The Challenges for Pharmacoepidemiologists Identifying Migraine in Electronic Healthcare Data Sources: A Systematic Literature Review

The Challenges for Pharmacoepidemiologists Identifying Migraine in Electronic Healthcare Data Sources: A Systematic Literature Review | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL Pharmacoepidemiology and Drug Safety This is a preprint and has not been peer reviewed. Data may be preliminary. 25 January 2025 V1 Latest version Share on The Challenges for Pharmacoepidemiologists Identifying Migraine in Electronic Healthcare Data Sources: A Systematic Literature Review Authors : Joan Forns 0000-0002-1066-0358 [email protected] , Alicia Abellan , Nuria Riera-Guardia 0000-0001-9482-7524 , Andrea Margulis 0000-0001-7388-6082 , and Elena Rivero-Ferrer Authors Info & Affiliations https://doi.org/10.22541/au.173781278.87655336/v1 Published Pharmacoepidemiology and Drug Safety Version of record Peer review timeline 299 views 163 downloads Contents Abstract Key Points Plain Language Summary Abstract Introduction Methods Results Discussion Conclusions Acknowledgements Tables Figures References Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Ascertaining migraine in electronic healthcare data is challenging because of likely diagnosis underrecording and treatment with over-the-counter analgesics, which cannot be used as disease proxies. Algorithm-identified migraine prevalence may depend on algorithm characteristics and target population. To describe migraine-identifying algorithms implemented in electronic healthcare data sources and summarize validation results and observed migraine prevalence, we searched PubMed for peer-reviewed, English-language, original research articles that identified migraine in adults using electronic algorithms in electronic healthcare data. We summarized algorithms, validation results, and migraine prevalence (PROSPERO: CRD42023491279). Of 360 unique titles and abstracts, 50 articles (14%) were selected for full-text review; of them, 41 articles (82%) were finally included: 16 were conducted in Europe, 13 in North America, and 12 in Asia. Sixteen studies (39%) identified migraine only using diagnosis codes, 5 (12%) only treatments, 9 (22%) diagnosis and/or treatment codes, and 11 (27%) diagnosis codes, treatments, and setting (e.g., primary care, specialist consultation). Reported migraine prevalence in the general population ranged between 4% and 17%. Only 2 studies reported validation results: one identified prevention-eligible patients with migraine (positive predictive value [PPV] = 97%), and one identified migraine on the basis of calculated probabilities with PPVs between 74% and 92%. Finding patients with migraine is feasible in various types of data sources; preferred algorithms vary; algorithm performance is mostly unknown. Identifying chronic migraine or other complex types of migraine requires combining diagnosis codes, treatments, and care settings, which is possible in only some data sources. The Challenges for Pharmacoepidemiologists Identifying Migraine in Electronic Healthcare Data Sources: A Systematic Literature Review Running title: Identifying Migraine in Electronic Data Sources Joan Forns, MPH, PhD 1 https://orcid.org/0000-0002-1066-0358; Alicia Abellan, MSc, PhD 1 https://orcid.org/0000-0001-6552-6060; Nuria Riera-Guàrdia, PharmD, PhD 1 https://orcid.org/0000-0001-9482-7524; Andrea V. Margulis, MD, ScD, FISPE 1 https://orcid.org/0000-0001-7388-6082; Elena Rivero-Ferrer, MD, MPH, FISPE 1 https://orcid.org/0000-0001-5093-2445 Author Affiliations: 1 Pharmacoepidemiology and Risk Management, RTI Health Solutions, Barcelona, Spain. Corresponding Author: Joan Forns RTI Health Solutions Av. Diagonal, 605, 9-1, 08028 Barcelona, Spain Telephone: +34.93.241.7761 Fax: +34.93.760.8507 [email protected] Prior presentation: Forns J, Abellan A, Riera-Guardia N, Margulis AV, Rivero-Ferrer E. Identification of migraine in electronic healthcare data sources: a headache for pharmacoepidemiologists – a systematic literature review. Poster presented at the 2024 ISPE Annual Meeting; August 27, 2024. Berlin, Germany. Declaration of interest: JF, AA, NR-G, AVM, and ER-F are employees of RTI Health Solutions. Funding: This study was funded by RTI Health Solutions. Key Points • Finding patients with migraine is feasible in various types of electronic healthcare data (healthcare claims data, electronic medical records, or other existing automated data sources). • The performance of the algorithms used to identify migraine is largely unknown. • Identifying chronic migraine or other complex types of migraine requires combining diagnosis codes, treatments, and care settings, which is possible in only some data sources. Plain Language Summary Finding whether a person has migraine in data arising from encounters between patients and health care providers is challenging; but it is necessary in order to evaluate migraine or the use, safety or effectiveness of migraine treatments in these health care data sources that reflect the real-world experience of patients. Therefore, we systematically identified and reviewed epidemiological studies that identified migraine in this type of data published between 2013 and 2023. In total, 41 articles were included. Of them, 39% identified migraine using diagnoses recorded by physicians in primary care or hospitals; 12% identified migraine from prescribed migraine treatments (like triptans); 22% identified migraine using a combination of diagnoses and/or treatments; finally, 27% used a combination of diagnoses and treatments and required additional information about the setting in which the diagnoses and/or treatments were made or prescribed (for example, primary care, specialist consultation). These studies reported that 4% to 17% of the general population has migraine. The performance of the methods was rarely reported ((for example, did they correctly identify all patients with migraine?). In conclusion, identifying migraine is possible in various types of electronic data sources. However, researchers do not know how well they can correctly identify migraine. Abstract Ascertaining migraine in electronic healthcare data is challenging because of likely diagnosis underrecording and treatment with over-the-counter analgesics, which cannot be used as disease proxies. Algorithm-identified migraine prevalence may depend on algorithm characteristics and target population. To describe migraine-identifying algorithms implemented in electronic healthcare data sources and summarize validation results and observed migraine prevalence, we searched PubMed for peer-reviewed, English-language, original research articles that identified migraine in adults using electronic algorithms in electronic healthcare data. We summarized algorithms, validation results, and migraine prevalence (PROSPERO: CRD42023491279). Of 360 unique titles and abstracts, 50 articles (14%) were selected for full-text review; of them, 41 articles (82%) were finally included: 16 were conducted in Europe, 13 in North America, and 12 in Asia. Sixteen studies (39%) identified migraine only using diagnosis codes, 5 (12%) only treatments, 9 (22%) diagnosis and/or treatment codes, and 11 (27%) diagnosis codes, treatments, and setting (e.g., primary care, specialist consultation). Reported migraine prevalence in the general population ranged between 4% and 17%. Only 2 studies reported validation results: one identified prevention-eligible patients with migraine (positive predictive value [PPV] = 97%), and one identified migraine on the basis of calculated probabilities with PPVs between 74% and 92%. Finding patients with migraine is feasible in various types of data sources; preferred algorithms vary; algorithm performance is mostly unknown. Identifying chronic migraine or other complex types of migraine requires combining diagnosis codes, treatments, and care settings, which is possible in only some data sources. Key words: Migraine, systematic literature review, algorithm, validation. Introduction Migraine is characterized by moderate-to-severe episodic, unilateral, pulsating headaches that last for 4 to 72 hours. The global age-standardized prevalence of any/episodic migraine in 2016 was estimated to be 14.4% using published, population-based, cross-sectional studies using the International Classification of Headache Disorders (95% confidence interval [CI], 13.8%-15.0%). 1 The prevalence of migraine differs by sex, with a higher prevalence among females (global age-standardized prevalence, 18.9% [95% CI, 18.1%-19.7%]) than among males (global age-standardized prevalence, 9.8% [95% CI, 9.4%-10.2%]). Additionally, the prevalence of migraine differs by age and, to a lesser extent, by environmental factors, geography, and genetics. 1 According to the International Headache Society, chronic migraine is a headache that occurs on 15 or more days per month for more than 3 months and that, on at least 8 days per month has the features of migraine headache. 2 The global prevalence of chronic migraine has been estimated to range from 1.4% to 2.2% based on self-reports. 3 A number of observational studies requested by regulatory authorities to evaluate the use, effectiveness, and safety of new medications to treat migraine are planned or ongoing. The majority of these studies are conducted using electronic healthcare automated data sources. However, the ascertainment of migraine in these data sources is challenging for a number of reasons: migraine diagnosis underrecording, use of over-the-counter analgesics as a treatment option (which cannot be used as disease proxies), treatment with analgesics that are also used for other pain conditions (these analgesics cannot be directly used as proxies for migraine), and provisional or even rule-out diagnoses. Therefore, the ability to identify the target population with migraine in these studies depends on the characteristics of the migraine-identifying algorithm. The present systematic literature review aimed to describe migraine-identifying algorithms implemented in observational studies that use healthcare claims data, electronic medical records, or other existing automated data sources; describe related validation efforts; and summarize the prevalence of migraine observed in those studies. Methods The protocol for this systematic literature review was registered with PROSPERO (CRD42023491279). Search Strategy The literature search was designed to identify observational studies conducted in electronic healthcare data describing migraine-identifying algorithms to find the study population, the exposure, or the outcomes. The search strategy was based on the PICOS framework (population, intervention, comparison, outcomes, study design; Table 1 and Supplementary Table S1). PubMed was searched in November 2023 for peer-reviewed, English-language, original research articles published since 2013. The literature search included search terms for patient/population (migraine), data sources, and adult age (Supplementary Table S2). Study Screening In the initial screening phase (level 1), the title and abstract of each study identified in PubMed were screened independently by 2 reviewers to ascertain eligibility according to the prespecified eligibility criteria listed in Table 1, using 2 Excel files of identical structure, 1 for each reviewer. Each excluded study was assigned a single reason for exclusion, even if multiple reasons were applicable, to facilitate quality control of numbers of included and excluded studies. In the event of a discrepancy regarding the inclusion of a study, consensus was reached through discussion between the 2 reviewers. In the second level of screening (level 2), the full text of each article that passed level 1 screening was independently evaluated by 2 researchers to ascertain eligibility in accordance with the same eligibility criteria (Table 1) and using the same process as in level 1 screening. Screening results were recorded in a comprehensive Excel file and illustrated in a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) diagram (Figure 1). 4 Data Extraction The data were extracted from the full-text publications. Any references to other publications were traced to original sources, where applicable. The data were extracted by 1 researcher using table templates in Microsoft Word format. A second researcher then checked the extracted data against the article from which the data were extracted. The researcher who conducted the primary data extraction corrected any mistakes. Any disagreements were resolved by discussion and consensus between the researchers, with escalation to the broader research team when necessary. The data extracted included the following elements: first author, year, journal, citation, data source name and country, migraine operational definition/algorithm, prevalence of migraine, and validation statistics (positive predictive value [PPV], negative predictive value [NPV], sensitivity, or specificity). Quality Assessment Quality assessment was aimed at evaluating the potential for bias in the reported prevalence of migraine in the selected studies. The assessment was conducted using a set of questions from the Joanna Briggs Institute’s critical appraisal tools for prevalence studies. 5 The tool was applied to each included article by a single researcher (see Supplementary Table S3). Results The literature search identified 360 unique articles. After reviewing titles and abstracts, 50 (14%) were selected for full-text review; of those, 41 (82%) were selected for data extraction (Figure 1). Sixteen studies were conducted in Europe, 13 in North America, and 12 in Asia. Among the 41 included articles, 16 (39%) identified migraine solely based on the basis of diagnosis codes, 5 (12%) used only migraine treatments, 9 (22%) used combinations of diagnosis codes and/or treatments, and 11 (27%) used combinations of diagnosis codes and treatments in different settings (e.g., primary care, hospital discharge, specialist consultations). (See Table 2 for a summary of results and Supplementary Table S4 for a more detailed presentation of the results). Of the 16 studies that identified migraine using only diagnosis codes, 12 were conducted in electronical medical record data sources and 4 in claims databases. Most studies required only 1 code to ascertain migraine, 6-15 whereas 3 studies required 2 or more codes. 16-19 Nearly all used the 346.xx code in the International Classification of Diseases, Ninth Revision (ICD-9) or the G43.xx code in the International Classification of Diseases, Tenth Revision (ICD-10). ICD codes of 4 or 5 digits were used to identify specific types of migraine, such as episodic migraine (ICD-9: 346.00, 346.10, 346.20, 346.90; ICD-10: G43.109, G43.009, G43.809, G43.909), chronic migraine (ICD-9: 346.7x; ICD-10: G43.7xx), treatment-resistant migraine or status migrainosus (ICD-9-Clinical Modification [ICD-9-CM] codes 346.92 and 346.93), or migraine without aura (ICD-9-CM: 346.1, 346.2, 346.4, 346.7, 346.8, 346.9; see Table 2). The prevalence of acute migraine in the general population in these studies ranged from 4.0% in German claims data 17 to 4.8% 12 in Finish electronic medical records data. Another study reported a prevalence of 6.1% in people with type 2 diabetes mellitus in a South Korean electronic medical record database. 8 Of the 5 studies that identified migraine using only migraine treatments, 3 were conducted in the Nordic National Health Registers, 20-22 and 2 in commercial claims databases. 2324, Triptans were the most common medication used to identify migraine (Anatomical Therapeutic Chemical Classification System [ATC] N02CC). Other drugs commonly used were ergotamines (N02CA); in one study, the authors used migraine prophylactic drugs (i.e., topiramate, beta-blockers, or tricyclic antidepressants). Three of these studies required only 1 prescription/dispensing, 202224, while 2 required more than 1 prescription. 2123, The prevalence of migraine reported by Egeberg et al. 20 was 7.3% and 12.1% in the reference population and in patients with rosacea, respectively. Nine studies used diagnosis codes and/or treatments to ascertain migraine (i.e., the algorithm included either one or both types of codes). Of them, 4 were conducted in national health registers, 4 in commercial claims databases, and 1 in an electronic medical record database . Almost all identified studies only required a single diagnosis code (ICD-9 346.xx and ICD-10 G43.xx) and/or a treatment code (mostly triptans). 25-32 One study used an algorithm requiring both a prior chronic migraine diagnosis code (ICD-9 346.7) and the initiation of a treatment with oral migraine preventive medications 33 . Only 3 studies reported prevalence of migraine ranging from 7.7% in a study using electronic medical records from the general population in Israel 2930, to 17.0% in a study of the general population in the Danish National Health Registers. 26 Finally, 11 studies required additional information to ascertain migraine, including diagnosis codes, treatments, and health care setting. 34-44 Of those, 6 were conducted in a commercial claims database, 3 in electronical medical record data sources (including an electronic health records database and a general practitioner database in Italy), 1 in the Danish National Health Registers, and 1 in health insurance data from Germany. A summary of the algorithms is included in Table 2. Most of the algorithms required at least 2 elements (e.g., 1 diagnosis code and 1 treatment), and most of them also required a time frame between qualifying codes. Other studies first identified patients with a diagnosis of migraine and then created subsequent cohorts to identify treatment-resistant patients. Two studies used a migraine probability algorithm (MPA), which is a validated approach to identify individuals diagnosed with migraine in electronic medical record data sources. The MPA results in a numeric score based on healthcare encounters with a primary or secondary diagnostic code for migraine; presence of a diagnostic code for migraine in the patient’s Significant Health Problem List; and prescriptions for migraine-specific abortive medications (i.e., triptans, ergotamines). The MPA score ranges from 0 to 101; scores greater than 10 were considered consistent with a diagnosis of migraine. 3544, Of the studies that required a combination of codes, treatments, and settings, 4 reported migraine prevalence. Using an electronic medical record database from the general population in the United States, Pressman et al. 44 estimated the prevalence of migraine was 11.9% using the MPA. Gaul et al. 38 reported a 2-year prevalence of 4.3% using a combination of codes and treatments in the general population in an electronic medical record database from Germany. Using a German claims database, Gendolla et al. 39 reported a prevalence of 4.8% using diagnosis codes from different care settings. Finally, Marconi et al. 42 used an Italian electronic medical record database in which the researchers identified chronic migraine using several algorithms that combined the timing of diagnosis codes and treatments; they reported a prevalence of chronic migraine ranging from 0.03% (algorithm 2A; i.e., at least 1 ICD-9-CM–coded, migraine-related contact per month for 3 consecutive months or 3 migraine-related contacts for 3 consecutive months combined with medication use for migraine attacks) to 0.28% (algorithm 1C; i.e., at least 1 ICD-9-CM–coded, migraine-related contact per month for 3 consecutive months or 2 contacts per month for 2 consecutive months and/or medication use for migraine attack ). Only 2 studies reported validation results. First, Foster et al. 37 used an algorithm that combined diagnosis codes and combinations of treatments and time to identify patients eligible for preventive treatment and achieved a PPV of 97.3%, specificity of 65.5%, and sensitivity of 81.7% (see Table 2 ). Second, Pressman et al. 44 validated the MPA and reported a PPV of 74.3% and sensitivity of 85.0%. When the gold standard was relaxed to include patients with definite or probable physician-diagnosed migraine, the PPV increased to 92.2% and the sensitivity decreased to 79.5%. Further details of the validation in this study are shown in Table 2. Results of the quality assessment indicated that all studies were of adequate quality. Discussion To our knowledge, this is the first systematic literature review to evaluate the ascertainment and validity of migraine diagnosis in automated electronic healthcare data sources. We reviewed 41 studies that used algorithms to identify migraine in data sources from Europe, North America, and Asia. Most studies included in this systematic review used diagnosis codes to identify migraine. Only a few studies that used a combination of diagnosis codes and treatments in different care settings identified specific migraine types and reported results on validation. As expected, the prevalence varied according to the type of algorithm and the underlying population. The findings from this systematic literature review may guide future studies to select the most appropriate migraine-identification algorithm for their research question and setting. Algorithms that use a combination of diagnosis codes, treatments, and other requirements, such as number of treatments, type of healthcare encounter, or time between different elements, have the potential to reduce or increase the risk of underascertainment of migraine (i.e., if diagnosis and treatments are required with “AND,” this may increase the underascertainment, while the requirement of diagnosis “OR” treatments may reduce the underascertainment). However, this type of algorithm requires a suitable data source; many electronic healthcare data sources have only a subset of the required information. Below, we discuss several example scenarios: • Studies that need to identify a population with “any” migraine, regardless of severity or clinical characteristics. Studies in this review used algorithms requiring 1 or 2 data elements. In this scenario, selecting the most appropriate algorithm depends on the type of data source available to conduct the study. For example, in claims data, codes for rule-out diagnosis may artificially increase the prevalence of “any” migraine. Therefore, in this case, a combination of diagnosis codes, treatments, and other requirements requiring more than 1 data element would be preferred. On the other hand, in almost all electronic medical records data sources, a simple algorithm that requires only 1 diagnosis code might be preferred because chronic conditions are recorded only once; algorithms that require 2 migraine diagnosis codes may artificially reduce the estimated prevalence. While the prevalence of “any” migraine has been reported to be 14.4% (95% CI, 13.8%-15.0%), 1 prevalences observed in this review were lower, except the 17% prevalence observed by Orimoloye et al. 26 in a study conducted in the Danish National Health Registers identifying patients from the general population. • Studies that need to identify a population with a specific type of migraine (e.g., chronic migraine, treatment-resistant migraine). This scenario requires more complex algorithms; 11 studies included in the current review were able to distinguish these specific and more complex types of migraine using combinations of codes, treatments, and additional requirements. If data sources available for research lack some of these data elements (e.g., hospital information or general practitioner information), simpler algorithms may still be helpful, although they may have a higher risk of underascertainment of migraine. • Studies that address the use, safety, or effectiveness of a new treatment for acute migraine. Identifying patients treated for acute migraine is possible through use of a simple algorithm requiring treatment codes that are only indicated for acute migraine like triptans or ergotamines. However, identifying a broader population of patients, including untreated patients, patients who use over-the-counter medications, or patients who use medications with additional indications (all of whom might benefit from treatments for acute migraine), requires more complex algorithms. In this review, we did not find studies that evaluated nonspecific medications to treat acute migraine. • Studies that address the use, safety, or effectiveness of a new treatment for chronic migraine. In this case, algorithms combining diagnosis codes and treatments would be required because most of the treatments for chronic migraine have nonmigraine indications that may contribute to misclassification of the population. Only in 1 study 42 was the prevalence of chronic migraine identified through a combination of codes, treatments, and timing of health encounters; it ranged from 0.03% to 0.28% depending on the complexity of the definition, which is lower than the global prevalences reported in a systematic review, ranging from 0.9% to 5.1% in the general population, with estimates typically in the range of 1.4%-2.2%. 3 Limitations of the present study include that our systematic literature search, designed by a specialist librarian and epidemiologists with clinical background, may have missed relevant articles; articles published before 2013 were not considered current and of interest, and we searched only the PubMed database. Also, identifying “administrative health care data sources” in the search strategy was not straightforward, as there is no MeSH (Medical Subject Headings) term to capture this type of study. Conclusions The ascertainment of migraine in electronic health care data sources is challenging but feasible in many types of data source and through the use of a variety of algorithms. Identifying patients with chronic migraine or other specific types of migraine requires more complex algorithms that combine diagnosis codes, treatments, care settings, or other elements that are available only in some electronic health care data sources. Acknowledgements The authors acknowledge the editorial assistance of John Forbes of RTI Health Solutions in the preparation of this manuscript. Tables Table 1. Study Eligibility Criteria for Screening Population Eligible: Adult population registered in an automated electronic health care data source Eligible: The study is on a human population There are no restrictions related to geographic region of the study population Intervention No particular intervention was required Outcomes Eligible, level 1 screening: The abstract contains an operational definition of migraine (or at least mentions that an operational definition/algorithm [i.e., diagnosis codes and/or treatments proxies] has been used) Eligible, level 2 screening: The full text contains a full description of the operational definition/algorithm used to ascertain migraine Not eligible: Studies in which migraine was a covariate or a patient characteristic for descriptive purposes only Study design/publication type Eligible: The study is original research; the publication has an abstract Not eligible: Letters, commentaries, editorial, expert opinion/reviews, clinical or other guidelines, case reports or case series, conference abstracts, preprints; reviews will be marked for future reference and for reference list review but are not eligible Eligible: Studies in English Table 2. Key Elements of Included Studies Orimoloye et al. 26 (2023) Denmark National Health Register Yes Yes 1 No 17% among controls (i.e., general population) Hvitfeldt Fuglsang et al. 21 (2023) Denmark National Health Register No Yes 2 No Not reported Peles et al. 29 (2023) Peles et al. 30 (2022) Israel EMR Yes Yes 1 No General population: 7.65% Females: 11.43% Males: 3.75% Gaul et al. 38 (2023) Germany EMR Yes Yes 2 Yes General population: 2-year “prevalence rate”: 4.3% Cheon et al. 8 (2022) South Korea EMR Yes No 1 No Among patients with type 2 diabetes mellitus: 6.1% Shao et al. 18 (2022) US Claims Yes No 2 No Not reported Nguyen et al. 23 (2022) US Claims No Yes 2 No Not reported Park et al. 28 (2022) South Korea EMR Yes Yes 1 No Not reported Gendolla et al. 39 (2022) Germany Claims Yes Yes 1 Yes Among general population: 4.8% (2019) Richer et al. 14 (2022) Canada EMR Yes No 1 No Not reported Elser et al. 35 (2021) US EMR Yes Yes All b Yes Not reported Wang et al. 32 (2021) Denmark National Health Register Yes Yes 1 for diagnosis codes, 2 for treatments No Not reported Orlando et al. 27 (2020) Italy EMR Yes Yes 1 for diagnosis codes, 2 for treatments No Not reported Foster et al. 37 (2020) US Claims Yes Yes 2 Yes Not reported Islamoska et al. 41 (2020) Denmark National Health Register Yes No 2 Yes Not reported Roessler et al. 17 (2020) Germany Claims Yes No 2 (only for outpatient diagnosis) No Among general population: 3.98% (1.60% in men and 6.09% in women) Marconi et al. 42 (2020) Italy EMR Yes Yes All b Yes “Prevalence rate” of chronic migraine: ranging from 0.03% to 0.28% Ford et al. 16 (2020) US Claims Yes No 2 (only for outpatient diagnosis) No Not reported Ford et al. 36 (2019) US Claims Yes Yes All b Yes Not reported Hoffman et al. 40 (2019) US Claims Yes Yes All b Yes Not reported Meyers et al. 43 (2019) Japan Claims Yes Yes All b Yes Not reported Nyberg et al. 22 (2019) Sweden National Health Register No Yes 1 No Not reported Wang et al. 19 (2019) Taiwan EMR Yes No 2 No Not reported Thomsen et al. 31 (2019) Denmark National Health Register Yes Yes 1 No Not reported Korolainen et al. 12 (2019) Finland EMR Yes No 1 No Overall prevalence: 4.8% Hwang et al. 11 (2018) Taiwan EMR Yes No 1 No Not reported Harnod et al. 10 (2018) Taiwan EMR Yes No 1 No Not reported Bonafede et al. 34 (2018) US Claims Yes Yes All b Yes Not reported Adelborg et al. 25 (2018) Denmark National Health Register Yes Yes 1 No Not reported Woolley et al. 24 (2017) US Claims No Yes 1 No Not reported Champaloux et al. 6 (2017) US Claims Yes No 1 No Not reported Hepp et al. 33 (2017) US Claims Yes Yes 1 No Not reported Egeberg et al. 20 (2017) Denmark National Health Register No Yes 1 No Reference population: 7.3% Patients with rosacea: 12.1% Chen et al. 7 (2016) Taiwan EMR Yes No 1 No Not reported Pressman et al. 44 (2016) US EMR Yes Yes All b Yes General population Women: 14.4% Men: 4.7% Patients aged ≥18 years Total prevalence using a broader definition of migraine: 11.9% Women: 17.6% Men: 5.4% Total prevalence using a stricter definition of migraine: 10.3% Women: 15.5% Men: 4.5% Wu et al. 45 (2016) Taiwan EMR Yes No 1 No Not reported Yang et al. 15 (2016) Taiwan EMR Yes No 1 No Not reported Lau et al. 13 (2015) Taiwan EMR Yes No 1 No Not reported Harnod et al. 46 (2015) Taiwan EMR Yes No 1 No Not reported Chu et al. 9 (2015) Taiwan EMR Yes No 1 No Not reported EMR = electronic medical record; US = United States. a For example, diagnosis from different settings or time between codes. b See Supplementary Table S3 for further details about each specific algorithm. Figures Figure 1. Study Selection Process in a PRISMA Diagram PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Reference: Page et al. 4 References 1. http://dx.doi.org/10.1016/s1474-4422(18)30322-3 http://dx.doi.org/10.1177/0333102417738202 http://dx.doi.org/10.1111/j.1468-2982.2009.01941.x http://dx.doi.org/10.1016/j.jclinepi.2021.03.001 https://jbi.global/sites/default/files/2020-08/Checklist_for_Prevalence_Studies.pdf http://dx.doi.org/10.1016/j.ajog.2016.12.019 http://dx.doi.org/10.1097/md.0000000000003670 http://dx.doi.org/10.1186/s12933-022-01705-3 http://dx.doi.org/10.1186/s10194-015-0547-z http://dx.doi.org/10.1186/s10194-018-0889-4 http://dx.doi.org/10.1001/jamaoto.2018.0939 http://dx.doi.org/10.1186/s10194-019-0964-5 http://dx.doi.org/10.1016/j.jpsychores.2015.10.005 http://dx.doi.org/10.1017/cjn.2021.68 http://dx.doi.org/10.1097/md.0000000000002646 http://dx.doi.org/10.1080/03007995.2019.1657716 http://dx.doi.org/10.1186/s10194-020-01154-x http://dx.doi.org/10.18553/jmcp.2022.28.11.1272 http://dx.doi.org/10.1136/bmjopen-2018-027251 http://dx.doi.org/10.1016/j.jaad.2016.08.055 http://dx.doi.org/10.1371/journal.pmed.1004238 http://dx.doi.org/10.1136/bmjopen-2019-029147 http://dx.doi.org/10.1186/s10194-022-01476-y http://dx.doi.org/10.1111/head.13157 http://dx.doi.org/10.1136/bmj.k96 http://dx.doi.org/10.1002/pbc.30385 http://dx.doi.org/10.1136/bmjopen-2020-038972 http://dx.doi.org/10.1186/s12913-022-08191-z http://dx.doi.org/10.1186/s10194-023-01558-5 http://dx.doi.org/10.1186/s10194-022-01513-w http://dx.doi.org/10.1002/pds.4723 http://dx.doi.org/10.1017/s2045796021000421 http://dx.doi.org/10.1177/0333102416678382 http://dx.doi.org/10.1111/head.13275 http://dx.doi.org/10.1186/s12940-021-00727-w http://dx.doi.org/10.18553/jmcp.2018.18058 http://dx.doi.org/10.1080/13696998.2020.1814790 http://dx.doi.org/10.1080/13696998.2023.2207413 http://dx.doi.org/10.1007/s40120-021-00304-w http://dx.doi.org/10.1177/0333102419856630 http://dx.doi.org/10.1186/s10194-020-01166-7 http://dx.doi.org/10.1177/0333102419889351 http://dx.doi.org/10.1177/0333102419851855 http://dx.doi.org/10.1177/0333102415590242 http://dx.doi.org/10.1097/md.0000000000003065 http://dx.doi.org/10.1093/qjmed/hcu215 1. GBD 2016 Headache Collaborators. Global, regional, and national burden of migraine and tension-type headache, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2018;17(11):954-976. Headache Classification Committee of the International Headache Society (IHS) The International Classification of Headache Disorders, 3rd edition. Cephalalgia. 2018;38(1):1-211. Natoli JL, Manack A, Dean B, et al. Global prevalence of chronic migraine: a systematic review. Cephalalgia. 2010;30(5):599-609. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. Journal of Clinical Epidemiology. 2021;134:178-189. The Joanna Briggs Institute. Checklist for prevalence studies. 2020. Accessed 8 February 2024. .6. Champaloux SW, Tepper NK, Monsour M, et al. Use of combined hormonal contraceptives among women with migraines and risk of ischemic stroke. Am J Obstet Gynecol. 2017;216(5):489.e481-489.e487. Chen HY, Lin CL, Kao CH. Does Migraine Increase the Risk of Glaucoma?: A Population-Based Cohort Study. Medicine (Baltimore). 2016;95(19):e3670. Cheon DY, Han K, Yang YS, et al. Associations between migraine and major cardiovascular events in type 2 diabetes mellitus. Cardiovasc Diabetol. 2022;21(1):275. Chu CH, Liu CJ, Lin LY, Chen TJ, Wang SJ. Migraine is associated with an increased risk for benign paroxysmal positional vertigo: a nationwide population-based study. J Headache Pain. 2015;16:62. Harnod T, Lin CL, Kao CH. Survival outcome and mortality rate in patients with migraine: a population-based cohort study. J Headache Pain. 2018;19(1):57. Hwang JH, Tsai SJ, Liu TC, Chen YC, Lai JT. Association of Tinnitus and Other Cochlear Disorders With a History of Migraines. JAMA Otolaryngol Head Neck Surg. 2018;144(8):712-717. Korolainen MA, Kurki S, Lassenius MI, et al. Burden of migraine in Finland: health care resource use, sick-leaves and comorbidities in occupational health care. J Headache Pain. 2019;20(1):13. Lau CI, Lin CC, Chen WH, Wang HC, Kao CH. Increased risk of chronic fatigue syndrome in patients with migraine: A retrospective cohort study. J Psychosom Res. 2015;79(6):514-518. Richer L, Wong KO, Martins KJB, et al. Characteristics of Adults with Migraine in Alberta, Canada: A Population-Based Study. Can J Neurol Sci. 2022;49(2):239-248. Yang FC, Lin TY, Chen HJ, et al. Increased Risk of Restless Legs Syndrome in Patients With Migraine: A Nationwide Population-Based Cohort Study. Medicine (Baltimore). 2016;95(5):e2646. Ford JH, Schroeder K, Buse DC, et al. Predicting initiation of preventive migraine medications: exploratory study in a large U.S. medical claims database. Curr Med Res Opin. 2020;36(1):51-61. Roessler T, Zschocke J, Roehrig A, Friedrichs M, Friedel H, Katsarava Z. Administrative prevalence and incidence, characteristics and prescription patterns of patients with migraine in Germany: a retrospective claims data analysis. J Headache Pain. 2020;21(1):85. Shao Q, Rascati KL, Lawson KA, Barner JC, Sonawane KB, Rousseau JF. Real-world opioid use among patients with migraine enrolled in US commercial insurance and risk factors associated with migraine progression. J Manag Care Spec Pharm. 2022;28(11):1272-1281. Wang QR, Lu YY, Su YJ, et al. Migraine and traumatic brain injury: a cohort study in Taiwan. BMJ Open. 2019;9(7):e027251. Egeberg A, Ashina M, Gaist D, Gislason GH, Thyssen JP. Prevalence and risk of migraine in patients with rosacea: A population-based cohort study. J Am Acad Dermatol. 2017;76(3):454-458. Hvitfeldt Fuglsang C, Pedersen L, Schmidt M, Vandenbroucke JP, Bøtker HE, Sørensen HT. Migraine and risk of premature myocardial infarction and stroke among men and women: A Danish population-based cohort study. PLoS Med. 2023;20(6):e1004238. Nyberg J, Gustavsson S, Linde M, et al. Cardiovascular fitness and risk of migraine: a large, prospective population-based study of Swedish young adult men. BMJ Open. 2019;9(8):e029147. Nguyen JL, Munshi K, Peasah SK, et al. Trends in utilization and costs of migraine medications, 2017-2020. J Headache Pain. 2022;23(1):111. Woolley JM, Bonafede MM, Maiese BA, Lenz RA. Migraine Prophylaxis and Acute Treatment Patterns Among Commercially Insured Patients in the United States. Headache. 2017;57(9):1399-1408. Adelborg K, Szépligeti SK, Holland-Bill L, et al. Migraine and risk of cardiovascular diseases: Danish population based matched cohort study. Bmj. 2018;360:k96. Orimoloye HT, Heck JE, Charles A, et al. Maternal migraine and risk of pediatric cancers. Pediatr Blood Cancer. 2023;70(7):e30385. Orlando V, Mucherino S, Monetti VM, Trama U, Menditto E. Treatment patterns and medication adherence among newly diagnosed patients with migraine: a drug utilisation study. BMJ Open. 2020;10(11):e038972. Park S, Han S, Suh HS. The disease burden of migraine patients receiving prophylactic treatments in Korea: a population-based claims database analysis. BMC Health Serv Res. 2022;22(1):902. Peles I, Asla M, Abayev M, et al. The variability of diagnosed migraine epidemiology amongst different municipalities in southern Israel. J Headache Pain. 2023;24(1):25. Peles I, Asla M, Abayev M, et al. Migraine epidemiology and comorbidities in Southern Israel: a clinical database study in a universal health coverage setting. J Headache Pain. 2022;23(1):160. Thomsen RW, Szépligeti SK, Xue F, Pedersen L, Sørensen HT, Ehrenstein V. Patterns of initial migraine treatment in Denmark: A population-based study. Pharmacoepidemiol Drug Saf. 2019;28(3):322-329. Wang H, He H, Miao M, et al. Maternal migraine and the risk of psychiatric disorders in offspring: a population-based cohort study. Epidemiol Psychiatr Sci. 2021;30:e55. Hepp Z, Dodick DW, Varon SF, et al. Persistence and switching patterns of oral migraine prophylactic medications among patients with chronic migraine: A retrospective claims analysis. Cephalalgia. 2017;37(5):470-485. Bonafede M, Sapra S, Shah N, Tepper S, Cappell K, Desai P. Direct and Indirect Healthcare Resource Utilization and Costs Among Migraine Patients in the United States. Headache. 2018;58(5):700-714. Elser H, Morello-Frosch R, Jacobson A, et al. Air pollution, methane super-emitters, and oil and gas wells in Northern California: the relationship with migraine headache prevalence and exacerbation. Environ Health. 2021;20(1):45. Ford JH, Schroeder K, Nyhuis AW, Foster SA, Aurora SK. Cycling Through Migraine Preventive Treatments: Implications for All-Cause Total Direct Costs and Disease-Specific Costs. J Manag Care Spec Pharm. 2019;25(1):46-59. Foster SA, Chen CC, Ding Y, et al. Economic burden and risk factors of migraine disease progression in the US: a retrospective analysis of a commercial payer database. J Med Econ. 2020;23(11):1356-1364. Gaul C, Seidel K, Heuck A, et al. Real-world treatment patterns and healthcare resource utilization among migraine patients: a German claims database analysis. J Med Econ. 2023;26(1):667-678. Gendolla A, Rauer N, Kraemer S, Schwerdtner I, Straube A. Epidemiology, Demographics, Triptan Contraindications, and Prescription Patterns of Patients with Migraine: A German Claims Database Study. Neurol Ther. 2022;11(1):167-183. Hoffman V, Xue F, Ezzy SM, et al. Risk of cardiovascular and cerebrovascular events and mortality in patients with migraine receiving prophylactic treatments: An observational cohort study. Cephalalgia. 2019;39(12):1544-1559. Islamoska S, Hansen Å M, Wang HX, et al. Mid- to late-life migraine diagnoses and risk of dementia: a national register-based follow-up study. J Headache Pain. 2020;21(1):98. Marconi E, Pecchioli S, Nica M, et al. Epidemiology and determinants of chronic migraine: A real-world cohort study, with nested case-control analysis, in primary care in Italy. Cephalalgia. 2020;40(5):461-469. Meyers JL, Davis KL, Lenz RA, Sakai F, Xue F. Treatment patterns and characteristics of patients with migraine in Japan: A retrospective analysis of health insurance claims data. Cephalalgia. 2019;39(12):1518-1534. Pressman A, Jacobson A, Eguilos R, et al. Prevalence of migraine in a diverse community–electronic methods for migraine ascertainment in a large integrated health plan. Cephalalgia. 2016;36(4):325-334. Wu SH, Chuang E, Chuang TY, et al. A Nationwide Population-Based Cohort Study of Migraine and Organic-Psychogenic Erectile Dysfunction. Medicine (Baltimore). 2016;95(10):e3065. Harnod T, Wang YC, Kao CH. High risk of developing subsequent epilepsy in young adults with migraine: a nationwide population-based cohort study in Taiwan. Qjm. 2015;108(6):449-455. Crossref Google Scholar Information & Authors Information Version history V1 Version 1 25 January 2025 Peer review timeline Published Pharmacoepidemiology and Drug Safety Version of Record 2 Dec 2025 Published Copyright This work is licensed under a Non Exclusive No Reuse License. Collection Pharmacoepidemiology and Drug Safety Keywords algorithm migraine systematic literature review validation Authors Affiliations Joan Forns 0000-0002-1066-0358 [email protected] RTI Health Solutions Barcelona View all articles by this author Alicia Abellan RTI Health Solutions Barcelona View all articles by this author Nuria Riera-Guardia 0000-0001-9482-7524 RTI Health Solutions Barcelona View all articles by this author Andrea Margulis 0000-0001-7388-6082 RTI Health Solutions Barcelona View all articles by this author Elena Rivero-Ferrer RTI Health Solutions Barcelona View all articles by this author Metrics & Citations Metrics Article Usage 299 views 163 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Joan Forns, Alicia Abellan, Nuria Riera-Guardia, et al. The Challenges for Pharmacoepidemiologists Identifying Migraine in Electronic Healthcare Data Sources: A Systematic Literature Review. Authorea . 25 January 2025. DOI: https://doi.org/10.22541/au.173781278.87655336/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.173781278.87655336/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ff53806f920aa64',t:'MTc3OTM4MzY4MQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();