusion criteria were: ICHD‐3 diagnosis of migraine 4 (documented in their electronic health record and/or via validated screening), having 6–14 headache days per month (confirmed via self‐report and diary monitoring), currently prescribed a migraine‐specific medication for acute migraine management, and aged 18–65 years. Exclusion criteria included not having a smartphone, medication overuse, a plan to change migraine medication during study participation, current pregnancy, or a plan to become pregnant during the study.
Procedure
Participants were screened remotely by master's‐level clinical psychology graduate students trained by the principal investigator, a licensed clinical psychologist, and included a final screening for inclusion/exclusion criteria, electronic informed consent (via Research Electronic Data Capture, a Health Insurance Portability and Accountability Act–compliant online data capture system), semi‐structured interviews assessing headache (Structured Diagnostic Interview for Headache–4 Brief Version 15 ; to confirm that migraine criteria were met) and psychiatric diagnoses (Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders Disorders 16 ), questionnaires assessing individual difference variables, education about optimal migraine management (to standardize migraine management recommendations), and headache diary use information. Eligible participants received instruction on using a mobile headache diary app and completed 1 month of daily diary collection (to confirm eligibility in the study); those who remained eligible completed an additional 2 months of daily diary collection, with a total of 3 months of baseline diary collection analyzed in the current secondary analysis.
Participants were contacted biweekly by the research coordinator to provide feedback and offer troubleshooting and were contacted if they failed to complete 2 consecutive days of diary entries. At the conclusion of the 3 months, participants were provided with a compensation gift card.
Measures
Headache status and anticipation
Diary time points were randomized to be on average 6 h apart. At each diary entry, participants reported their headache status by endorsing either: (1) I do not anticipate a headache, (2) I think I might get a headache (“headache anticipation”), (3) I am currently experiencing a headache (“current head pain”), or (4) I had a headache since the last entry, but I am now free of pain. Additionally, they reported any acute medication use.
Prodromal symptoms
After each diary entry during which no head pain was present, participants reported the presence of 11 common prodromal symptoms, selected based on their identification in prior studies to represent common sensory, cognitive, affective, and somatic prodromal symptoms. 10 , 17 , 18 These symptoms included stiff neck, sensitivity to light, sensitivity to sound, feeling tired or weary, difficulty thinking or concentrating, irritability, yawning, nausea, thirst, blurred vision, and dizziness.
Data analysis
Because this study is a secondary analysis, no a priori power analyses were conducted for the specific hypotheses tested here. The sample size was determined by the number of participants and diary observations available from the parent study.
Statistical analyses were performed using SPSS version 20.0 (IBM, Armonk, N.Y., USA). Inferential tests were two‐tailed with 95% confidence intervals (CIs); a two‐sided alpha level of 0.05 was considered statistically significant for all analyses. All variable distributions were visualized. Descriptive statistics, summarized with means and standard deviations, and frequencies were obtained for demographic information and headache days. Prodromal symptoms reported during headache‐free intervals were evaluated using frequencies, counts, and percentages and visualized using histograms. Cross‐tabulations described the frequency of prodromal symptoms during each reported headache status.
Model assumptions were evaluated prior to analyses. Distributions of variables were inspected visually, and collinearity among predictors was assessed. Model convergence and parameter estimates were reviewed for all models.
Generalized estimating equations (GEE) with a binomial distribution and logit link function evaluated the relationships between reported prodromal symptoms and same‐day headache anticipation, with odds ratios (ORs) and 95% CIs quantifying these relationships. For this analysis, as well as all GEE analyses described below, the analyses accounted for the repeated measures design, where data points were nested within participants by day. An unstructured correlation matrix was chosen to account for within‐participant correlations across repeated diary entries because headache diary intervals were not equidistant. Analyses incorporated all available observations without imputing missing values under the assumption that data were missing completely at random. Exploratory analyses of reporting patterns did not reveal a systematic pattern (i.e., increased missingness following reported headaches).
To evaluate the temporal relationship between prodromal symptoms, headache anticipation, and subsequent headache onset, lagged variables were created using the LEAD function in SPSS version 20.0 (IBM). Specifically, prodromal symptoms and headache anticipation values were shifted forward by one to six time points (representing ~6, 12, 18, 24, 30, and 36 h prior to head pain onset). Lagged time points were conceptualized as cumulative of all other time points between the lag and head pain onset (i.e., “prodromal symptoms 12 h prior to head pain onset” included symptoms reported ~6 and 12 h prior to head pain onset).
GEE models examined the relationship between headache anticipation and head pain onset. ORs and 95% CIs were reported for headache anticipation's relationship with head pain onset at each lagged time point (~6, 12, 18, 24, 30, and 36 h).
GEE models were also used to assess the relationship between individual prodromal symptoms and subsequent headache onset at each lagged time point (~6, 12, 18, 24, 30, and 36 h). Only observations for which participants did not report current head pain were included in the analysis. ORs and 95% CIs were reported for each model.
To evaluate the predictive utility of individual prodromal symptoms on headache onset, positive predictive values (PPVs) and negative predictive values (NPVs) with 95% CIs were calculated for each prodromal symptom.
To assess whether headache anticipation moderated the relationship between prodromal symptoms reported 6 h prior to headache onset and the occurrence of headache, a series of moderation analyses were conducted. Specifically, 12 separate binomial GEE models with a logit function incorporating an interaction term between the prodromal symptom and headache anticipation were conducted, one for each of the 11 individual prodromal symptoms and one for the presence of any prodromal symptom reported approximately 6 h prior to headache onset. In each model, the prodromal symptom served as the predictor, headache onset served as the outcome, and headache anticipation was included as the moderator. Significant interactions indicated a significant moderation.
RESULTS
Participant characteristics
Of the 4380 individuals identified with migraine, 4166 were excluded either due to non‐responsiveness (n = 2956) or prescreening (n = 1424). Of those, 214 individuals were assessed for eligibility; 157 individuals were excluded for not meeting inclusion criteria (n = 97), declining to participate (n = 15), software issues (n = 12), or non‐responsiveness (n = 33); 57 individuals were enrolled in the study; and 43 individuals completed the study and were included for analysis. The mean age of participants was 37.1 years (SD = 11.2). The majority of the sample was female (n = 36, 83.7%) and White, non‐Hispanic (n = 33, 76.7%). At baseline, the mean number of headache days in the last month was 6.1 (SD = 3.8) with a mean baseline pain intensity of 7.0/10.0 (SD = 1.8). The mean number of completed headache diaries per person was 244.4 (SD = 0.6). A total of 10,500 observations were recorded during the study; 2247 entries (21.4%) were missing or excluded due to incomplete symptom or headache reports.
Prodromal frequencies and associations with headache anticipation
Table 1 displays prodrome symptom frequencies. A total of 7061 observations were recorded during headache‐free periods, with at least one prodromal symptom reported in 52.7% (n = 3721) of entries. All participants reported experiencing at least one prodromal symptom during the study period. The most commonly reported prodromal symptoms during headache‐free periods were fatigue (n = 2405, 34.1%), stiff neck (n = 1158, 16.4%), and thirst (n = 1179, 16.7%). The least commonly reported symptoms were sensitivity to light (n = 276, 3.9%) and sound (n = 196, 2.8%). Every prodromal symptom was reported more frequently when participants anticipated a headache compared to when they did not. At least one prodromal symptom was reported in 87.1% (n = 937/1076) of entries with headache anticipation, compared to 47.6% (n = 2850/5985) without headache anticipation. Feeling tired/weary and difficulty thinking/concentrating showed the largest increases in endorsement during headache anticipation. Additionally, the majority of participants reported prodromal symptoms during periods when they anticipated a headache compared to when they did not, except for blurred vision, dizziness, and nausea.
TABLE 1.
Frequency and proportion of prodrome symptoms during headache‐free diary entries and number of participants reporting each symptom, in total and when anticipating and not anticipating a headache.
| Prodromal symptom | Total | By headache anticipation | |||
|---|---|---|---|---|---|
| Observations | Participants | No headache anticipated observations | Headache anticipated observations | Participants during headache anticipation | |
| n (%) | n (%) | n (%) | n (%) | n (%) | |
| n = 7061 | N = 43 | n = 5985 | n = 1076 | N = 43 | |
| Stiff neck | 1158 (16.4%) | 35 (81.4%) | 892 (14.9%) | 267 (24.8%) | 35 (81.4%) |
| Unusually sensitive to light | 276 (3.9%) | 32 (74.4%) | 65 (1.1%) | 211 (19.6%) | 31 (72.1%) |
| Unusually sensitive to sound | 196 (2.8%) | 19 (44.2%) | 50 (0.8%) | 146 (13.6%) | 19 (44.2%) |
| Feeling tired or weary | 2405 (34.1%) | 41 (95.4%) | 1722 (28.8%) | 683 (63.5%) | 39 (90.7%) |
| Difficulty thinking or concentrating | 1022 (14.5%) | 32 (74.4%) | 609 (10.2%) | 413 (38.4%) | 31 (72.1%) |
| Irritability | 591 (8.4%) | 34 (79.1%) | 310 (5.7%) | 281 (16.1%) | 33 (76.7%) |
| Yawning | 389 (5.5%) | 33 (76.7%) | 240 (4.0%) | 149 (13.8%) | 24 (55.8%) |
| Nausea | 217 (3.1%) | 26 (60.5%) | 129 (2.2%) | 88 (8.2%) | 18 (41.9%) |
| Thirst | 1179 (16.7%) | 33 (76.7%) | 815 (12.6%) | 364 (33.8%) | 32 (74.4%) |
| Blurred vision | 242 (3.4%) | 13 (30.2%) | 162 (2.7%) | 80 (7.4%) | 13 (30.2%) |
| Dizziness | 139 (2.0%) | 15 (34.9%) | 73 (1.2%) | 66 (6.1%) | 15 (34.9%) |
| Presence of at least 1 prodromal symptom | 3721 (52.7%) | 43 (100%) | 2850 (47.6%) | 937 (87.1%) | 39 (90.70%) |
Note: “Observations” represents the number of headache diary entries during headache‐free periods in which each prodromal symptom was reported. “Participants” indicates the number of participants who reported each symptom at least once during headache‐free periods. “No headache anticipated” and “Headache anticipated” columns reflect the number of symptom endorsements during time points when participants did or did not anticipate a headache, respectively. “Participants during headache anticipation” indicates the percentage of participants who reported each symptom during periods of headache anticipation. Percentages are calculated within columns. Percentages across rows do not sum to 100% because postdrome observations were not included. Total number of observations during headache‐free periods = 7061. Total number of observations of not anticipating a headache = 5985. Total number of observations of anticipating a headache = 1076. Total number of participants = 43.
GEEs revealed that every prodromal symptom was significantly associated with an increased odds of reporting headache anticipation at the same time point; see Table 2. Unusual sensitivity to light (OR = 18.57, 95% CI [11.65, 26.62]) and sound (OR = 20.00, 95% CI [10.39, 38.48]) were associated with the highest odds of headache anticipation.
TABLE 2.
Odds ratios examining the associations between prodromal symptoms and headache anticipation during headache‐free diary entries.
| Prodromal symptom | OR | 95% CI [LL, UL] |
|---|---|---|
| Stiff neck | 2.09 | [1.48, 2.96] |
| Unusually sensitive to light | 18.57 | [11.65, 26.62] |
| Unusually sensitive to sound | 20.00 | [10.39, 38.48] |
| Feeling tired or weary | 2.76 | [2.02, 3.77] |
| Difficulty thinking or concentrating | 3.99 | [2.99, 5.33] |
| Irritability | 4.98 | [3.75, 6.62] |
| Yawning | 2.43 | [1.46, 4.04] |
| Nausea | 3.87 | [2.36, 6.34] |
| Thirst | 2.21 | [1.47, 3.34] |
| Blurred vision | 2.67 | [1.80, 3.94] |
| Dizziness | 4.01 | [2.58, 6.25] |
| Presence of at least 1 prodromal symptom | 3.24 | [2.15, 4.88] |
Note: Bold text indicates significance such that the CI does not include 0. Values represent unadjusted ORs. Analyses accounted for the repeated measures design, where data points were nested within participants by day. An unstructured correlation matrix was chosen to account for within‐participant correlations across repeated diary entries because assessment intervals were not equidistant. Missing data were not included in the model.
Abbreviations: CI, confidence interval; LL, lower limit; OR, odds ratio; UL, upper limit.
Associations between lagged headache anticipation, prodromal symptoms, and head pain onset
GEEs revealed that headache anticipation ~6, 12, and 36 h prior to head pain onset was significantly associated with head pain onset; see Table 3.
TABLE 3.
Odds ratios examining the associations of headache onset ~6, 12, 18, 24, 30, and 36 h after headache anticipation.
| Variable | OR | 95% CI [LL, UL] |
|---|---|---|
| Headache presence 6 h after anticipation | 3.35 | [2.57, 4.36] |
| Headache presence 12 h after anticipation | 1.98 | [1.56, 2.51] |
| Headache presence 18 h after anticipation | 1.16 | [0.89, 1.51] |
| Headache presence 24 h after anticipation | 0.81 | [0.47, 1.39] |
| Headache presence 30 h after anticipation | 1.13 | [0.96, 1.35] |
| Headache presence 36 h after anticipation | 1.37 | [1.13, 1.65] |
Note: Bold text indicates significance such that the CI does not include 0. Values represent unadjusted ORs. Analyses accounted for the repeated measures design, where data points were nested within participants by day. An unstructured correlation matrix was chosen to account for within‐participant correlations across repeated diary entries because assessment intervals were not equidistant. Missing data were not included in the model.
Abbreviations: CI, confidence interval; LL, lower limit; OR, odds ratio; UL, upper limit.
GEEs evaluating individual prodromal symptoms' association with head pain onset revealed various significant patterns of associations between putative prodromal symptoms and head pain onse; see Table 4.
TABLE 4.
Odds ratios examining the associations of headache onset ~6, 12, 18, 24, 30, and 36 h after each reported prodromal symptom.
| Variable | OR | 95% CI [LL, UL] |
|---|---|---|
| Stiff neck | – | – |
| 6 h pre‐head pain | 1.13 | [0.90, 1.42] |
| 12 h pre‐head pain | 1.08 | [0.90, 1.30] |
| 18 h pre‐head pain | 0.95 | [0.78, 1.15] |
| 24 h pre‐head pain | 0.92 | [0.78, 1.07] |
| 30 h pre‐head pain | 0.91 | [0.79, 1.05] |
| 36 h pre‐head pain | 0.86 | [0.74, 1.00] |
| Unusually sensitive to light | – | – |
| 6 h pre‐head pain | 1.34 | [0.68, 1.91] |
| 12 h pre‐head pain | 1.86 | [1.48, 2.34] |
| 18 h pre‐head pain | 1.13 | [0.83, 1.55] |
| 24 h pre‐head pain | 1.05 | [0.84, 1.31] |
| 30 h pre‐head pain | 0.76 | [0.58, 0.99] |
| 36 h pre‐head pain | 1.16 | [0.99, 1.37] |
| Unusually sensitive to sound | – | – |
| 6 h pre‐head pain | 2.45 | [1.63, 3.68] |
| 12 h pre‐head pain | 1.88 | [1.40, 2.53] |
| 18 h pre‐head pain | 1.03 | [0.74, 1.44] |
| 24 h pre‐head pain | 1.27 | [1.06, 1.51] |
| 30 h pre‐head pain | 0.85 | [0.65, 1.12] |
| 36 h pre‐head pain | 1.29 | [1.73, 1.54] |
| Feeling tired or weary | – | – |
| 6 h pre‐head pain | 1.05 | [0.83, 1.32] |
| 12 h pre‐head pain | 0.96 | [0.79, 1.78] |
| 18 h pre‐head pain | 0.84 | [0.62, 1.23] |
| 24 h pre‐head pain | 0.94 | [0.79, 1.24] |
| 30 h pre‐head pain | 0.71 | [0.56, 0.90] |
| 36 h pre‐head pain | 0.90 | [0.78, 1.05] |
| Difficulty thinking or concentrating | – | – |
| 6 h pre‐head pain | 1.05 | [0.86, 1.30] |
| 12 h pre‐head pain | 1.30 | [1.13, 1.48] |
| 18 h pre‐head pain | 1.14 | [0.94, 1.38] |
| 24 h pre‐head pain | 1.06 | [0.93, 1.21] |
| 30 h pre‐head pain | 0.88 | [0.74, 1.06] |
| 36 h pre‐head pain | 0.95 | [0.84, 1.07] |
| Irritability | – | – |
| 6 h pre‐head pain | 1.00 | [0.82, 1.22] |
| 12 h pre‐head pain | 0.93 | [0.79, 1.10] |
| 18 h pre‐head pain | 0.85 | [0.70, 1.04] |
| 24 h pre‐head pain | 0.81 | [0.69, 0.96] |
| 30 h pre‐head pain | 0.74 | [0.63, 0.87] |
| 36 h pre‐head pain | 0.80 | [0.66, 0.95] |
| Yawning | – | – |
| 6 h pre‐head pain | 0.97 | [0.68, 1.37] |
| 12 h pre‐head pain | 1.70 | [0.85, 1.35] |
| 18 h pre‐head pain | 0.92 | [0.71, 1.21] |
| 24 h pre‐head pain | 1.11 | [0.96, 1.28] |
| 30 h pre‐head pain | 1.08 | [0.94, 1.24] |
| 36 h pre‐head pain | 0.92 | [0.82, 1.05] |
| Nausea | – | – |
| 6 h pre‐head pain | 1.86 | [1.28, 2.69] |
| 12 h pre‐head pain | 1.00 | [0.76, 1.32] |
| 18 h pre‐head pain | 1.28 | [0.96, 1.71] |
| 24 h pre‐head pain | 1.28 | [1.01, 1.61] |
| 30 h pre‐head pain | 0.83 | [0.63, 1.08] |
| 36 h pre‐head pain | 0.88 | [0.63, 1.23] |
| Thirst | – | – |
| 6 h pre‐head pain | 1.05 | [0.83, 1.32] |
| 12 h pre‐head pain | 1.31 | [1.23, 153] |
| 18 h pre‐head pain | 1.28 | [1.08, 1.51] |
| 24 h pre‐head pain | 1.39 | [1.21, 1.61] |
| 30 h pre‐head pain | 1.30 | [1.16, 1.46] |
| 36 h pre‐head pain | 1.24 | [1.08, 1.43] |
| Blurred vision | – | – |
| 6 h pre‐head pain | 1.29 | [1.08, 1.55] |
| 12 h pre‐head pain | 1.67 | [1.35, 2.08] |
| 18 h pre‐head pain | 1.73 | [1.42, 2.11] |
| 24 h pre‐head pain | 1.53 | [1.33, 1.77] |
| 30 h pre‐head pain | 1.33 | [1.14, 1.56] |
| 36 h pre‐head pain | 1.43 | [1.21, 1.68] |
| Dizziness | – | – |
| 6 h pre‐head pain | 1.17 | [0.76, 1.79] |
| 12 h pre‐head pain | 1.54 | [1.24, 1.90] |
| 18 h pre‐head pain | 1.38 | [1.10, 1.73] |
| 24 h pre‐head pain | 1.31 | [1.06, 1.62] |
| 30 h pre‐head pain | 1.06 | [0.89, 1.27] |
| 36 h pre‐head pain | 1.39 | [1.15, 1.67] |
| Presence of at least 1 prodromal symptom | – | – |
| 6 h pre‐head pain | 0.90 | [0.63, 1.28] |
| 12 h pre‐head pain | 1.05 | [0.89, 1.24] |
| 18 h pre‐head pain | 0.88 | [0.65, 1.19] |
| 24 h pre‐head pain | 0.98 | [0.83, 1.17] |
| 24 h pre‐head pain | 0.70 | [0.59, 0.83] |
| 36 h pre‐head pain | 0.85 | [0.74, 0.98] |
Note: Bolded font indicates significance such that the CI does not include 0. Analyses accounted for the repeated measures design, where data points were nested within participants by day. An unstructured correlation matrix was chosen to account for within‐participant correlations across repeated diary entries because assessment intervals were not equidistant. Missing data were not included in the model.
Abbreviations: CI, confidence interval; LL, lower limit; OR, odds ratio; UL, upper limit.
Unusual sensitivity to sound (phonophobia) produced the strongest relationships with head pain onset such that experiencing phonophobia more than doubled the odds of head pain onset 6 h later; phonophobia remained associated with future head pain onset at 12, 24, and 36 h (Table 4). Other cardinal (nausea and photophobia) and non‐cardinal (difficulty thinking or concentrating) migraine symptoms were associated with higher odds of head pain in at least one of time points within 24 h of symptom reporting (Table 4). Blurred vision was associated with higher odds of future head pain at all time points evaluated, up to 36 h after the symptom was reported (Table 4). Although not associated with head pain at the most proximal time points, both thirst and dizziness were associated with higher odds of future head pain at time points between 18–36 h (Table 4). Notably, a few isolated findings suggested certain putative prodromal symptoms were associated with lower odds of future head pain (e.g., photophobia and fatigue were associated with lower odds of head pain in 30 h; irritability was associated with lower odds of head pain after 24 h) (Table 4).
Positive and negative predictive values of prodromal symptoms
To evaluate the predictive utility of individual prodromal symptoms, PPVs and NPVs were calculated for symptoms reported within ~6 h prior to headache onset, see Table 5. Overall, PPVs for individual symptoms ranged from 12.1% (95% CI [0.11, 0.24]; dizziness) to 22.0% (95% CI [0.16, 0.28]; sensitivity to sound), indicating that the presence of a single symptom was a modest predictor of subsequent headache. In contrast, NPVs were consistently high across symptoms, ranging from 85.2% (95% CI [0.84, 0.86]) to 85.8% (95% CI [0.85, 0.87]), suggesting that the absence of symptoms was more reliably associated with the absence of a headache.
TABLE 5.
Predictive performance of prodromal symptoms reported ~6 h preceding headache onset on headache onset.
| Prodromal symptom | PPV (%) | 95% CI [LL, UL] | NPV (%) | 95% CI [LL, UL] |
|---|---|---|---|---|
| Stiff neck | 13.3% | [0.11, 0.15] | 85.4% | [0.85, 0.86] |
| Unusually sensitive to light | 21.0% | [0.16, 0.26] | 85.8% | [0.85, 0.87] |
| Unusually sensitive to sound | 22.0% | [0.16, 0.28] | 85.8% | [0.85, 0.87] |
| Feeling tired or weary | 13.5% | [0.12, 0.15] | 85.2% | [0.84, 0.86] |
| Difficulty thinking or concentrating | 14.1% | [0.12, 0.16] | 85.5% | [0.85, 0.86] |
| Irritability | 13.0% | [0.10, 0.16] | 85.5% | [0.85, 0.86] |
| Yawning | 15.7% | [0.12, 0.20] | 85.6% | [0.85, 0.86] |
| Nausea | 16.4% | [0.12, 0.22] | 85.6% | [0.85, 0.86] |
| Thirst | 14.2% | [0.12, 0.16] | 85.5% | [0.85, 0.86] |
| Blurred vision | 13.7% | [0.10, 0.19] | 85.5% | [0.85, 0.86] |
| Dizziness | 12.1% | [0.11, 0.24] | 85.6% | [0.85, 0.86] |
| Presence of at least 1 prodromal symptom | 13.1% | [0.12, 0.14] | 84.6% | [0.83, 0.86] |
Note: Bolded font indicates significance such that the 95% CI does not include 0.
Abbreviations: CI, confidence intervals; LL, lower limit; NPV, negative predictive value; PPV, positive predictive value; UL, upper limit.
Headache anticipation as a moderator between prodromal symptoms and headache occurrence
To evaluate whether headache anticipation moderated the relationship between prodromal symptoms reported 6 h prior to headache onset and headache occurrence, a series of moderation analyses were conducted, see Figure 1.
FIGURE 1.
Interaction of prodromal symptoms (presence of at least one prodromal symptom, feeling tired or weary, difficulty thinking or concentrating, thirst, blurred vision, unusually sensitive to light, unusually sensitive to sound, nausea, and dizziness) and headache onset moderated by headache anticipation.
Headache anticipation moderated the relationship between the presence of any premonitory symptom in the past ~6 h and headache occurrence such that the relationship between premonitory symptom presence and headache occurrence was only significant when headache when participants did not anticipate a headache (OR = 0.42, 95% CI [0.29, 0.60]). Even adjusted for the presence of any premonitory symptom and the interaction, headache anticipation remained a significant predictor of headache occurrence (OR = 6.26, 95% CI [4.03, 9.73]).
The same pattern was observed for the individual premonitory symptoms of feeling tired/weary, difficulty thinking or concentrating, thirst, and blurred vision. Headache anticipation moderated the relationship between each of these symptoms and headache anticipation such that they were each only associated with headache occurrence during times when participants did not anticipate a headache (Tired OR = 0.50, 95% CI [0.37, 0.68]; Difficulty Thinking OR = 0.37, 95% CI [0.24, 0.58]; Thirst OR = 0.42, 95% CI [0.31, 0.58]; Blurred Vision OR = 0.63, 95% CI [0.42, 0.93]). Even adjusted for each of these symptoms and the interaction, headache anticipation remained a significant predictor of headache occurrence in each moderation analysis (Headache Anticipation adjusted for Tired OR = 4.60, 95% CI [3.26, 6.47]; Headache Anticipation adjusted for Difficulty Thinking OR = 4.18, 95% CI [3.08, 5.66]; Headache Anticipation adjusted for Thirst OR = 3.95, 95% CI [3.01, 5.18]; Headache Anticipation adjusted for Blurred Vision OR = 3.36, 95% CI [2.59, 4.37]).
Additionally, the association between both light and sound sensitivity and headache occurrence was weaker when participants anticipated a headache (Light OR = 0.15, 95% CI [0.07, 0.33]; Sound OR = 0.19, 95% CI [0.12, 0.28]); when the interaction was included in the model, both sensitivity to light and sound (Light OR = 2.89, 95% CI [1.90, 4.41]; Sound OR = 4.34, 95% CI [3.27, 5.76]) and headache anticipation (Headache Anticipation adjusted for Light OR = 3.74, 95% CI [2.78, 5.03]; Headache Anticipation adjusted for Sound OR = 3.41, 95% CI [2.70, 4.32]) were still significantly associated with higher odds of headache occurrence. Similarly, the association between both nausea and dizziness and headache occurrence was weaker when participants anticipated a headache (Nausea OR = 0.33, 95% CI [0.17, 0.65]; Dizziness OR = 0.17, 95% CI [0.08, 0.35]); when the interaction was included in the model, both nausea and dizziness (Nausea OR = 2.29, 95% CI [1.30, 4.05]; Dizziness OR = 2.54, 95% CI [1.71, 3.79]) and headache anticipation (Nausea OR = 3.42, 95% CI [2.63, 4.46]; Dizziness OR = 3.47, 95% CI [2.66, 4.52]) remained significantly associated with headache occurrence.
DISCUSSION
This EMA study evaluated the frequency and predictive utility of prodromal symptoms in the hours leading up to a migraine attack in episodic migraine. Headache anticipation was most strongly associated with headache in the ~6 and 12 h prior to headache attack as well as 36 h prior. Prodromal symptoms demonstrated heterogeneous temporal patterns. Some symptoms (e.g., sensory sensitivities, nausea, cognitive difficulty) associated with head pain onset within 24 h, whereas others (e.g., thirst, blurred vision, dizziness) showed broader associations extending to later time points. Relationships between prodromal symptoms and headache onset were strongest when headache was not anticipated. These findings underscore the value of within‐person, real‐time monitoring to better understand the prodrome phase and to develop and implement interventions before the onset of head pain.
As predicted, experiencing prodromal symptoms was consistently associated with increased odds of headache anticipation. Every prodromal symptom in this study significantly increased the odds of anticipating a headache. This robust pattern suggests that the experience of prodromal symptoms plays a direct role in shaping individuals' prediction of an impending migraine. However, predictive performance was modest because positive predictive values were low (≈20%), indicating that prodrome symptoms frequently occurred without a headache following in the specified window. Accordingly, prodromal symptoms may be better conceptualized as early signals that can contribute to clinical decision‐making when considered alongside other features rather than as definitive stand‐alone indicators for initiating treatment. 19 , 20 Consistent with this interpretation, evidence supporting the effectiveness of treating during the prodrome is currently limited and medication‐specific, with the strongest data for ubrogepant, 21 and there is limited to no evidence that treating prodromal symptoms with other acute therapies (e.g., triptans, nonsteroidal anti‐inflammatory drugs) improves outcomes. Notably, predictive values were calculated using all available diary entries as independent observations to align with the time point‐level nature of electronic diary data, which may overestimate precision due to non‐independence. Thus, these estimates should be considered preliminary.
Interestingly, the relationship between prodromal symptoms and subsequent headache occurrence was strongest when participants did not anticipate a headache. Prior research has focused on identifying prodromal symptoms in general, 8 , 10 but little work has explored how individuals' conscious anticipation interacts with those symptoms. The fact that anticipating a headache weakened the predictive relationship between prodromal symptoms (e.g., light and sound sensitivity, feeling tired/weary, difficulty thinking/concentrating, nausea, blurred vision, and dizziness) and headache onset suggests a more complex interplay between these variables. It is possible that individuals who consciously anticipate a headache, perhaps in part due to experiencing prodromal symptoms, may initiate early migraine treatments (e.g., resting, taking medication, reducing stimulation), which may in turn affect the occurrence of a headache. Although this would be beneficial to participants' migraine management, it would alter the natural progression of symptoms, therefore affecting the observed associations in this study. This cannot be directly evaluated with the present data because participants were not asked whether they engaged in acute treatment after reporting headache anticipation. Alternatively, heightened anticipatory worry may lead to an increased attention to internal bodily sensations, resulting in elevated symptom reporting, which then weakens the predictive value of prodromal symptoms. This is consistent with prior research suggesting that anticipatory attention and pain‐related hypervigilance may reduce habituation and contribute to increased symptom sensitivity. 22
Notably, feeling tired/weary, difficulty thinking/concentrating, blurred vision, and thirst were associated with headache onset only in the absence of headache anticipation. This finding suggests that individuals may lack awareness of certain prodromal symptoms, especially symptoms that may be more mundane. Therefore, enhancing individuals' ability to recognize these symptoms as part of the migraine prodrome could be a valuable target for education and self‐monitoring efforts. Improving symptom recognition can aid in the implementation of acute treatments or behavioral strategies, which could potentially reduce the severity, duration, or overall burden of the migraine attack.
Although stiff neck and fatigue were among the most frequently reported prodromal symptoms, consistent with prior studies 23 , 24 , in the current study, they did not significantly increase the odds of headache occurrence. Within this study, experiencing cardinal migraine symptoms during a non‐head pain period were all associated with higher odds of future head pain (phonophobia, photophobia, and nausea). Importantly, several other neurologic and homeostatic symptoms emerged as important symptoms predictive of future head pain, including blurred vision (significant at every time point assessed from 6 to 36 h), dizziness, and thirst. This contributes to growing evidence that the presence of certain prodromal symptoms can be used as indicators to initiate migraine interventions. 10 , 19
Interestingly, fatigue did not emerge as an independent predictor of headache onset, despite prior evidence to the contrary, and in fact was associated with decreased odds of head pain at 30 h prior to head pain. 10 This discrepancy may be explained by the nonspecific nature of fatigue and its overlap with common, everyday experiences. Similarly, irritability showed a similar pattern despite being commonly reported during prodrome period. 8 , 10 , 25 One possibility is that certain symptoms, like fatigue and irritability, are less specific and more easily influenced by contextual factors such as mood, distraction, or sleep quality. The fact that anticipation moderated the effect of fatigue on headache onset suggests that it is more meaningful in the context of headache anticipation. Additionally, it is possible that the binary classification of the prodromal symptoms (i.e., yes/no response) may miss subclinical cognitive changes that fall below the threshold of conscious reporting. Notably, photophobia and phonophobia were assessed using the qualifier “unusual” sensitivity, which may have helped participants distinguish prodromal experiences from their baseline. This added specificity may have enhanced their predictive utility because individuals may have been more likely to report them only when they clearly deviated from their norm. In contrast, symptoms assessed without such qualifiers, such as fatigue and irritability, may have been reported more frequently and non‐selectively, reducing their specificity and weakening their predictive utility.
The approximate 6–24 h window prior to headache onset emerged as a significant period during which headache anticipation predicted subsequent headache occurrence. This time frame is broader than what has been reported in many prior studies, which have identified a more immediate window, typically within 6 h because this window is the most predictive for headache onset based on prodrome symptoms. 10 , 19 However, in the present study, more prodromal symptoms were specifically associated with headache onset in 24 h before headache onset. This finding supports prior research showing that prodromal symptoms intensify as headache onset approaches, making the hours immediately preceding pain the most critical for identifying early indicators. 6 , 7 , 9 Although some prodromal symptoms in this study were not significantly associated with headache onset at any of the time points assessed, which have been seen in other studies, this may be explained by sample characteristics. For instance, Schwedt et al. 19 reported stronger associations between specific prodromal symptoms and headache onset; however, they exclusively included individuals who had demonstrated an ability to accurately identify their prodrome phase. In contrast, the current study did not differentiate between participants based on their predictive accuracy.
LIMITATIONS
Several limitations should be noted: The sample consisted of individuals with episodic migraine; therefore, findings may not generalize to those with chronic migraine. Future studies should evaluate these patterns in more diverse headache types to assess whether these prodromal patterns are migraine‐specific or shared across headache types.
Additionally, it is possible that symptoms reported during headache‐free periods, particularly those that did not predict migraine onset or that were reported further from a subsequent headache, may reflect interictal symptomatology rather than prodromal features. Future analyses distinguishing between the temporal boundaries of the interictal, prodromal, and ictal phases may help clarify which symptoms function as reliable warning signs of impending headache during the prodrome versus symptoms experienced during the interictal period, which may be less closely tied to headache anticipation or headache prediction.
All data were based on self‐report, and although the ecological design minimized retrospective bias, the lack of objective physiological data limits the conclusions that can be drawn. Future research incorporating ambulatory physiological monitoring (e.g., phone screen brightness, phone volume level, and walking speed) could offer objective data during migraine prodrome as well as reduce participant burden, increase data granularity, and help identify objective correlates of prodromal symptoms. Additionally, the present study examined a predefined set of 11 commonly reported prodromal symptoms; future studies should evaluate a broader range of symptoms, including osmophobia, to more comprehensively characterize the prodrome phase.
Additionally, the EMA design placed a burden on participants who were required to complete a headache diary multiple times per day throughout the study. Although this approach increases the accuracy of symptom tracking and the identification of patterns, it can be tiring for participants to consistently record their symptoms and experiences, which may have resulted in less detailed reporting over time, especially when symptoms were mild or less disruptive. This could potentially affect the quality of the data, particularly in the later stages of the study, and may reduce the sensitivity of the analysis in detecting changes in symptom patterns as headaches approach.
Another limitation was that the small sample size limits the generalizability of the findings. Whereas the EMA design provided many time points for analysis, the relatively small number of participants may have hindered the ability to detect meaningful relationships between prodromal symptoms and headache onset. Prodromal symptoms are likely to be highly individualized, and with a limited sample size, the study may not capture the full range of symptom experiences. Furthermore, the study's sample primarily consisted of women and individuals who identified as White and non‐Hispanic, which may limit the generalizability of the findings to more diverse populations. Given these limitations, future research should aim to include larger, more diverse samples to better capture the variability of prodromal symptoms across different individuals.
CONCLUSION
This study found that photophobia and phonophobia were predictors of headache onset, particularly within the 24 h window prior to a migraine. However, predictive performance was modest, indicating that prodromal symptoms help inform clinical decision‐making alongside other factors rather than as stand‐alone indicators for initiating early intervention. Although all prodromal symptoms increased the likelihood of headache anticipation, the predictive strength of many of these symptoms on actual headache onset was attenuated when headache anticipation was also present, suggesting that conscious awareness of impending headache may influence symptom interpretation and reporting.
AUTHOR CONTRIBUTIONS
Megan Lacritz: Conceptualization; investigation; writing – original draft; methodology; visualization; formal analysis; validation; software; data curation; project administration. Richard B. Lipton: Conceptualization; writing – review and editing; supervision. Christopher Metts: Software; resources; writing – review and editing. Elizabeth K. Seng: Funding acquisition; conceptualization; writing – review and editing; validation; supervision; resources.
CONFLICT OF INTEREST STATEMENT
Megan Lacritz reports no disclosures. Richard B. Lipton has received research support from the National Institutes of Health, the US Food and Drug Administration (FDA), and the National Headache Foundation as well as grants from Abbvie (Allergan), Biohaven, ElectroCore, Eli Lilly, FDA, Lundbeck, Pfizer, and Teva. He serves on the editorial board of Neurology, senior advisor to Headache, and associate editor to Cephalalgia. He has reviewed for the National Institute on Aging and the National Institute of Neurological Disorders and Stroke, holds stock and stock options in Biohaven Holdings, CoolTech, and NuVieBio; serves as consultant, advisory board member, or has received honoraria from: Abbvie (Allergan), American Academy of Neurology, American Headache Society, Amgen, Axon, Axsome, Biohaven, Biovision, Boston Scientific, Clexio, Dr. Reddy's (Promius), Electrocore, Eli Lilly, eNeura Therapeutics, Equinox, GlaxoSmithKline, Grifols, Ipsen, Kallyope, Karuna, Lundbeck (Alder), Manistee, Pfizer, Satsuma, Scilex, ShiraTronics, Supernus, Teva, Tonix, and Trigemina. He receives royalties from Wolff's Headache and Other Head Pain, 7th and 8th editions (Oxford Press University, 2001, 2009), Wiley, and Informa. Christopher Metts is the developer of the iPhone application used in this project. Elizabeth K. Seng has consulted or served on an advisory board for GlaxoSmithKline, Abbvie, Pfizer, and Theranica and received research funding from the National Center for Complementary and Integrative Health (R01AT011005‐01A1 MPIs: Seng and Shallcross, AT011875 PI: Minen), Veteran's Health Administration (the Headache Center of Excellence Research and Evaluation Center and VA HSR&D, IRP 20–002 PI: Damush), the American Heart Association (12MRFCSD1077180 PI:Seng), and the Cystic Fibrosis Foundation Therapeutics (PIs: Georgiopoulos & Dhingra; Quittner).
REFERENCES
- 1. Stovner LJ, Hagen K, Linde M, Steiner TJ. The global prevalence of headache: an update, with analysis of the influences of methodological factors on prevalence estimates. J Headache Pain. 2022;23(1):34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Ashina M, Katsarava Z, do TP, et al. Migraine: epidemiology and systems of care. Lancet. 2021;397(10283):1485‐1495. [DOI] [PubMed] [Google Scholar]
- 3. Maniyar FH, Sprenger T, Monteith T, Schankin CJ, Goadsby PJ. The premonitory phase of migraine – what can we learn from it? Headache. 2015;55(5):609‐620. [DOI] [PubMed] [Google Scholar]
- 4. Headache Classification Committee of the International Headache Society (IHS) the International Classification of Headache Disorders. Cephalalgia. 2018;38(1):1‐211. [DOI] [PubMed] [Google Scholar]
- 5. Karsan N, Goadsby PJ. Biological insights from the premonitory symptoms of migraine. Nat Rev Neurol. 2018;14(12):699‐710. [DOI] [PubMed] [Google Scholar]
- 6. Kelman L. The premonitory symptoms (prodrome): a tertiary care study of 893 migraineurs. Headache. 2004;44(9):865‐872. [DOI] [PubMed] [Google Scholar]
- 7. Maniyar FH, Sprenger T, Monteith T, Schankin C, Goadsby PJ. Brain activations in the premonitory phase of nitroglycerin‐triggered migraine attacks. Brain. 2014;137(1):232‐241. [DOI] [PubMed] [Google Scholar]
- 8. Quintela E, Castillo J, Muñoz P, Pascual J. Premonitory and resolution symptoms in migraine: a prospective study in 100 unselected patients. Cephalalgia. 2006;26(9):1051‐1060. [DOI] [PubMed] [Google Scholar]
- 9. Houtveen JH, Sorbi MJ. Prodromal functioning of migraine patients relative to their interictal state‐an ecological momentary assessment study. PLoS One. 2013;8(8):e72827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Giffin N, Giffin NJ, Ruggiero L, et al. Premonitory symptoms in migraine: an electronic diary study. Neurology. 2003;60(6):935‐940. [DOI] [PubMed] [Google Scholar]
- 11. Giffin NJ, Lipton RB, Silberstein SD, Olesen J, Goadsby PJ. The migraine postdrome: an electronic diary study. Neurology. 2016;87(3):309‐313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Schoonman G, Evers DJ, Terwindt GM, van Dijk J, Ferrari MD. The prevalence of premonitory symptoms in migraine: a questionnaire study in 461 patients. Cephalalgia. 2006;26(10):1209‐1213. [DOI] [PubMed] [Google Scholar]
- 13. Waelkens J. Warning symptoms in migraine: characteristics and therapeutic implications. Cephalalgia. 1985;5(4):223‐228. [DOI] [PubMed] [Google Scholar]
- 14. Penzien DB, Andrasik F, Freidenberg BM, et al. Guidelines for trials of behavioral treatments for recurrent headache: American Headache Society Behavioral Clinical Trials Workgroup. Headache. 2005;45:S110‐S132. [DOI] [PubMed] [Google Scholar]
- 15. Penzien D. Structured Diagnostic Interview for Headache. University of Mississippi Headache Clinic; 1991. [Google Scholar]
- 16. First MB, Caban D. Structured clinical interview for the DSM (SCID). The SAGE Encyclopedia of Abnormal and Clinical Psychology, volume 7. SAGE Publications; 2017:3364‐3366. [Google Scholar]
- 17. Pellegrino ABW, Davis‐Martin RE, Houle TT, Turner DP, Smitherman TA. Perceived triggers of primary headache disorders: a meta‐analysis. Cephalalgia. 2018;38(6):1188‐1198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Wöber C, Wöber‐Bingöl Ç. Triggers of migraine and tension‐type headache. Handbook of Clinical Neurology. 2010;97:161‐172. 10.1016/S0072-9752(10)97012-7 [DOI] [PubMed] [Google Scholar]
- 19. Schwedt TJ, Lipton RB, Goadsby PJ, et al. Characterizing prodrome (premonitory phase) in migraine: results from the PRODROME trial screening period. Neurol Clin Pract. 2025;15(1):e200359. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Dodick DW, Goadsby PJ, Schwedt TJ, et al. Ubrogepant for the treatment of migraine attacks during the prodrome: a phase 3, multicentre, randomised, double‐blind, placebo‐controlled, crossover trial in the USA. Lancet. 2023;402(10419):2307‐2316. [DOI] [PubMed] [Google Scholar]
- 21. Goadsby PJ, Ailani J, Dodick DW, et al. Ubrogepant for the treatment of migraine prodromal symptoms: an exploratory analysis from the randomized phase 3 PRODROME trial. Nat Med. 2025;31(7):2179‐2185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Lev R, Granovsky Y, Yarnitsky D. Enhanced pain expectation in migraine: EEG‐based evidence for impaired prefrontal function. Headache. 2013;53(7):1054‐1070. [DOI] [PubMed] [Google Scholar]
- 23. Eigenbrodt AK, Christensen RH, Ashina H, et al. Premonitory symptoms in migraine: a systematic review and meta‐analysis of observational studies reporting prevalence or relative frequency. J Headache Pain. 2022;23(1):140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Schwedt TJ, Nahas SJ, Fanning KM, et al. Characterizing pre‐headache (prodrome) features of migraine attacks: results from the CaMEO study (S4.004). Neurology. 2022;98(18_supplement):1208. [Google Scholar]
- 25. Lipton RB, Stokes J, Evans CJ, et al. Characterizing the patient experience during the prodrome phase of migraine: a qualitative study of symptoms and their timing. Headache. 2025;65(8):1355‐1368. [DOI] [PMC free article] [PubMed] [Google Scholar]