Abstract
Background Previous work has highlighted that migraine is underdiagnosed and undertreated in primary care.
Aim To characterise diagnosis and treatment patterns among patients with migraine in England.
Design & setting A retrospective cohort study using the Clinical Practice Research Datalink (CPRD) Aurum linked to the 2019 Index of Multiple Deprivation dataset.
Method The study cohort included patients presenting to general practices with migraine at age ≥18 years from September 2012 to May 2023 (index event). Prescribed medications 12 months after index were compared across sociodemographic stratifications (age, sex, ethnic group, and deprivation). Medication overprescription was defined as ≥10 days’ (opioids and triptans) or ≥15 days’ (analgesics, antipyretics, and non-steroidal anti-inflammatory drugs [NSAIDs]) supply in 3 consecutive months.
Results In total, 1 534 807 patients with any headache or migraine were observed attending primary care in England. Of these, 876 233 (57.1%) were coded with undifferentiated or unclassified headache and 606 928 (39.5%) with a primary headache disorder. Migraine was the most coded primary headache disorder, with 476 191 adults (78.5%), constituting the final sample used for subsequent analyses. Only 36.5% of the migraine cohort were prescribed preventive medication. Regarding prescribed dose, 17.9% of those prescribed amitriptyline, 31.3% prescribed propranolol, and 31.2% prescribed topiramate reached Scottish Intercollegiate Guidelines Network (SIGN) 155 and National Institute for Health and Care Excellence (NICE) Clinical Knowledge Summaries (CKS) recommended doses. In total, 62.6% of the migraine cohort were prescribed an acute medication: triptan (36.6%), and opioid (9.6%). Regarding possible medication overuse, 23.6% of triptan users and 44.3% of opioid users exhibited medication overprescription, as defined by the International Classification of Headache Disorders, 3rd edition (ICHD-3).
Conclusion The majority of headache presenting to primary care remains undifferentiated or unclassified. Only approximately one-third of patients diagnosed with migraine receive triptans or preventive medication. Acute medication overprescription is common and preventive medicines are poorly optimised.
How this fits in
Previous work has highlighted that migraine is underdiagnosed and undertreated in primary care. This research provides updated information on diagnosis and treatment patterns among patients with migraine in England, highlighting continued and substantial migraine underdiagnosis and undertreatment.
Introduction
Migraine affects an estimated 10 million adults in the UK, with prevalence approximately three times higher in females than in males.1 Globally, migraine is the second leading cause of years lived with disability among adults aged 15–49 years, after lower back pain, and the leading cause among women in this age group.2
The majority of episodic headaches presenting to primary care with sufficient severity to prompt consultation are attributable to migraine.3 Headache associated with nausea, photophobia, and functional impairment has previously been found to have high diagnostic sensitivity for migraine.4 Despite this, a review of primary care records between 1987 and 2005 by Kernick et al highlighted that most headache presentations to general practice remain undifferentiated.5 Furthermore, a 2021 report by The Migraine Trust identified persistent challenges in obtaining a timely diagnosis in the UK, and recent evidence suggests that treatment remains suboptimal even after diagnosis.6
Since the publication of Kernick et al’s study, there have been substantial advances in the understanding of migraine pathophysiology, leading to the development of targeted therapies and updated clinical guidelines. NICE issued guidance in 2012 to support the diagnosis and management of headache in primary care, and this guidance was updated in June 2025.7
This study aimed to evaluate patterns in the diagnosis and medical management of migraine in general practice in England since the publication of NICE guidance in 2012, using routinely collected health records. Prescribing trends were characterised among patients with a coded diagnosis of migraine, stratified by age, sex, ethnic group, and deprivation, to explore potential disparities in care.
Method
Data source
This retrospective cohort study used data from the Clinical Practice Research Datalink (CPRD) Aurum obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The data are provided by patients within the primary care setting and collected by the NHS as part of their care and support. Only coded structured data were available, with no access to free-text medical notes. The CPRD Aurum June 2024 build that was used for this analysis is made up of data from n = 1569 currently contributing practices in England, representing 19.9% of all practices in the UK, with 16 227 262 current patients.8 Data were linked to the 2019 English Index of Multiple Deprivation (IMD),9 which provided a relative patient-level composite measure of deprivation based on income, employment, education and skills, health, housing, crime, access to services, and living environment.
Study design and population
At least one migraine diagnosis in the index period (19 September 2012 to 1 May 2023) was required; the first observed migraine diagnosis constituted the index date. The index period start date represented the date of the CG150 NICE guidance,7 and the end date allowed a minimum of 12 months follow-up based on the latest available data at the time of access. In the absence of evidence around average time from diagnosis to first-line medication for migraine in the UK, 12 months was deemed sufficient to observe initial medication. Medication prescriptions were based on prescription events occurring during the 12 months starting from and including the index date.
Full eligibility criteria are detailed below:
at least one migraine diagnosis in the index period;
aged ≥18 years on the index date;
continuously registered with a GP practice for ≥12 months before and following index date;
medical record data considered of acceptable research quality by CPRD;
GP practice in England.
Definitions
Diagnoses (migraine, headache, comorbidities) were defined based on the presence of medcodeIDs or Read codes in medical records. Code lists were developed by the study team and reviewed by a clinical expert. Charlson Comorbidity Index (CCI) was derived based on a previously published algorithm.10 Medication prescriptions were assessed based on the presence of prescription events and prodcodeIDs and were reported for those with no observed migraine-specific medications (triptan and flunarizine) before index to understand first-line use onwards. Medication overprescription was defined as ≥10 days’ (opioids and triptans) or ≥15 days’ (analgesics, antipyretics, and non-steroidal anti-inflammatory drugs [NSAIDs]) supply in 3 consecutive months (90 days), respectively. Days’ supply was taken directly from prescription records in CPRD; if a patient had multiple prescriptions in the 12 months follow-up period, a cumulative value was taken. To understand alignment with Scottish Intercollegiate Guidelines Network (SIGN) 155 and NICE Clinical Knowledge Summaries (CKS) guidance,11,12 dose data were also taken directly from prescription records in CPRD.
Sociodemographic stratifications included age, sex, ethnic group, and deprivation (defined by IMD quintiles), which were all based on pre-coded administrative data, apart from ethnic group, which was based on healthcare professional-recorded medcodeIDs or Read codes.
Analyses
Descriptive analyses were conducted to describe patient characteristics and treatment patterns. Continuous variables were summarised using means, standard deviations, medians, and interquartile ranges, while categorical variables were reported as counts and percentages. All missing values were reported as ‘missing’ while non-missing but uncategorised values (for example, ‘indeterminate’ for sex) were reported as ‘unknown’. Inferential statistics were used to facilitate identification of clinical and treatment differences across sociodemographic strata. T-tests were used for the analysis of continuous variables, while χ2 or Fisher’s exact tests, as appropriate, were used for categorical variables at a two-sided significance level of 0.05. Missing and unknown values were excluded from inferential analyses owing to their limited clinical interpretability. No correction for multiple comparisons was applied, consistent with the exploratory nature of the study. Further details, including full statistical outputs, are provided in the Supplementary Tables. To maintain patient confidentiality, values representing data from fewer than five patients (that is, 1–4) have been suppressed in line with CPRD requirements. All analyses were conducted with Stata (version 18, 18.5, and 19.5).
Results
In total, 3 206 682 patients were observed attending primary care in England with a headache event (Figure 1). After applying exclusion criteria, the pooled headache and migraine cohort was 1 534 807.
The figure is a patient disposition flow diagram showing how the study cohort was derived from an initial extract of patients with a migraine diagnosis or headache event.The initial extract included 3 206 682 patients with a migraine diagnosis or headache event ever recorded. After excluding 160 347 patients, 3 046 335 patients remained with an index event, defined as a migraine diagnosis or headache event between 19 September 2012 and 1 May 2023.The next criterion required a continuous medical record for at least 12 months before and after the index event. After excluding 1 240 976 patients, 1 805 359 patients remained.The cohort was then restricted to patients registered with a general practice in England. After excluding 6275 patients, 1 799 084 patients remained.The final eligibility criterion required patients to be adults, defined as aged 18 years or older at the index date. After excluding 264 277 patients, the final pooled headache and migraine cohort included 1 534 807 adults.The pooled cohort was divided into three diagnostic groups: Undifferentiated headache: 876 233 patients, representing 57.1% of the pooled cohort. Primary headache disorders: 606 928 patients, representing 39.5% of the pooled cohort. Secondary headache disorders: 51 646 patients, representing 3.4% of the pooled cohort.The primary headache disorders group was further subdivided into: Primary headache: tension-type headache: 117 789 patients, representing 19.4% of the primary headache disorders group. Migraine: 476 191 patients, representing 78.5% of the primary headache disorders group. This was the group used for analyses. Primary headache: cluster or other: 12 948 patients, representing 2.3% of the primary headache disorders group.Overall, the diagram presents the sequential exclusion criteria used to identify the eligible adult study population in England and shows that the analytic migraine group comprised 476 191 patients.
The majority of headache presenting to primary care in England was undifferentiated or unclassified (n = 876 233, 57.1%), with primary headache disorders accounting for 39.5% (n = 606 928) of headache presentations. Migraine was the most common primary headache disorder (n = 476 191, 78.5%) followed by tension-type headache (n = 117 789, 19.4%).
Most of the migraine cohort was female (78.0%) or of White ethnic group (n = 262 229, 55.1%, where known: 78.7%), with the overall median age at index 40.0 years (interquartile range: 30.0 years, 52.0 years) (Table 1). The cohort was evenly distributed across deprivation quintiles.
Among the migraine cohort, 33.6% had a recorded history of migraine before the index event, while 66.4% had no prior migraine diagnosis and were considered newly recorded cases.
Almost two-thirds (62.6%) of the migraine cohort who had no prior migraine-specific medication were prescribed medication for acute treatment of migraine within 12 months of index (Table 2). The most prescribed acute medication classes were triptans (36.6%) and NSAIDs (19.9%). Opioids (9.6%) and combinations including an opioid (14.8%) were also commonly prescribed. Medication overprescription was high, particularly in opioid users (44.3%) and analgesic and antipyretic users (40.9%), with >40% of users prescribed doses that are consistent with the International Classification of Headache Disorders, 3rd edition (ICHD-3) medication overuse (MO) definition (Figure 2). Among the medication classes studied, medication overprescription was lowest in triptan users (23.6%).
The figure is a two-panel vertical bar chart showing treatment use during the 12-month follow-up period after the index date. Both panels use purple bars against a white background with a percentage scale from 0% to 100% on the y-axis.Panel A shows possible medication overuse for acute migraine or headache treatments. The panel reports the percentage of patients prescribed each acute medication type in the 12 months after the index date.The acute medication categories and percentages are: Opioid: 44.3% of 33 089 patients. Analgesic and antipyretic: 40.9% of 20 861 patients. NSAID: 30.7% of 68 324 patients. Combination including opioid: 28.1% of 50 901 patients. Triptan: 23.6% of 125 676 patients.In Panel A, opioid had the highest percentage of possible medication overuse, followed by analgesic and antipyretic. Triptan had the lowest percentage among the acute medication categories shown.Panel B shows titration to SIGN 155 and NICE CKS recommended doses for preventive migraine treatments. The panel reports the percentage of patients prescribed each preventive medication who reached the specified recommended daily dose in the 12 months after the index date.The preventive medication categories and percentages are: Amitriptyline, at least 25 mg/day: 17.9% of 41 234 patients. Propranolol, at least 80 mg/day: 31.3% of 41 609 patients. Topiramate, at least 50 mg/day: 31.2% of 7417 patients. Candesartan, at least 16 mg/day: 32.7% of 4252 patients. Valproate, at least 400 mg/day: 35.5% of 1330 patients.In Panel B, valproate had the highest percentage of patients titrated to the recommended dose, followed by candesartan, propranolol, and topiramate, which were similar. Amitriptyline had the lowest percentage of titration to the recommended dose.
CKS = Clinical Knowledge Summaries. MO = medication overuse. NICE = National Institute for Health and Care Excellence. NSAID = non-steroidal anti-inflammatory drug. SIGN = Scottish Intercollegiate Guidelines Network
Just over one-third (36.5%) of the migraine cohort were prescribed a preventive medication in the 12 months after index date (Table 3). The most prescribed preventive medication classes were tricyclic antidepressants and beta blockers. Prescriptions of serotonin antagonists, angiotensin-converting enzyme inhibitor (ACEi) or angiotensin receptor blocker (ARB), calcium channel blockers, and selective serotonin reuptake inhibitors (SSRIs) or serotonin-norepinephrine reuptake inhibitors (SNRIs) were uncommon (≤5%). For medications covered by SIGN 155 and NICE CKS dose guidelines, relatively low percentages reached recommended doses (17.9%–35.5%), as shown in Figure 2. 17.9% of those prescribed amitriptyline, 31.3% prescribed propranolol, and 31.2% prescribed topiramate reached the SIGN 155/NICE CKS recommended doses.
Medication use in the 12 months after index stratified by sociodemographic stratifications is shown in Figure 3 and Supplementary Tables. A higher percentage of 45–54-year-olds, females, or the most deprived were prescribed an acute medication. Triptans were more commonly prescribed in younger adults, females, and use increased as deprivation increased. Opioids and combinations including an opioid were more commonly prescribed in older adults, females, and the most deprived. A higher percentage of ≥55-year-olds, females, and those of White ethnic group were prescribed any preventive medication, and use increased as deprivation increased.
The figure is a grouped bar chart showing medication use in the 12 months after the index date, stratified by sociodemographic characteristics. The y-axis shows the percentage of people with no prior triptan or flunarizine use who received each treatment during follow-up, from 0% to 100%.For any acute medication, the overall percentage was 62.6%. By index age, use was 61.4% for ages 18–44 years, 65.6% for ages 45–54 years, and 63.6% for ages over 55 years. By sex, use was 59.2% for males, 63.6% for females, and 53.3% for unknown sex. By ethnic group, use was 62.6% for White, 69.2% for Black, 68.4% for Asian, 62.3% for Mixed, 62.2% for Other, and 60.7% for unknown ethnic group. By deprivation, use increased from 56.5% in the 1st, least deprived quintile to 70.2% in the 5th, most deprived quintile; the missing deprivation group was 62.5%.For triptan, the overall percentage was 36.6%. By index age, use was 40.8% for ages 18–44 years, 38.7% for ages 45–54 years, and 21.0% for ages over 55 years. By sex, use was 34.2% for males, 37.3% for females, and 46.7% for unknown sex. By ethnic group, use was 36.4% for White, 38.8% for Black, 37.2% for Asian, 39.7% for Mixed, 38.9% for Other, and 36.2% for unknown ethnic group. By deprivation, use ranged from 34.8% in the 1st, least deprived quintile to 38.6% in the 5th, most deprived quintile; the missing deprivation group was 36.6%.For combinations including opioid, the overall percentage was 14.8%. By index age, use was 12.9% for ages 18–44 years, 16.7% for ages 45–54 years, and 19.5% for ages over 55 years. By sex, use was 12.9% for males and 15.4% for females; the value for unknown sex is shown as an asterisk (which means data were suppressed in line with CPRD requirements due to the small n number of between 1 to 4). By ethnic group, use was 14.7% for White, 20.8% for Black, 20.0% for Asian, 15.5% for Mixed, 13.4% for Other, and 13.5% for unknown ethnic group. By deprivation, use increased from 9.8% in the 1st, least deprived quintile to 21.3% in the 5th, most deprived quintile; the missing deprivation group was 15.2%.For opioid, the overall percentage was 9.6%. By index age, use was 7.7% for ages 18–44 years, 11.7% for ages 45–54 years, and 14.2% for ages over 55 years. By sex, use was 8.0% for males, 10.1% for females, and 0.0% for unknown sex. By ethnic group, use was 10.4% for White, 7.2% for Black, 7.0% for Asian, 7.1% for Mixed, 8.1% for Other, and 9.4% for unknown ethnic group. By deprivation, use increased from 7.4% in the 1st, least deprived quintile to 12.7% in the 5th, most deprived quintile; the missing deprivation group was 9.7%.For any preventive medication, the overall percentage was 36.5%. By index age, use was 32.9% for ages 18–44 years, 41.2% for ages 45–54 years, and 43.8% for ages over 55 years. By sex, use was 35.0% for males, 36.9% for females, and 40.0% for unknown sex. By ethnic group, use was 37.7% for White, 32.8% for Black, 35.3% for Asian, 31.1% for Mixed, 34.1% for Other, and 35.4% for unknown ethnic group. By deprivation, use increased from 31.8% in the 1st, least deprived quintile to 42.7% in the 5th, most deprived quintile; the missing deprivation group was 36.8%.Overall, the figure shows that acute medication use was common across all sociodemographic groups. Triptan use was lower in the oldest age group than in younger age groups. Opioid and combination opioid use generally increased with age and with greater deprivation. Preventive medication use was also higher in older age groups and in more deprived quintiles.
Discussion
Summary
Overall, findings suggest substantial migraine underdiagnosis and undertreatment in England.
This study highlights the high frequency of headache presentations in primary care, with 3 206 682 patients consulting for headache in England observed in this dataset. Although the coding of headache in primary care has improved compared with the previous Kernick et al study,5 where more than 70% of headache presenting to primary care were undifferentiated or unclassified between 1987 and 2005, 57.1% of the 1 534 807 patients included in this study still lacked specific diagnostic classification.
While the percentage of the migraine cohort prescribed any preventive treatment is similar to previous studies,13 this present study suggests that in addition to being underutilised in primary care, the majority of users are not prescribed the SIGN 155- and NICE CKS-recommended doses. And further, opioids are still commonly prescribed for acute treatment of migraine despite NICE guidance clearly stating not to offer these agents.7 Medication overprescription of acute medication was common, suggestive of possible medication overuse.
Strengths and limitations
This study offers current, population-based evidence on the real-world management of migraine in England. CPRD Aurum data are considered broadly comparable with the UK general population in terms of sociodemographic features, where coded diagnoses have high positive predictive values.14
While it is possible that observed migraine diagnoses were incorrect, this is unlikely to represent a significant proportion of the study population (in line with previous Landmark study findings)3 and unlikely to explain observed low medication use. Key limitations relate to coding of medical events and operationalised definitions. Reason for prescriptions could not be ascertained, therefore, while this study’s cohort had diagnosed migraine, it is possible that observed prescriptions were prescribed for an alternative comorbidity, posing a risk of over-representing medication prescriptions for migraine in the present study. Conversely, over-the-counter medications or those administered in specialist or secondary care are typically not observed in CPRD, posing a risk of under-representing prescriptions. The medication overprescription status reported in this study is based on days’ supply of prescribed medications; it is unknown whether prescriptions were dispensed and/or taken continuously, hence the definition is an approximation of medication overuse only, as defined by ICHD-3. A prospective study surveying patients with migraine to ascertain true medication use and expert review of case notes to determine coding accuracy would be needed to overcome these limitations.
In addition, as the cohort used was so large, statistical comparisons were not reported owing to risk of over-powered results, which may achieve statistical significance without necessarily reflecting clinical meaningfulness. For inclusion in this study, patients were required to have at least 12 months of continuous data before and following index, which resulted in around 1.2 million patients being excluded (Figure 1). Although this may have introduced elements of selection bias, this was deemed essential to fully observe outcomes and remove patients with limited data availability.
Comparison with existing literature
The previous Kernick study, a similar retrospective cohort of primary care records but between 1987 and 2005, found that 70.1% of headache presentations were undifferentiated, meaning the patient did not receive a diagnosis.5 This study shows that in the nearly 20 years since publication of Kernick et al,5 coding of headache presentations has improved but that more than half (57.1%) of headache presentations are still undifferentiated. Based on the Landmark study,3 most patients coded with undifferentiated or unclassified headache in primary care likely have migraine. Investigating this group further (for example, review of free-text notes or prescribing trends) could therefore provide a more accurate understanding of the burden of migraine and access to treatments.
Migraine preventive medication was prescribed to a similar percentage of patients in this study when compared with previous studies in the UK, 36.5% versus 42.4%.13 The present study likely had a slightly lower percentage of preventive medication use compared with Kernick et al,13 as the follow-up period was 12 months rather than 3 years.
Acute and preventive medications were generally prescribed more frequently to those in the most deprived groups, which may be owing to the accessibility of free NHS prescriptions through the NHS Low Income Scheme. Previous research has also found that more prescriptions, specifically of dependency-forming medicines (such as opioids) are issued in the most deprived areas of England.15,16 Regarding acute medications, triptans are prescribed most to those younger in age (18–44 years and 45–54 years compared with ≥55 years), which is likely a reflection of limited systematically evaluated safety and efficacy data for triptans in patients aged ≥65 years.17 Opioids exhibit a similar trend to opioid combinations, with prescriptions being higher in older adults, reflecting previous research from the UK18 and may be indicative of the management of health conditions other than migraine. Regardless of the primary indication, the finding that 14.8% of patients diagnosed with migraine received a combination including an opioid and 9.6% received an opioid is cause for concern given the NICE guidance against offering these agents.
Implications for research and practice
In conclusion, the findings of the present study underscore that migraine remains underdiagnosed and undertreated despite national guidelines and local pathways.
Qualitative research may be required to understand the needs of primary care clinicians to improve diagnosis and management of migraine in primary care. In adults, although acute medications (primarily triptans and NSAIDs) largely reflect current clinical guidance, the notable prescription of opioid medications, outside of recommended guidelines, suggests opportunities for improving adherence to best practice. Future research could consider community pharmacy as a mechanism to reduce opioid use, including those purchased over the counter, for acute treatment of migraine.
Preventive medication appears to be underused and poorly titrated to recommended doses. Again, future research could consider whether pharmacy can support titration to NICE CKS- and SIGN 155-recommended doses for migraine preventive medication and improve outcomes accordingly.
Notes
Funding
This study was funded by Pfizer Ltd, NISAMR ID 2024032114.
Ethical approval
The protocol for this research was approved by CPRD’s Research Data Governance Process (protocol number: #24_004055). This study is based in part on data from the CPRD obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The data are provided by patients and collected by the NHS as part of their care and support. Linked data were re-used with the permission of The Health & Social Care Information Centre, all rights reserved. The interpretation and conclusions contained in this study are those of the authors alone.
Provenance
Freely submitted; externally peer reviewed.
Data
Data are available on request from the CPRD. Their provision requires the purchase of a licence, and this licence does not permit authors to make data publicly available.
Acknowledgements
Data analysis was supported by Kostas Alexiou, an employee of Adelphi Real World at the time of the study.
Competing interests
David PB Watson has received consulting fees from Pfizer Ltd. Lucinda Camidge, Elke Rottier, Hannah Gowman, and Sarah Law worked for Adelphi Real World at the time of the study, which received funding from Pfizer Ltd to conduct the study and develop this manuscript. Alannah Williams has received undergraduate placement support from Pfizer Ltd. Grant O’Neil, Robert Pawinski, Maciej Czachorowski, Marzieh Araghi, Rowena Randall, and Alannah Williams were employees of Pfizer Ltd at the time of the study and held financial equities in the company.
- Received July 21, 2025.
- Revision received November 24, 2025.
- Accepted December 22, 2025.
- Copyright © 2026, The Authors
This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)
LinkedIn
Jump to section
Keywords
More in this TOC Section
Related Articles
Cited By...
