Changes in Migraine-Related Outcomes and Functioning at 1 Year: Results of the Longitudinal OVERCOME (US) Study

Abstract

Introduction

Migraine is a common neurological disorder that can severely diminish quality of life. Quantifying migraine-related changes in a real-world population over time can increase understanding of the natural history of migraine and improve clinical care of individuals with migraine.

Methods

The Observational survey of the Epidemiology, treatment, and Care Of MigrainE (OVERCOME) is a web-based study that longitudinally assessed migraine-related outcomes in a demographically representative adult population with migraine in the United States. Migraine disease status was classified as “overall improved,” “overall worsened,” or “no net change” based on changes in monthly headache days (MHDs), the Migraine Disability Assessment Scale (MIDAS) score, and the Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive (MSQ-RFR) domain score over 1 year. Migraine-related characteristics and patient-reported outcomes were evaluated for each group.

Results

Among 11,634 individuals with migraine who completed the baseline and 1-year follow-up surveys, 40.4% were classified as having “overall improved,” 24.1% as having “overall worsened,” and 35.5% as having “no net change” migraine status at 1 year. The “overall improved” group reported decreases in MHDs [mean change from baseline (SD), –3.2 (5.0)] and MIDAS score [–15.4 (26.2)] and improvements in MSQ-RFR score [+ 13.7 (21.6)]. Additionally, the impact of migraine on work decreased, there was less stigma, and this group was less likely to overuse acute medications. The “overall worsened” group experienced more MHDs [+ 3.4 (5.7)], increased MIDAS score [+ 18.8 (29.0)], and worsening MSQ-RFR score [–14.1 (21.1)]. Migraine had a greater negative impact on their working lives, and they had a greater need for acute and preventive medications for migraine.

Conclusion

MHD, MIDAS score, and MSQ-RFR score together provide a robust characterization of overall change in migraine over time. This study demonstrates the profound effect of improvement or worsening of migraine symptoms on workplace productivity, migraine-related stigma, physical function, and need for treatment.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s40120-026-00946-8.

Key Summary Points

Why carry out this study?

Understanding and characterizing changes in migraine-related symptoms and patient reported outcome measures (PROMs) over time in a real-word population is an essential prelude to improving clinical care. Previous longitudinal real-world studies only used a single measure for worsening or improvement (i.e., monthly headache days), but conceptualization of disease worsening and improvement have expanded to include other constructs such as burden/impact and health-related quality of life.

This analysis focused on the 1-year longitudinal data from OVERCOME (US) to describe changes in migraine-related symptoms and PROMs in populations with migraine that improved, worsened, or showed no net change over 1 year of follow-up based on monthly headache days (MHDs), migraine-related disability, and migraine-related quality of life.

What was learned from the study?

Participants with migraine that improved over 1 year, based on improvements in MHDs, migraine-related disability, and migraine-related quality of life, also had improved work productivity, reduced migraine-related stigma, physical function, and health-related quality of life (HRQoL), but those with migraine that worsened experienced greater negative impact on their work and HRQoL, and had an increased need for acute and preventive medications for migraine.

MHDs, Migraine Disability Assessment Scale (MIDAS) score, and Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive subscale score together provide a robust picture of overall migraine disease status and can be used to obtain a good picture of overall improvement or worsening of migraine in other domains, such as physical function, migraine-related stigma, and work productivity.

This work provides important insights for clinicians who treat people with migraine and serves as a reminder of the importance of diagnosing migraine and optimizing treatment to improve functioning and well-being and to prevent further decline over time.

Introduction

Migraine is a primary headache disorder and chronic neurological disease that affects approximately 12–15% of adults in the United States (US) [1, 2]. It is also the fourth most common cause of years lived with disability worldwide and the second leading cause of disability in women aged 15–49 years [3]. The personal and economic burden of migraine is significant, in particular with regard to health-related quality of life (HRQoL) and ability to actively participate in work, school, family, and social activities [4–6]. Fluctuations over time in the number of monthly days with migraine or headache, symptom intensity, and disability are common, and influenced by treatment, life stage and aging, comorbidities, and exogenous events, among other factors [2, 7]. Quantifying the changes that occur for people with migraine and understanding related predictors and associated outcomes of worsening or improvement in migraine disease status are valuable, both for gaining a greater understanding of the natural history of migraine and for improving outcomes in clinical care.

Population-based studies aid in understanding migraine frequency, natural history, and associated burden (Table S1) [2, 8–13]. The current study focused on the 1-year longitudinal data from OVERCOME (US) to understand changes in migraine-related characteristics and patient-reported outcomes (PROs) in populations with migraine categorized as “overall improved,” “overall worsened,” or “no net change” over the 1-year follow-up period, based on monthly headache days (MHDs), migraine-related disability, and migraine-related quality of life, using operational definitions described in the Methods section.

Methods

Study Design and Data Source

The Observational survey of the Epidemiology, treatment and Care Of MigrainE (OVERCOME) is a web-based, multicohort, longitudinal study that recruited annually in a demographically representative sample of people with migraine from the US population. OVERCOME (US) aimed to assess symptomatology, diagnosis, treatment, impact, and outcomes of migraine in the same population over time. Details on participant recruitment and methodology have been published previously [14–16]. OVERCOME (US) recruited a sample representative of the US population by geographic region, age, race, and sex. The sample was recruited in three waves in 2018, 2019, and 2020, with follow-up surveys conducted at 6, 12, 18, and 24 months. Briefly, individuals from US web panels were contacted via email and invited to participate in a general health survey; once they provided voluntary informed consent to participate, passed a demographic screener, and met criteria for migraine [via the American Migraine Study (AMS)/American Migraine Prevalence and Prevention (AMPP) diagnostic questionnaire based on modified International Classification of Headache Disorders, 3rd edition criteria) [17, 18], they completed the full migraine baseline questionnaire. Participants were required to be ≥ 18 years of age, a US resident, a member of the online survey panel, able to read and write in English, have access to the internet, and provide electronic informed consent. The study was approved by the Sterling Institutional Review Board (IRB ID #6425–001). It was conducted in accordance with the Helsinki Declaration of 1964 and its later amendments, consistent with Good Pharmaco epidemiology Practices and applicable laws.

Figure 1 shows the sampling details for the OVERCOME (US) baseline cohort. The current analysis used pooled data from the three baseline surveys and examined changes in migraine-related characteristics and patient reported outcome measures (PROMs) in populations with migraine that were considered improved, worsened, or net unchanged over the allotted time, as defined by MHDs, migraine-related disability, and migraine-related quality of life at 1 year of follow-up. All included participants completed the baseline and 1-year follow-up survey; however, participants may or may not have completed the 6-month follow-up questionnaire.

Fig. 1.

Consort diagram for OVERCOME (US) 2018–2020 (all cohorts 1-year follow-up) [20]. CONSORT Consolidated Standards of Reporting Trials, ICHD-3 International Classification of Headache Disorders, 3rd edition, US United States. ^aTargeted sampling to represent the US adult population in terms of key demographic characteristics (age, sex, race, and geography) was applied. ^bConsists of disqualified before consent (n = 2) and did not consent (n = 56,559). ^cConsists of did not complete screener (n = 79,014), did not pass data check/inconsistent response (n = 1526), and did not meet inclusion criteria (n = 1374). ^dConsists of no/do not remember headache in the past 12 months (n = 152,484), did not meet migraine criteria (n = 49,878), and over quota (n = 16,254). ^eConsists of did not complete survey (n = 9589) and disqualified (n = 25)

Sociodemographic, Clinical, and Migraine-related Measures

Sociodemographic characteristics collected in OVERCOME (US) included age, race, ethnicity, sex at birth, marital status, education (high school or less, high school diploma/some college, and college degree and beyond), employment status (full-time, part-time, retired, homemaker, or other), and annual household income (< US$15,000, $15,000–49,999, $50,000–99,999, $100,000–199,999, and ≥ $200,000).

Migraine and clinical characteristics included age at diagnosis of migraine, time between first attack and diagnosis, presence of self-reported comorbidities (including cardiovascular, joint or pain, digestive, or psychiatric comorbidities), and several migraine-related PROMs including:

MHDs

The number of days an individual reported headache in a given month on average over the preceding 3 months at the time of survey.

Migraine Disability Assessment Scale (MIDAS)

Scores the number of days a patient missed or had reduced productivity at work/school/social events in the preceding 3 months [19]. This analysis examined the mean change from baseline (CFB) in MIDAS score between baseline (sum of the preceding 3 months) and 1 year (sum of months 9–12). This scoring algorithm has been previously defined [20].

Migraine-Specific Quality-of-Life Questionnaire (MSQ) v2.1

The MSQ v2.1 consists of three dimensions quantifying the degree to which normal activities are limited by migraine (the RFR domain), the degree to which performance of normal activities is interrupted by migraine [the Role Function-Preventive (RFP) domain], and the emotional effects of migraine [the Emotional Function (EF) domain] [21]. Scores from the MSQ-RFR were included in the classification of migraine change over time analysis. The mean CFB at 1 year in each of the MSQ dimensions was also assessed.

Other PROs assessed are detailed in Table 1. For information on how these variables were used in our analysis, see Table S2.

Table 1.

Other patient-reported outcomes included in the analysis

Measure	Description	Categories/Scale
4-item Migraine Interictal Burden Scale (MIBS-4) [36]	A 4-item questionnaire measuring the burden of migraine between attacks over the previous 4 weeks	0 = None, 1−2 = Mild, 3−4 = Moderate, and ≥ 5 = Severe
Migraine Treatment Optimization Questionnaire (mTOQ) [46]	A self-administered, validated questionnaire to assess an individual’s perceived efficacy of current acute migraine treatment	0 = Very Poor, 1−5 = Poor, 6−7 = Moderate, 8 = Maximum
Migraine-Related Stigma (MiRS) questionnaire [14, 47, 48]	A validated questionnaire to measure how frequently an individual with migraine perceives stigma because of their condition. Consists of 2 subscales: (1) Secondary Gain (SG): how much the respondent with migraine feels that others perceive him/her as using migraine and associated disability to achieve secondary gains, and (2) Minimizing Burden (MB): how much the respondent with migraine believes/perceives that others minimize or underestimate the amount of pain, distress, disability, and burden that the respondent experiences due to migraine	Never, Rarely or Sometimes, Often/Very Often (MiRS-SG only), Often/Very Often (MiRS-MB only), Often/Very Often (both MiRS-MB and SG)
4-item Patient Health Questionnaire for Depression and Anxiety (PHQ-4)[39]	A screening scale for anxiety and depression with cut scores for clinically significant anxiety and depressive symptomology. Assesses the frequency with which key symptoms of depression and anxiety were experienced during the previous 2 weeks	Total score range 0−12. Higher scores represent higher depression and anxiety
12-item Allodynia Symptom Checklist (ASC-12)[34]	A 12-item questionnaire used to assess the presence and severity of ictal cutaneous allodynia symptoms during migraine attacks	0−2 = None, 3−5 = Mild, 6−8 = Moderate, ≥ 9 = Severe
Work Productivity and Activity Impairment (WPAI) questionnaire [36]	A 6-item instrument that measures impairments in paid and unpaid work over the previous 7 days. Based on responses to the 6 questions, scores for absenteeism (percent of work time missed), presenteeism (percent impairment while working), overall work impairment, and non–work-related activity impairment (both expressed as percentages) are calculated	Higher numbers indicate greater impairment and less productivity
Questionnaire to understand impact on an individual’s life due to severe headache	Respondents were asked to select 1 out of 4 responses to the following question: “Which best describes how your most severe type of headache usually affects you?”	(1) I work or function normally; (2) working ability or activity impairment to some degree; (3) working ability or activity severely impaired, but no bed rest required; or (4) bed rest required

Classification of Migraine Change Over Time

Migraine status was classified as “overall improved,” “overall worsened,”’ or “no net change” after 1 year based on changes in three PROMs: MHDs, MIDAS score, and MSQ-RFR score. Specific cutoffs are described in Table 2; these have been used previously to explore factors associated with migraine progression in these three domains [20, 22].

Table 2.

Criteria for classifying change in MHD, MIDAS score, and MSQ-RFR score after 1 year

Classification	Definition
MHD improvement	A decrease of ≥ 50% in MHDs at the 1-year follow-up compared to baseline
MIDAS score improvement	A decrease of ≥ 5 points at the 1-year follow-up compared to baseline when MIDAS baseline score is 11–20, OR a decrease of ≥ 30% at the 1-year follow-up compared to baseline when MIDAS baseline score is > 20
MSQ-RFR score improvement	An increase of ≥ 25.71 points at the 1-year follow-up compared to baseline
MHD worsening	An increase of ≥ 5 MHDs at the 1-year follow-up compared to baseline
MIDAS score worsening	An increase of ≥ 5 points at the 1-year follow-up compared to baseline
MSQ-RFR score worsening	A decrease of ≥ 25.71 points at the 1-year follow-up compared to baseline

MHD monthly headache day, MIDAS Migraine Disability Assessment Scale, MSQ-RFR Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive

For MHDs, improvement was defined as a decrease of ≥ 50% in MHDs at the 1-year follow-up compared to baseline. Worsening of MHDs was defined as an increase of ≥ 5 MHDs at the 1-year follow-up compared to baseline. Migraine status that did not fall into either category was classified as “no net change” based on MHDs. The rationale behind these cut offs is that an increase of approximately 5 MHDs can significantly affect disability, comorbidity, and HRQoL in people with migraine [23, 24]. Further, the consensus statement from the American Headache Society (AHS) suggests preventive treatment should be considered for patients with ≥ 5 MHDs and no disability. Lastly, a change of approximately 5 MHDs during a trial period has been reported in multiple phase 3 clinical trials of preventive medications for migraine [25–29].

For MIDAS score, improvement was defined as a decrease of ≥ 5 points at the 1-year follow-up compared to baseline when MIDAS baseline score was between 11 and 20, or a decrease of ≥ 30% at the 1-year follow-up when baseline score was > 20. MIDAS score worsening was defined as an increase of ≥ 5 points at the 1-year follow-up compared to baseline. If scores did not fall into either category, migraine status was classified as “no net change” based on MIDAS. Previous research has indicated that these cut offs for improvement and worsening in MIDAS score are clinically meaningful for patients [30].

For MSQ-RFR score, improvement was defined as an increase of ≥ 25.71 points in MSQ-RFR score, while worsening was defined as a decrease of ≥ 25.71 points. If scores did not fall into either category, migraine status was classified as “no net change” based on MSQ-RFR. Previous research suggests that these cut offs are clinically meaningful for patients [31].

Criteria for the classification of migraine status as “overall improved,” “overall worsened,” or “no net change” at 1 year were based on grouping of all permutations of potential outcomes for change in MHDs, MIDAS score, and MSQ-RFR score into the three descriptive groups of “overall improved,” “overall worsened,” or “no net change” (Fig. 2). Of note, in some cases, there may have been improvement in some PROMs and worsening in others which was classified as “no net change”. Briefly, participants were assigned + 1 point if they met criteria for improvement in each of the domains of MHDs, MIDAS or MSQ-RFR, – 1 point if they met criteria for worsening, and 0 points if they met neither improvement nor worsening criteria. Participants with net positive points across all three domains were classified as having “overall improved,” those with net negative points as having “overall worsened,” and those with net neutral or 0 points as having “no net change” migraine status.

Fig. 2 .

Classification criteria for overall migraine status at 1 year based on changes in MHDs, MIDAS score, and MSQ-RFR score. Participants were assigned + 1 point if they met criteria for improvement (i.e., decrease of ≥ 50% in MHDs, decrease of ≥ 5 MIDAS points when MIDAS baseline score was 11–20 or a decrease of ≥ 30% when MIDAS baseline score was > 20, or an increase of ≥ 25.71 points in MSQ-RFR score over the 1-year follow-up) in each of the domains of MHDs, MIDAS score, or MSQ-RFR score (indicated by the green box), – 1 point if they met criteria for worsening (i.e., increase of ≥ 5 MHDs, increase of ≥ 5 MIDAS points, or decrease of ≥ 25.71 points on the MSQ-RFR over 1 year; indicated by the red box), or 0 points if they met neither improvement nor worsening criteria (indicated by the yellow box). Those with net positive points across all three domains were classified as having “overall improved,” those with net negative points as having “overall worsened,” and those with net neutral points as having “no net change” migraine status. MHD monthly headache day, MIDAS Migraine Disability Assessment, MSQ-RFR Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive

Need For Treatment and Medication Use

Respondents were stratified by their need for initiation/modification of prescription treatment according to the categories: No medication need, Acute medication need, Preventive medication need, or Preventive and Acute medication need. Classification was based on the AHS 2021 Consensus Statement recommendations using MHDs, MIDAS score, and Migraine Treatment Optimization Questionnaire-4 score [22, 29].

Information on medication use was collected via a series of questions regarding respondents’ use of acute [including previous and current prescription and over-the-counter (OTC)] and preventive migraine medication. Recommended acute and preventive medications for migraine were based on the AHS Consensus Statement [29].

For each acute medication that respondents indicated they were currently taking, they were asked to report for how many days in the previous 30 days they had taken this medication. They were considered to be overusing a certain medication if intake was ≥ 10 days for prescription medications and OTC combination migraine or headache formula medications, or ≥ 15 days for simple analgesics such as acetaminophen, aspirin, ibuprofen, or naproxen sodium [18, 32].

Statistical Analysis

Sociodemographic, clinical, and migraine-related characteristics were summarized using means [± standard deviation (SD)] for continuous variables and percentages for ordinal or categorical variables. Standardized mean differences (SMDs) were calculated to compare characteristics between groups. SMD values < 0.20 were considered to indicate no difference, 0.20–0.49 a small difference, 0.50–0.79 a moderate difference, and > 0.80 a large difference [33].

To understand factors associated with not completing the 1-year follow-up survey and to test whether non-completion was random or due to a systematic reason, supervised machine learning (random forest with 1000 individual trees) was used to identify factors that were predictive of not completing the 1-year survey from among a list of 59 sociodemographic, clinical, and migraine-related candidate factors (Table S2). Continuous variables were standardized for the analysis. Machine learning algorithms are efficient prediction tools, as they excel at handling large complex, nonlinear datasets, ensuring all possible interactions are accounted for. The random forest performance was evaluated by randomly splitting the data into training (60%) and test (40%) sets. The random forest was trained on the training dataset and computed area under the curve on the test dataset. Random forest modeling has been used in previous OVERCOME (US) analyses to understand factors predictive of a certain outcome [16]. Impurity reduction was measured by the Gini Index, using out-of-bag (OOB) observations to estimate prediction accuracy and variable importance. A logistic regression was run using the top 10 variables identified by the random forest algorithm to understand the directionality of the association. Odds ratios and 95% confidence intervals are reported for each variable in the regression. Variance inflation factors were calculated and evaluated to determine the degree of multicollinearity between variables in the logistic regression models. Changes in continuous variables at 1 year were reported as absolute change or mean CFB (SD) as applicable. Changes in categorical variables were reported as absolute change in the proportion of participants in each category. For all measured variables, SMDs were calculated between baseline and 1 year and between the mean CFB values for the “overall improved” and “overall worsened” groups. SAS Enterprise Guide Version 7.15 (SAS Institute, Cary, NC, USA) was used for all analyses.

Results

Patient Demographics and Characteristics

Among the 11,634 participants who completed the baseline and 1-year follow-up surveys, 4703 (40.4%) were classified as having “overall improved,” 2804 (24.1%) as having “overall worsened,” and 4127 (35.5%) as having “no net change” migraine status at 1 year using the algorithm compilation of MHDs, MIDAS score, and MSQ-RFR score. Demographics were consistent across all three groups (Table 3).

Table 3.

Patient demographics and characteristics grouped by overall change in migraine status at 1 year

Characteristic	“Overall improved” N = 4703	“No net change” N = 4127	“Overall worsened” N = 2804	All participants N = 11,634
Age at baseline, mean (SD)	47.7 (14.2)	49.6 (14.2)	47.0 (13.7)	48.2 (14.1)
Years with migraine, mean (SD)	22.1 (15.9)	23.9 (16.9)	22.4 (15.6)	22.8 (16.2)
Sex at birth, n (%)
Female	3521 (74.9)	3101 (75.1)	2171 (77.4)	8793 (75.6)
Male	1182 (25.1)	1026 (24.9)	633 (22.6)	2841 (24.4)
Race, n (%)
White	3929 (83.5)	3496 (84.7)	2389 (85.2)	9814 (84.4)
Black	327 (7.0)	257 (6.2)	173 (6.2)	757 (6.5)
Asian	134 (2.8)	139 (3.4)	72 (2.6)	345 (3.0)
Other	294 (6.3)	218 (5.3)	158 (5.6)	670 (5.8)
Prefer not to answer	19 (0.4)	17 (0.4)	12 (0.4)	48 (0.4)
Region of residence, n (%)
Midwest	1084 (23.0)	992 (24.0)	625 (22.3)	2701 (23.2)
Northeast	898 (19.1)	829 (20.1)	544 (19.4)	2271 (19.5)
West	910 (19.3)	858 (20.8)	558 (19.9)	2326 (20.0)
South	1811 (38.5)	1448 (35.1)	1077 (38.4)	4336 (37.3)
Ethnicity, n (%)
Non-Hispanic	4193 (91.3)	3748 (92.9)	2541 (92.5)	10,482 (90.1)
Hispanic	398 (8.7)	286 (7.1)	205 (7.5)	889 (7.6)
Urban yes, n (%)	3887 (82.6)	3408 (82.6)	2279 (81.3)	9574 (82.3)
Married or living with partner, n (%)	2774 (59.0)	2510 (60.8)	1605 (57.2)	6889 (59.2)
Household income, n (%)
Less than $15,000	448 (9.5)	312 (7.6)	291 (10.4)	1051 (9.0)
$15,000—49,999	1732 (36.8)	1403 (34.0)	1058 (37.7)	4193 (36.0)
$50,000—99,999	1540 (32.7)	1420 (34.4)	935 (33.3)	3895 (33.5)
$100,000—199,999	731 (15.5)	735 (17.8)	391 (13.9)	1857 (16.0)
$200,000 or over	122 (2.6)	121 (2.9)	50 (1.8)	293 (2.5)
Prefer not to answer	130 (2.8)	136 (3.3)	79 (2.8)	345 (3.0)
Employment status, n (%)
Full-time employment	1976 (42.0)	1798 (43.6)	1124 (40.1)	4898 (42.1)
Part-time employment	575 (12.2)	509 (12.3)	359 (12.8)	1443 (12.4)
Retired	501 (10.7)	420 (10.2)	336 (12.0)	1257 (10.8)
Homemaker	774 (16.5)	765 (18.5)	390 (13.9)	1929 (16.6)
Other	850 (18.1)	618 (15.0)	566 (20.2)	2034 (17.5)
Prefer not to answer	27 (0.6)	17 (0.4)	29 (1.0)	73 (0.6)
Education, n (%)
Highschool or less	1048 (22.3)	855 (20.7)	631 (22.5)	2534 (21.8)
Some college	1826 (38.8)	1514 (36.7)	1146 (40.9)	4486 (38.6)
College degree or beyond	1810 (38.5)	1746 (42.3)	1016 (36.2)	4572 (39.3)
Prefer not to answer	19 (0.4)	12 (0.3)	11 (0.4)	42 (0.4)
Health Insurance yes, n (%)	4157 (88.4)	3729 (90.4)	2478 (88.4)	10,364 (89.1)
Tobacco use yes, n (%)	1320 (28.1)	929 (22.5)	822 (29.3)	3071 (26.4)
Marijuana use yes, n (%)	1076 (22.9)	662 (16.0)	623 (22.2)	2361 (20.3)
Migraine diagnosis class, n (%)
Migraine diagnosis only	906 (19.3)	834 (20.2)	528 (18.8)	2268 (19.5)
Other headache diagnosis only	1089 (23.2)	915 (22.2)	597 (21.3)	2601 (22.4)
No migraine or other headaches	960 (20.4)	1060 (25.7)	530 (18.9)	2550 (21.9)
Migraine plus other headache diagnosis	1748 (37.2)	1318 (31.9)	1149 (41.0)	4215 (36.2)
Cardiovascular comorbiditiesa, n (%)
Number of cardiovascular comorbidities
0	2222 (47.2)	2014 (48.8)	1349 (48.1)	5585 (48.0)
1	1166 (24.8)	1006 (24.4)	683 (24.4)	2855 (24.5)
≥ 2	1315 (28.0)	1107 (26.8)	772 (27.5)	3194 (27.5)
Triptan contraindication	789 (16.8)	584 (14.2)	460 (16.4)	1833 (15.8)
Noncardiovascular comorbidities, n (%)
Number of joint or pain comorbiditiesb
0	2965 (63.0)	2806 (68.0)	1645 (58.7)	7416 (63.7)
1	1030 (21.9)	891 (21.6)	699 (24.9)	2620 (22.5)
≥ 2	708 (15.1)	430 (10.4)	460 (16.4)	1598 (13.7)
Number of digestive comorbiditiesc
0	3011 (64.0)	2768 (67.1)	1740 (62.1)	7519 (64.6)
≥ 1	1692 (36.0)	1359 (32.9)	1064 (37.9)	4115 (35.4)
Number of psychiatric comorbiditiesd
0	2291 (48.7)	2408 (58.3)	1255 (44.8)	5954 (51.2)
1	882 (18.8)	685 (16.6)	508 (18.1)	2075 (17.8)
2	881 (18.7)	629 (15.2)	596 (21.3)	2106 (18.1)
≥ 3	649 (13.8)	405 (9.8)	445 (15.9)	1499 (12.9)
Allergies	1815 (38.6)	1574 (38.1)	1132 (40.4)	4521 (38.9)
Asthma	865 (18.4)	668 (16.2)	564 (20.1)	2097 (18.0)
Sleep apnea	569 (12.1)	437 (10.6)	350 (12.5)	1356 (11.7)

SD standard deviation

^aIncluded self-reported medical diagnosis of prediabetes, diabetes, high cholesterol, and hypertension

^bIncluded self-reported chronic back pain, fibromyalgia, osteoarthritis, and rheumatoid arthritis

^cIncluded self-reported acid reflux/gastroesophageal reflux disease, inflammatory bowel disease, Crohn’s disease, or ulcerative colitis

^dIncluded self-reported anxiety, depression, or panic disorder

A total of 48.2% of participants met improvement criteria (Table 1) for at least one of the specified assessments (MHD, MIDAS, or MSQ-RFR), 14.9% for at least 2 assessments, and 2.8% for all 3 assessments. A total of 31.8% of participants met the criteria for worsening for at least one assessment, 8.4% for at least 2 assessments, and 1% for all 3 assessments (Table 4).

Table 4.

Participants meeting the improvement or worsening criteria in 1, 2, or all 3 of the domains of MHDs, MIDAS score, and MSQ-RFR score at 1 year

Classification	Population, n (%) N = 11,634
Improved over time in ≥ 1 variable
Met MHD improvement criteria	3712 (31.9)
Met MIDAS score improvement criteria	2418 (20.8)
Met MSQ-RFR score improvement criteria	1534 (13.2)
Met ≥ 1 of the above 3 improvement criteria	5607 (48.2)
Met ≥ 2 of the above 3 improvement criteria	1728 (14.9)
Met all 3 of the above improvement criteria	329 (2.8)
Worsened over time in ≥ 1 variable
Met MHD worsening criteria	1087 (9.3)
Met MIDAS score worsening criteria	2519 (21.7)
Met MSQ-RFR score worsening criteria	1186 (10.2)
Met ≥ 1 of the above 3 worsening criteria	3703 (31.8)
Met ≥ 2 of the above 3 worsening criteria	979 (8.4)
Met all 3 of the above worsening criteria	110 (1.0)

MHD monthly headache day, MIDAS Migraine Disability Assessment Scale, MSQ-RFR Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive

Baseline Factors Predicting Participant Non-completion of the 1-Year Follow-Up Survey

Among participants who completed the baseline survey, 19.7% (11,634/59,001) completed the 1-year follow-up survey (Fig. 1). The random forest analysis incorporating 59 sociodemographic, clinical, and migraine-related factors (Table S2) found that younger age was the most important factor predictive of non-completion of the 1-year survey (OOB Gini 0.00651) (Fig. 3).

Fig. 3.

Random forest and logistic regression analysis of factors associated with non-completion of the 1-year follow-up survey. First, a random forest algorithm was used to identify factors that were predictive of not completing the 1-year survey from among a list of 59 sociodemographic, clinical, and migraine-related candidate factors. The OOB Gini index is commonly used as the splitting criterion in classification trees. Second, a logistic regression was run using the top 10 variables identified by the random forest algorithm to understand the directionality of the association. Forest plot shows ORs with 95% CIs. For ORs with no visible CI, the CI was too small to be plotted on the graph. ASC-12 12-item Allodynia Symptom Checklist, CI confidence interval, MIDAS Migraine Disability Assessment Scale, MiRS Migraine-Related Stigma questionnaire, OOB out-of-bag, OR odds ratio, PHQ-4 4-item Patient Health Questionnaire for Depression and Anxiety

Changes in MHDs, MIDAS Score, and MSQ-RFR Score Over 1 Year of Follow-Up

We examined mean changes in MHDs, MIDAS score, and MSQ-RFR score in the three outcome groups. As these variables were used to define the outcome groups, changes in these measures are inherent to the grouping and therefore expected (see Discussion). Participants in the “overall improved” group experienced fewer MHDs at the 1-year follow-up [mean CFB (SD), –3.2 (5.0)], while those in the “overall worsened” group experienced more MHDs [mean CFB (SD), + 3.4 (5.7); SMD = 1.23] (Fig. 4a). In the “overall improved” group, the proportion of participants with 0–3 MHDs increased by 24.7% (50.9% at baseline vs. 75.6% at 1 year, SMD = 0.53) (Fig. 4b). Conversely, in the “overall worsened” group, the proportion of participants experiencing more MHDs increased by 9.7% for 8–14 MHDs (10.2% at baseline vs. 19.9% at 1 year, SMD = 0.27) and by 11.6% for 15 + MHDs (10.6% at baseline vs. 22.2% at 1 year, SMD = 0.32) (Fig. 4b).

Fig. 4.

CFB in MHDs, MIDAS score, and MSQ-RFR score at 1 year by “overall improved,” “overall worsened,” or “no net change” migraine status. a Mean CFB (SD) in the number of MHDs. b Change in the proportion of participants in each MHD frequency category. c Mean CFB (SD) in MIDAS score. d Change in the proportion of participants at each MIDAS disability level. e Mean CFB (SD) in MSQ-RFR score. CFB change from baseline, MHD monthly headache day, MIDAS Migraine Disability Assessment Scale, MSQ-RFR Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive, SD standard deviation, SMD standardized mean difference. # SMD between 0.2 and < 0.5; ## SMD between 0.5 and < 0.8

MIDAS score decreased in participants in the “overall improved” group [mean CFB (SD), –15.4 (26.2)] and increased in the “overall worsened” group [mean CFB (SD), + 18.8 (29.0); SMD = 1.24] over the 1-year follow-up (Fig. 4c). In the “overall improved” group, the proportion of participants with little or no disability increased by 28.6% (34.6% at baseline vs. 63.2% at 1 year, SMD = 0.60), while, in the “overall worsened” group, the proportion with severe disability increased by 27.9% (17.8% at baseline vs. 45.7% at 1 year, SMD = 0.63) (Fig. 4d).

MSQ-RFR score improved in participants in the “overall improved” group [mean CFB (SD), + 13.7 (21.6); 51.5 (24.1) at baseline vs. 65.2 (25.4) at 1 year, SMD = 0.55] and worsened in the “overall worsened” group [mean CFB (SD), –14.1 (21.1); 57.3 (23.4) at baseline vs. 43.3 (21.6) at 1 year, SMD = 0.62]. The SMD between the groups was 1.3 (Fig. 4e).

MHDs, MIDAS score, and MSQ-RFR score were not significantly different at 1 year compared to those at baseline in the “no net change” group (SMD < 0.2 in all instances).

Changes in Migraine-Related PROMs Over 1 Year of Follow-Up

This section details PROMs not used to define the groups. MSQ-RFP score increased significantly on average in the “overall improved” group at the 1-year follow-up. Mean CFB (SD) in this group was + 11.0 (22.8) [63.6 (26.7) at baseline vs. 74.7 (25.7) at 1 year, SMD = 0.42]. MSQ-RFP score decreased in the “overall worsened” group, with a mean CFB (SD) of –10.3 (23.2) [67.2 (25.2) at baseline vs. 56.9 (26.1) at 1 year, SMD = 0.40]. The mean CFB differed between the “overall improved” and “overall worsened” groups (SMD = 0.93) (Fig. 5a). Mean CFB in MSQ-EF score also differed between the “overall improved” and “overall worsened” groups (SMD = 0.89). MSQ-EF score increased by 11.7 (25.1) on average in the “overall improved” group [60.2 (29.6) at baseline vs. 71.9 (28.2) at 1 year, SMD = 0.40] and decreased by 10.7 (25.3) on average in the “overall worsened” group [63.5 (28.5) at baseline vs. 52.8 (28.3) at 1 year, SMD = 0.38] (Fig. 5b). There was a mean reduction from baseline in headache pain severity of 0.9 (2.2) in the “overall improved” group [7.1 (1.8) at baseline vs. 6.2 (2.4])at 1 year, SMD = 0.42] but no change in the “overall worsened” group. The SMD between groups was 0.58 (Fig. 5c). There were no significant within-group differences in allodynia at 1 year, as measured by the 12-item Allodynia Symptom Checklist [34], but the SMD between the “overall improved” and “overall worsened” group was 0.36 (Fig. 5d). The interictal burden, as measured by the 4-item Migraine Interictal Burden Scale [35], increased in the “overall worsened” group; in particular, the proportion of individuals with no interictal burden decreased by 9.4% (30.4% at baseline vs. 21.0% at 1 year, SMD = 0.22), and the proportion of individuals with severe interictal burden increased by 12.1% (34.3% at baseline vs. 46.4% at 1 year; SMD = 0.25) (Fig. 5e).

Fig. 5.

CFB to 1 year in migraine-related PROMs by migraine change status. a Mean CFB (SD) in MSQ-RFP score. b Mean CFB (SD) in MSQ-EF score. c Mean CFB (SD) in pain severity during migraine. d Change in the proportion of participants at each level of allodynia symptom severity (as measured by the ASC-12). e Change in the proportion of participants at each level of interictal burden severity (as measured by the MIBS-4). ASC-12 12-item Allodynia Symptom Checklist, CFB change from baseline, MIBS-4 4-item Migraine Interictal Burden Scale, MSQ-EF Migraine-Specific Quality-of-Life Questionnaire–Emotional Function, MSQ-RFR Migraine-Specific Quality-of-Life Questionnaire–Role Function-Restrictive, PROM patient-reported outcome measure, SD standard deviation, SMD standardized mean difference. # SMD between 0.2 and < 0.5

Changes in Treatment Need, Medication Use and Overuse, and Treatment Response Over 1 Year of Follow-Up

Among participants in the “overall improved” group, there was a decreased need for medication at 1 year, with the proportion of individuals with no medication needs increasing by 26.6% (19.5% at baseline vs. 46.1% at 1 year, SMD = 0.59) and the proportion of individuals with a need for acute and preventive medication decreasing by 24.6% (42.3% at baseline vs. 17.7% at 1 year, SMD = 0.56) (Fig. 6a). Conversely, participants in the “overall worsened” group showed higher medication needs at 1 year, with an 18.0% decrease in the proportion of individuals with no medication needs (24.8% at baseline vs. 6.8% at 1 year, SMD = 0.51) and a 26.7% increase in the proportion with a need for acute and preventive medication (32.2% at baseline vs. 58.9% at 1 year, SMD = 0.56). Interestingly, use of recommended acute medications decreased in all groups irrespective of migraine change status (SMD > 0.2), and there was no notable change in the use of recommended preventive medications for any group (SMD < 0.2) (Fig. 6b).

Fig. 6.

CFB to 1 year in medication candidacy, use, and efficacy by migraine change status. Graphs show the proportion of participants with “overall improved,” “overall worsened,” and “no net change” migraine status across a treatment candidacy, b medication use, c medication overuse, and d acute medication efficacy. mTOQ Migraine Treatment Optimization Questionnaire, NSAID nonsteroidal anti-inflammatory drug, OTC over-the-counter, SMD standardized mean difference. # SMD between 0.2 and < 0.5; ## SMD between 0.5 and < 0.8

Overuse of OTC medications and nonsteroidal anti-inflammatory drugs (NSAIDs) decreased by 8.3% and 6.0%, respectively, in the “overall improved” group at the 1-year follow-up (OTC: 13.5% at baseline vs. 5.2% at 1 year, SMD = 0.29; NSAIDs: 8.7% at baseline vs. 2.7% at 1 year, SMD = 0.26). No significant difference in OTC or NSAID overuse was observed in the “overall worsened” group at the 1-year follow-up (SMD < 0.2). Opioid overuse also remained unchanged in all groups (SMD < 0.2), with a general trend toward decrease across all groups (Fig. 6c). There were no significant changes in acute medication efficacy at the 1-year follow-up in any of the 3 groups (SMD < 0.2) (Fig. 6d).

Changes in Impact on Functioning, Migraine-related Stigma, Psychiatric Symptomology, and Work Productivity

Participants in the “overall improved” group reported a reduction in the impact of migraine on their daily functioning, with a 6.8% increase in the proportion with normal function (9.7% at baseline vs. 16.5% at 1 year, SMD = 0.2) (Fig. 7a). This group also experienced overall improvement in Work Productivity and Activity Impairment (WPAI) scores [36]: daily activity loss decreased by 8.4% (43.6% at baseline vs. 35.2% at 1 year, SMD = 0.28), presenteeism decreased by 6.7% (41.1% at baseline vs. 34.4% at 1 year, SMD = 0.23), and productivity loss decreased by 6.6% (44.9% at baseline vs. 38.3% at 1 year, SMD = 0.21) (Fig. 7b). Participants in the “overall worsened” group did not experience significant change in their daily functioning (SMD < 0.2 in all instances) but did see increased impairment in their work. Daily activity loss increased by 9.2% (41.6% at baseline vs. 50.8% at 1 year, SMD = 0.31), presenteeism increased by 7.9% (38.1% at baseline vs. 46.0% at 1 year, SMD = 0.27), and productivity loss increased by 8.0% (41.9% at baseline vs. 49.9% at 1 year, SMD = 0.25).

Fig. 7.

CFB to 1 year in stigma and mental health PROs by migraine change status. a Change in the proportion of participants in each daily functioning category. b Change in mean work and activity impairment (as measured by the WPAI). c Change in the proportion of participants experiencing each frequency category of migraine-related stigma (as measured by the MiRS questionnaire). d Change in the proportion of participants hiding migraine. e Mean CFB (SD) in the frequency of depression and anxiety symptoms (as measured by the PHQ-4, total score range 0–12). CFB change from baseline, MiRS Migraine-Related Stigma, MiRS-MB Migraine-Related Stigma–Minimizing Burden subscale, MiRS-SG Migraine-Related Stigma–Secondary Gain subscale, PHQ-4 4-item Patient Health Questionnaire for Depression and Anxiety, PRO patient-reported outcome, SD standard deviation SMD standardized mean difference, WPAI Work Productivity and Activity Impairment questionnaire. # SMD between 0.2 and < 0.5

Participants in the “overall improved” group experienced less migraine-related stigma, with an 11.8% increase in the proportion of individuals reporting on the Migraine-Related Stigma questionnaire [14, 37, 38] that they never experience stigma (13.5% at baseline vs. 25.3% at 1 year, SMD = 0.30) (Fig. 7c). Participants in this group were also less likely to hide their migraine, with an 11.5% decrease in the proportion of participants hiding their migraine at the 1-year follow-up (69.2% at baseline vs. 57.7% at 1 year, SMD = 0.24) (Fig. 7d). While clinically significant depressive and anxious symptomology, as measured by the 4-item Patient Health Questionnaire for Depression and Anxiety (PHQ-4) [39], was not different between baseline and the 1-year follow-up within the “overall improved” and “overall worsened” groups (SMD < 0.20), there was a difference in mean CFB between the “overall improved” and “overall worsened” groups (SMD = 0.39) (Fig. 7e).

Discussion

The longitudinal nature of the OVERCOME (US) study allowed for the assessment of rates, patterns, and changes in migraine characteristics and PROs in individuals with migraine disease status classified as “overall improved,” “overall worsened,” or “no net change” based on changes in MHDs, MIDAS score, and MSQ-RFR score at the 1-year follow-up. Our results show that individuals with migraine that improved overall over 1 year were also more likely to experience improvement in other domains, such as other quality of life, functional impairment, and other migraine-related symptoms, and these individuals also had less need for medication. Conversely, individuals with migraine that overall worsened over the 1-year follow-up were more likely to experience decline in other measures of function and HRQoL and have more medication needs. These findings suggest that improvement in some domains predict improvement in other domains, potentially highlighting that treatment may deliver broader benefits than those typically measured.

Previous population-based studies in migraine have been both cross-sectional (AMS I and AMS II) [10, 13] and longitudinal [AMPP, the Chronic Migraine Epidemiology and Outcomes (CaMEO) study, and the Migraine in America Symptoms and Treatment (MAST) study] [8, 9, 11] in methodology, and both have advantages. While longitudinal studies prove better at establishing relationships and patterns, cross-sectional studies, which examine a moment in time, are often larger and more diverse and do not have the potential for selective attrition. However, both study designs may be associated with self-selection, recall, and non-completion biases. In OVERCOME (US), the proportion of individuals who completed the 1-year follow-up survey among those who completed the baseline survey was 19.7% (n = 11,634/59,001). Differences in survey response rates have been prominent in previous population-based studies of migraine. Notably, 1-year follow-up rates in AMPP, a mail-based survey conducted in 2004–2009, were 54% [47], whereas follow-up rates in CaMEO, a web-based survey similar to OVERCOME conducted in 2012–2013, were 35.2% [7]. Understanding the lower follow-up rates is essential, as nonrespondents may differ from respondents in meaningful ways. To understand the non-completion rates in OVERCOME, we conducted a random forest and logistic regression analysis of 59 sociodemographic, clinical, and migraine-related factors associated with non-completion of the 1-year follow-up survey. This analysis found that younger age was the most important predictive factor for non-completion of the OVERCOME (US) 1-year follow-up survey. Younger age has been identified as a predictive factor for follow-up survey nonresponse in multiple previous studies [40–42]. One reason why older individuals may be more likely to complete follow-up surveys for studies such as OVERCOME (US) and AMPP could be that they have less work and family demands (e.g., retired or homemaker vs. working part or full-time), and thus may have more time to participate in surveys. It could also be that participation in a survey gives older individuals, who may not have a busy social life, the feeling that they are able to communicate with someone about how they are impacted by their migraine. However, it is also important to consider that those who drop out or do not complete a survey may do so not only because they are too busy but also because their migraine has worsened to such an extent that they can no longer participate in an online survey such as this one or that they perceive their migraine has improved so significantly that they do not feel that a survey is relevant to them. Hence interpretation of the results should take the possibility of differential drop out into account.

Our results indicate that improvement or worsening of migraine at the 1-year follow-up was associated with significant changes in a broader range of variables (emotional function, well-being, and workplace productivity) in the expected direction. The magnitude of change in the domains used to define outcome status at follow-up was substantial; participants with migraine status classified as “overall improved” reported decreases in headache frequency (mean CFB in MHDs –3.2) and migraine-related disability (mean CFB in MIDAS score –15.4) as well as improvements in HRQoL (mean CFB in MSQ-RFR score + 13.7). A reduction of more than 3 MHDs and a reduction of MIDAS score of more than 15 exceeded the clinical meaningful change thresholds [30] by a factor of 3. In the “overall improved” group, there were notable positive changes in PROMs not used to define outcome status. The negative impact of migraine on work was reduced, emotional function and well-being improved, and stigma was reduced. By contrast, individuals with migraine classified as “overall worsened” experienced more frequent headaches (mean CFB in MHDs + 3.4), increased disability (mean CFB in MIDAS score + 18.8), and decreased HRQoL (mean CFB in MSQ-RFR score –14.1). Individuals with “overall worsened” migraine also experienced higher levels of interictal burden and deterioration in emotional well-being. These results are consistent with prior findings on depression, anxiety, and stigma in migraine [37, 43, 44].

Importantly, this study also evaluated workplace productivity over time and showed that changes in migraine status over 1 year were accompanied by meaningful and directionally consistent changes in workplace productivity (as measured by the WPAI). Participants with migraine classified as “overall improved” experienced reductions in work-related impairment, including lower presenteeism and reduced overall productivity loss, along with improved daily functioning. In contrast, those with migraine classified as overall worsened” showed significant increases in work impairment, with higher presenteeism and greater productivity loss over time, indicating a growing impact of migraine on job performance. These findings demonstrate that improvement or worsening of migraine disease status translates into tangible changes in workplace functioning, underscoring that effective migraine management has important implications for productivity beyond symptom reduction alone.

Medication use and need changed for both the “overall improved” and “overall worsened” populations in this analysis. Individuals with “overall improved” migraine had decreased medication need, acute prescription medication use, and medication overuse. In contrast, those with “overall worsened” migraine were more likely to be candidates for acute and preventive treatment optimization at 1 year. These individuals were less likely to use recommended acute medications as well, which may reflect movement toward preventive treatment. However, our data show that even among those with migraine that worsened over 1 year, only 13.7% were taking recommended preventive medications- a rate that, while unfortunately low, is in line with results from other epidemiologic studies and highlights an unmet need among these individuals. While it might be expected that use of acute medications may be higher among individuals with migraine that worsened over 1 year, a decrease in the use of recommended acute medications was actually observed but with an increase in the overuse of acute OTC treatments. Prior population-based migraine studies have consistently shown that patterns of acute medication use over time do not necessarily mirror changes in migraine severity. Large cohort studies such as CaMEO and AMPP have demonstrated that use of migraine-specific acute medications, including triptans and other prescription therapies, often declines longitudinally, even among individuals with persistent or worsening migraine burden [42, 45]. At the same time, these studies report increased reliance on nonspecific OTC analgesics and NSAIDs, reflecting substitution rather than a reduced need for acute treatment. Proposed explanations include perceived loss of efficacy of prescription acute therapies, treatment fatigue, tolerability concerns, cost or access barriers, and disengagement from formal medical care. Consistent with these observations, our findings suggest that participants with migraine classified as “overall worsened” may reduce use of guideline-recommended acute migraine medications while increasingly relying on OTC agents, highlighting the importance of distinguishing medication type from overall acute treatment burden when interpreting longitudinal changes in real-world migraine populations. Participants with “no net change” in migraine status at 1 year experienced no change in their medication need or rates of acute medication overuse but used fewer recommended acute medications. Medication use is confounded by indication. People who require medication may have had more severe disease at the start, and it is difficult to pinpoint whether a patient’s condition is improving due to medication use or due to natural fluctuations in migraine symptomatology [7].

Although participants with migraine classified as “overall improved” or “overall worsened” showed clear and consistent changes across the primary defining domains (MHDs, MIDAS score, and MSQ-RFR score), improvement or worsening was not uniformly observed across all secondary outcomes. For example, within-group changes in clinically significant depression and anxiety symptoms (as measured via the PHQ-4) were modest and showed no meaningful change at 1 year in either the “overall improved” or “overall worsened” groups despite clear differences in the primary defining domains. Similarly, less pronounced change was seen in absenteeism than in presenteeism and overall productivity loss, suggesting that migraine-related work impact may be more likely to manifest as reduced job performance than as complete work absence. In addition, acute medication efficacy remained largely unchanged across groups, indicating that improvements in migraine status were not necessarily accompanied by perceived gains in acute treatment effectiveness. These findings underscore the heterogeneity of migraine outcomes over time and highlight that meaningful improvement in migraine disease status does not automatically translate into improvement across all psychosocial or treatment-related domains, reinforcing the importance of multidimensional assessment and targeted interventions beyond headache frequency alone.

Migraine improvement and worsening are inherently multidimensional processes that do not occur uniformly across headache frequency, disability, and HRQoL. OVERCOME (US) has demonstrated substantial discordance across these domains: the previously published OVERCOME progression analysis by Buse et al. showed that classification of migraine progression or improvement varied markedly depending on whether changes in MHDs, MIDAS score, or MSQ-RFR score were considered individually or in combination, with considerable non-overlap across definitions [20]. These findings highlight that reliance on any single domain captures only a subset of individuals experiencing meaningful change. Building on this evidence, the composite, net change approach used in the current analysis was intentionally designed to account for domain-specific discordance while summarizing overall disease trajectory across symptom frequency, disability, and HRQoL. The consistency of observed changes across multiple downstream outcomes not used to define the composite further supports the clinical relevance of this multidimensional framework.

This study had many strengths, including the large representative sample population with individuals who may or may not be receiving treatment from a healthcare provider. Participants were identified using the validated AMS/AMPP migraine diagnostic questionnaire and responded to a wide range of PROs, including validated PROMs. This allowed for an in-depth look at patient health and well-being at 1 year.

This study also has several limitations. The response rate of 19.7% for the 1-year follow-up survey highlights the potential of participation bias in follow-up surveys. Loss to follow-up was not random, with younger age identified as the strongest predictor of non-completion; therefore, the 1-year analytic sample may be biased toward older participants and may not fully represent the baseline population, and no adjustment was applied to account for this attrition. The analysis of factors associated with outcomes including generally improved, generally worsened, or neither worsened or improved at 1 year was detailed; however, there may be unmeasured confounders, for example dietary factors such as caffeine intake or fasting or change in body mass index, that were not included in our analyses. Gini impurity, used in this study to categorize variable importance, can favor continuous variables such as age, potentially inflating importance. The requirement for access to the internet to complete the survey, while not unusual today, may have also resulted in participation bias. Finally, as the results were self-reported, they may be subject to recall bias.

Conclusion

MHDs, MIDAS score, and MSQ-RFR score together provide a robust picture of overall migraine disease status. In this analysis of individuals with migraine from the general population, these 3 variables were used to characterize change in migraine status over the course of 1 year, revealing profound effects of overall improvement or worsening of migraine on physical, functional, and emotional function. These results illustrate the significant effect that improvement in migraine disease status can have on disability, HRQoL, and perception of stigma for an individual with migraine. This work provides important insights for clinicians who treat individuals with migraine and serves as a reminder of the importance of diagnosing migraine and optimizing treatment as quickly as possible to improve functioning and well-being and to prevent further decline over time.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Dawn C. Buse, E. Jolanda Muenzel, Robert A. Nicholson, Richard B. Lipton, and Sait Ashina conceived the analysis. Anthony J. Zagar, Dawn C. Buse, E. Jolanda Muenzel, Robert A. Nicholson, Richard B. Lipton, and Sait Ashina designed the analysis. Ali Sheikhi Mehrabadi, Anthony J. Zagar and Armen Zakharyan analyzed the data. Anthony J. Zagar, Dawn C. Buse, E. Jolanda Muenzel, Gilwan Kim, Robert A. Nicholson, Richard B. Lipton, Robert E. Shapiro and Sait Ashina interpreted the data. Anthony J. Zagar, Dawn C. Buse, and E. Jolanda Muenzel drafted the work. Anthony J. Zagar, Ali Sheikhi Mehrabadi, Dawn C. Buse, E. Jolanda Muenzel, Gilwan Kim, Robert A. Nicholson, Richard B. Lipton and Robert E. Shapiro critically reviewed the work.

Funding

This work, and the rapid service fee, were funded by Eli Lilly and Company (Indianapolis, IN, USA).

Data Availability

Lilly will provide access to anonymized individual participant data collected during the study. The data will be available to request on vivli.org after the study team has completed analyses and publications. Access will be provided after a proposal has been approved by an independent review committee identified for this purpose and after receipt of a signed data sharing agreement. After a proposal is approved, data and documents, including the study protocol, will need to be provided in a secure data sharing environment. For details on submitting a request, see the instructions provided at vivli.org.

Declarations

Conflict of Interest

Dawn C. Buse has received research support from the United States Food and Drug Administration (FDA) and the National Headache Foundation. She serves as consultant or advisory board member or has received honoraria or research support from AbbVie/Allergan, Amgen, Biohaven, Eli Lilly and Company, Ipsen, Lundbeck, Novartis, Teva, and Theranica. E. Jolanda Muenzel is an employee and minor stockholder of Eli Lilly and Company. Ali Sheikhi Mehrabadi was an employee and minor stockholder of Eli Lilly and Company at the time of study. He is now an employee of the School of Medicine, University of Limerick, Limerick, Ireland. Anthony J. Zagar was an employee of Eli Lilly and Company and remains a minor stockholder. Armen Zakharyan is employed by TechData, a company contracted by Eli Lilly and Company (sponsor) to perform the statistical analysis. Sait Ashina has received support for consulting, teaching, and honoraria from Allergan/AbbVie, Amgen, Axsome Therapeutics, Biohaven, Eli Lilly and Company, Impel NeuroPharma, Linpharma, Lundbeck, Novartis, Satsuma, Supernus, Percept, Pfizer, Teva, and Theranica. Robert E. Shapiro serves as consultant or advisory board member or has received honoraria or research support from Eli Lilly and Company, AbbVie, Theranica, Amneal Pharmaceuticals, and Lundbeck. Gilwan Kim is an employee and minor stockholder of Eli Lilly and Company. Robert A. Nicholson is an employee and minor stockholder of Eli Lilly and Company. Richard B. Lipton has received research support from the National Institutes of Health, the FDA, and the National Headache Foundation. He serves as consultant or advisory board member or has received honoraria or research support from AbbVie/Allergan, Amgen, Biohaven, Dr. Reddy's Laboratories (Promius), electroCore, Eli Lilly and Company, GlaxoSmithKline, Eli Lilly and Company, Lundbeck, Merck, Novartis, Pfizer, Teva, Vector, and Vedanta Research. He receives royalties from Wolff's Headache, 8th edition (Oxford University Press, 2009), and Informa. He holds stock/options in Axon, Biohaven, Cooltech, and Manistee.

Ethical Approval

The study was approved by the Sterling Institutional Review Board (IRB ID #6425–001). All participants provided informed consent to participate in the study. The study was conducted in accordance with the Helsinki Declaration of 1964 and its later amendments, consistent with Good Pharmacoepidemiology Practices and applicable laws.

Disclosures

Anthony J Zagar is now retired and Ali Sheikhi Mehrabadi is now an employee of the School of Medicine, University of Limerick, Limerick, Ireland.