The epidemiology, management, and the associated burden of migraine in Australian primary care: a retrospective analysis of electronic health record data.

This was a real-world retrospective, secondary use of data, cross-sectional and longitudinal descriptive study within Australian general practice. Data were captured from the MedicalDirector national clinical practice management software from a 14-year index period (1 January 2010 until 28 February 2024). As of August 2023, the database consisted of approximately 1,300 users (primary care practitioners (90–95%) and specialists) who consented for their de-identified and aggregated patient record data to be used for research. These consenting practitioners represented 3,524,138 medical records. The study protocol was reviewed and approved for secondary use of data by the Independent Ethics Committee Bellberry Limited (approval number 2023-11-1427). The total eligible population included records with registered demographic information (age and sex) at the time of data extraction and one or more general practice encounters since the start of the index period (1 January 2010 to 28 February 2024). For demographic endpoints, summaries were conducted in the migraine set which included all patients with a diagnosis of migraine (migraine, common migraine, classical migraine, or hemiplegic migraine). The index date was the date in which the first diagnosis of migraine appeared in the database within the index period. For secondary endpoints for assessment of treatment patterns, patients with a minimum of 2-year follow-up were included (migraine set with follow-up), to examine initial as well as subsequent treatments. The point prevalence of diagnosed migraine on 28 February 2024 was estimated overall and stratified by age at diagnosis (or age at first visit for undiagnosed patients) and sex. The point prevalence represented people who were alive, who had or have had migraine in the past, and have had at least one clinical encounter between1 January 2010 until 28 February 2024. Clinical encounters occurring during this period did not have to be migraine related to qualify as a prevalent diagnosed migraine record. The incidence of diagnosed migraine was estimated based on patients with new onset migraine during the index period (1 January 2010 to 28 February 2024). People with a diagnosis of migraine prior to the index period were excluded. Incidence rates (unadjusted), stratified by age and sex, were calculated and adjusted estimates were calculated based on modelling. The following variables were extracted from the MedicalDirector national clinical practice management software: age at diagnosis, sex, comorbid conditions, geography (state of practice), socioeconomic status (of practice), remoteness (of practice), diagnosis of migraine, referral to neurologist, migraine medications (acute, preventive, antidepressants and opioids), date of diagnosis (for seasonality assessment) and date of acute migraine treatment initiation (for seasonality assessment). The diagnoses in MedicalDirector were recorded using the Doctor Command Language (DOCLE) classification system, while prescriptions were mapped to Anatomic Therapeutic Codes (ATC). All eligible patients were to be included. A priori, the precision of the estimated prevalence was evaluated. Based on a prevalence of approximately 1% and an approximate total pool of around 3 million patients, the 95% CI around the prevalence of diagnosed migraine was expected to be at most ± 0.02% from the estimate. The data were analysed descriptively. Models were fitted for the primary endpoints of incidence and prevalence to provide adjusted estimates. Subgrouping factors of interest were added individually in a univariate model; a multivariate model was then built, starting with age and sex, and then adding other factors in the order of their significance in univariate modelling. Covariates with p  < 0.05, when added to those already in the model, were considered statistically significant and included in the final model. Sociodemographic factors were categorised according to features relating to the patient and the clinical practice. Patient age at diagnosis (to 4 years; 5-to- 9 years; 10-to-14 years; etc.; 70-to-79 years, and > 80 years), and sex (male and female), were extracted from each record. The location of the clinical practice was categorised by state (Australian Capital Territory (ACT), New South Wales (NSW), Northern Territory (NT), Queensland (QLD), South Australia (SA), Tasmania (TAS), Victoria (VIC), or Western Australia (WA)), and by remoteness (major city, inner regional, outer regional, remote and very remote). Socioeconomic status was also determined by practice location according to the Index of Relative Socioeconomic Disadvantage (IRSD) Socioeconomic Indexes for Areas (SEIFA) quintiles 1–5, where quintile 1 is the most disadvantaged and quintile 5 is the least disadvantaged. Adjusted prevalence and incidence estimates within each subgroup were reported using a modelling approach. Females and NSW were the reference groups in the presentation of modelling results for rate ratios. For other subgrouping factors, the group with the lowest values were generally used as the reference group. Statistical analyses were performed using R version 3.6.3 (R Project, Vienna, Austria, https://www.R-project.org/ ).

There were 3,524,138 cases in the MD Clinical ® Electronic Medical Record database screened for inclusion into this study (Fig.  1 ). Patients without a visit recorded during the index period, or those with age or sex missing were excluded, leaving a total of 3,448,369 cases, around 98% of the total database, that qualified as the total eligible population. Of these, 37,579 cases were included in the migraine set (i.e. those with a clinical diagnosis of headache– migraine, common migraine, classical migraine, or hemiplegic migraine). Of the migraine set, 30,417 patients were included in the migraine set with follow-up (i.e. a minimum of 2 years follow-up). Fig. 1 *Migraine set is patients from the eligible population with a visit or diagnosis with a migraine related Doctor Command Language (DOCLE) code between 1 January 2010 and 28 February 2024. **Migraine set with follow-up is patients from the migraine set where their first visit or diagnosis of migraine is at least 730 days (2 years) before 28 February 2024 *Migraine set is patients from the eligible population with a visit or diagnosis with a migraine related Doctor Command Language (DOCLE) code between 1 January 2010 and 28 February 2024. **Migraine set with follow-up is patients from the migraine set where their first visit or diagnosis of migraine is at least 730 days (2 years) before 28 February 2024 Demographics of the analysis sets are reported in Table  1 . Of the 37,579 patients in the migraine set, 74.15% were female. The median age was 38 years (interquartile range (IQR) 27 to 50), and the mean age was 39.24 years (standard deviation (SD) 15.63). The majority of patients (94.03%) were aged ≥ 18 years. Table 1 Demographics and other baseline characteristics Demographic Migraine Set ( n  = 37,579) Migraine Set with Follow up ( n  = 30,417) n % n % Number of cases diagnosed 37,579 30,417 Age at diagnosis, y Mean 39.24 38.93 SD 15.63 15.54 Median 38 37 Q1 27 27 Q3 50 49 IQR 23 22 Age group, n Paediatric (4-17y) 2,230 5.93% 1,825 6.00% Adult (≥ 18y) 35,334 94.03% 28,580 93.96% Age category, n 0-4y 26 0.07% 24 0.08% 5-9y 237 0.63% 198 0.65% 10-14y 835 2.22% 680 2.24% 15-19y 2,172 5.78% 1,779 5.85% 20-24y 3,751 9.98% 3,139 10.32% 25-29y 4,506 11.99% 3,731 12.27% 30-34y 4,445 11.83% 3,657 12.02% 35-39y 4,337 11.54% 3,479 11.44% 40-44y 4,057 10.80% 3,263 11.42% 45-49y 3,659 9.74% 2,964 9.74% 50-54y 3,093 8.23% 2,462 8.09% 55-59y 2,253 6.00% 1,787 5.88% 60-64y 1,622 4.32% 1,264 4.16% 65-69y 1,131 3.01% 877 2.88% 70-74y 697 1.85% 527 1.73% 75-79y 401 1.07% 303 1.00% 80 + y 355 0.94% 283 0.93% Length of diagnosis, y Median 1.8 2.74 Q1 0.23 0.54 Q3 4.73 5.58 IQR 4.5 5.04 Sex, n Male 9,713 25.85% 7,991 26.27% Female 27,866 74.15% 22,426 73.73% Geography of practice, n ACT 450 1.2% 339 1.12% NSW 13,254 35.37% 11,029 36.38% NT 489 1.3% 342 1.13% QLD 8,409 22.44% 6,661 21.97% SA 1,775 4.74% 1,223 4.03% TAS 621 1.66% 468 1.54% VIC 11,404 30.43% 9,441 31.14% WA 1,075 2.87% 813 2.68% Remoteness, n Major cities of Australia 26,842 71.62% 22,252 73.4% Inner regional Australia 6,755 18.02% 5,228 17.24% Outer regional Australia 3,199 8.54% 2,357 7.77% Remote Australia 614 1.64% 428 1.41% Very remote Australia 69 0.18% 53 0.17% Socioeconomic status, n 1 (most disadvantaged) 8,548 23.11% 6,795 22.67% 2 4,934 13.34% 3,875 12.93% 3 7,655 20.7% 6,072 20.26% 4 8,253 22.31% 6,858 22.88% 5 (least disadvantaged) 7,599 20.54% 6,375 21.27% Time period, n Before 1 July 2018 17,556 46.72% 17,556 57.72% On or after 1 July 2018 20,023 53.28% 12,861 42.28% n number of people, Length of diagnosis Difference between the first instance of migraine being recorded in either reason for visit or diagnosis, and the final visit in the study period for any reason, Geography of practice This is determined using the postcode of the practice. The practice postcode is mapped against postcode data from the 2016 census. For the 22 postcodes with multiple states (Example: 872, sitting in NT, WA and SA), the state with the most "localities" was selected, Remoteness Determined using census data from 2016, Socioeconomic status Determined using census data from 2016The scores are split into quintiles as per https://www.abs.gov.au/ausstats/[email protected]/Lookup/by Subject/2033.0.55.001~2016~Main Features~SEIFA Measures~14 Demographics and other baseline characteristics Paediatric (4-17y) n number of people, Length of diagnosis Difference between the first instance of migraine being recorded in either reason for visit or diagnosis, and the final visit in the study period for any reason, Geography of practice This is determined using the postcode of the practice. The practice postcode is mapped against postcode data from the 2016 census. For the 22 postcodes with multiple states (Example: 872, sitting in NT, WA and SA), the state with the most "localities" was selected, Remoteness Determined using census data from 2016, Socioeconomic status Determined using census data from 2016The scores are split into quintiles as per https://www.abs.gov.au/ausstats/[email protected]/Lookup/by Subject/2033.0.55.001~2016~Main Features~SEIFA Measures~14 Baseline demographics in the migraine with follow-up set were broadly similar to the migraine set, apart from a longer median length of diagnosis in the follow-up set. The median length of diagnosis was 1.8 years (IQR 0.23 to 4.73) in the migraine set and 2.74 years (IQR 0.54 to 5.58) in the migraine set with follow-up. Comorbidities were reported significantly more frequently in the migraine set compared to the total eligible population (Table  2 ). Upper respiratory tract infection (URTI) ranked among the most frequent comorbidities, reported in 27.29% and 9.05% of the migraine set and the total eligible population, respectively. This was followed by gastroesophageal reflux disease (GORD) (18.98% vs. 3.73%), headache (13.75% vs. 1.08%) and urinary tract infection (UTI) (12.81% vs. 2.86%). Depression occurred in 12.70% vs. 2.34% of the migraine set and the total eligible population respectively. Anxiety occurred in 12.42% vs. 2.10% respectively. Hypertension ranked as the second most common comorbidity in the total eligible population, whereas it ranked 7th in the migraine set. All other comorbidities had a similar order of appearance. In a breakdown of cardiovascular comorbidities, hypertension was reported in 12.36% of migraine set patients, and ischaemic heart disease (IHD) occurred in 1.45% of patients. Table 2 Common comorbidities in people in Australia diagnosed with migraine Comorbidities Migraine Set Total Eligible Population Odds Ratio 95% CI Lower 95% CI Upper p -value ( n  = 37,579) ( n  = 3,448,369) n % n % URTI 10,257 27.29 311,994 9.05 3.77 3.69 3.86 < 0.001 GORD 7,131 18.98 128,661 3.73 6.04 5.89 6.2 < 0.001 Headache 5,168 13.75 37,327 1.08 14.57 14.13 15.03 < 0.001 UTI 4,812 12.81 98,568 2.86 4.99 4.84 5.15 < 0.001 Depression 4,771 12.7 80,800 2.34 6.06 5.87 6.25 < 0.001 Anxiety 4,666 12.42 72,356 2.1 6.61 6.41 6.83 < 0.001 Hypertension 4,646 12.36 165,032 4.79 2.81 2.72 2.9 < 0.001 Back pain 4,368 11.62 69,325 2.01 6.41 6.21 6.62 < 0.001 Asthma 3,818 10.16 87,428 2.54 4.35 4.2 4.5 < 0.001 Dermatitis 3,817 10.16 98,327 2.85 3.85 3.72 3.99 < 0.001 Insomnia 3,609 9.6 58,450 1.7 6.16 5.95 6.38 < 0.001 Sinusitis 3,574 9.51 79,383 2.3 4.46 4.31 4.62 < 0.001 Fibromyalgia 526 1.4 3,899 0.11 12.54 11.44 13.74 < 0.001 Endometriosis 381 1.01 3,502 0.1 10.08 9.06 11.2 < 0.001 Common comorbidities in people in Australia diagnosed with migraine For prevalence analysis, of the cases that had a clinical practice encounter during the active record period, there were 37,579 eligible migraine prevalent cases (i.e. people who were alive and who had or have had migraine in the past). The overall adjusted point prevalence of diagnosed migraine was estimated as 7.02 per 1,000 persons (95% CI 6.79 to 7.25), which equates to 0.702% after adjusting for the interaction of age and sex (Table  3 ). Overall point prevalence was highest in the 40-to-44-year age range (adjusted 14.67 per 1,000 persons; 95% CI 13.95 to 15.44). Females had the highest prevalence in the 40-to-44-year age range (adjusted 25.31 per 1,000 persons; 95% CI 24.09 to 26.59) compared to males (adjusted 8.47per 1,000 persons; 95% CI 7.81 to 9.18), however peak prevalence in males was in the 35-to-39 year age range (adjusted 9.07 per 1,000 persons; 95% CI 8.43 to 9.77) (Table  3 ). Prevalent cases by age and sex are presented in Fig.  2 . Table 3 The estimated point prevalence of diagnosed migraine in the Australian population Stratification Metric Overall 0–9 years 10–14 years 15–19 years 20–24 years 25–29 years 30–34 years 35–39 years 40–44 years 45–49 years 50–54 years 55–59 years 60–64 years 65–69 years 70–74 years 75–79 years 80 + years Overall Point Prevalence (per 1,000 persons), adjusted 7.02 0.49 5.18 10.02 11.72 12.23 13.02 14.28 14.67 13.8 12.82 9.68 7.71 5.62 4.61 3.82 2.57 Overall 95% CI, adjusted (6.79, 7.25) (0.43, 0.57) (4.75, 5.65) (9.42, 10.65) (11.13, 12.34) (11.65, 12.83) (12.42, 13.65) (13.61, 14.98) (13.95, 15.44) (13.08, 14.57) (12.13, 13.54) (9.09, 10.32) (7.17, 8.28) (5.14, 6.14) (4.14, 5.13) (3.34, 4.36) (2.24, 2.95) Female Point Prevalence (per 1,000 persons), adjusted 10.61 0.5 6.18 13.89 17.17 17.98 19.24 22.4 25.31 25.2 21.08 16.13 13.13 9.77 7.66 5.78 3.45 Female 95% CI, adjusted (10.25, 10.98) (0.41, 0.62) (5.53, 6.9) (13.02, 14.82) (16.3, 18.08) (17.12, 18.88) (18.32, 20.21) (21.32, 23.52) (24.09, 26.59) (23.96, 26.5) (19.97, 22.26) (15.15, 17.16) (12.24, 14.09) (8.96, 10.64) (6.88, 8.52) (5.02, 6.66) (2.98, 4.0) Male Point Prevalence (per 1,000 persons), adjusted 4.64 0.48 4.34 7.22 7.99 8.3 8.79 9.07 8.47 7.52 7.77 5.8 4.51 3.22 2.77 2.52 1.91 Male 95% CI, adjusted (4.44, 4.83) (0.4, 0.59) (3.82, 4.94) (6.55, 7.95) (7.37, 8.66) (7.71, 8.93) (8.18, 9.45) (8.43, 9.77) (7.81, 9.18) (6.88, 8.21) (7.11, 8.49) (5.22, 6.44) (3.99, 5.1) (2.77, 3.76) (2.31, 3.32) (2.02, 3.15) (1.52, 2.41) Any cell with data from < 11 patients or coming from < 3 practices will be hidden CI confidence interval, Prevalent Cases  Is any patient with a migraine diagnosis recorded between 1 January 2010 and 28 February 2024 for the Consented MD Clinical ® GPs The estimated point prevalence of diagnosed migraine in the Australian population Any cell with data from < 11 patients or coming from < 3 practices will be hidden CI confidence interval, Prevalent Cases  Is any patient with a migraine diagnosis recorded between 1 January 2010 and 28 February 2024 for the Consented MD Clinical ® GPs Fig. 2 The estimated point prevalence of diagnosed migraine in the Australian population The estimated point prevalence of diagnosed migraine in the Australian population Among the states and territories, point prevalence was estimated to be highest in NSW (adjusted 8.33 per 1,000 persons (95% CI 8.02 to 8.64)) (Supplementary Table 1). Point prevalence was also estimated to be higher in the IRSD quintile 1 (most disadvantaged) compared with quintile 5 (least disadvantaged) (adjusted 8.93 per 1,000 persons; 95% CI 8.6 to 9.28 versus 5.89 per 1,000 persons; 95% CI 5.64 to 6.16) (Supplementary Table 2). For incidence analysis, 36,462 migraine incident cases were extracted (i.e. those patients in the migraine set who were diagnosed during the index period). The overall incidence rate (IR) was estimated as 3.48 per 1,000 person-years (95% CI 3.38 to 3.58) after adjusting for the interaction of age and sex (Table  4 ). The IR was estimated to be highest in the 20-to-24-year age range (adjusted IR 8.98 per 1,000 person-years; 95% CI 8.59 to 9.38). The estimated IR per 1,000 person-years was 5.14 (95% CI 4.99 to 5.30) for females and 2.35 (95% CI 2.27 to 2.44)for males (adjusted IRs). Table 4 The estimated incidence of diagnosed migraine in the Australian population Stratification Metric Overall 0–9 years 10–14 years 15–19 years 20–24 years 25–29 years 30–34 years 35–39 years 40–44 years 45–49 years 50–54 years 55–59 years 60–64 years 65–69 years 70–74 years 75–79 years 80+ years Overall Incidence Rate (per 1,000 person-years), adjusted 3.48 0.25 2.54 5.83 8.98 8.93 8.1 7.52 6.67 5.7 5.05 3.8 2.88 2.38 2.01 1.7 1.48 Overall 95% CI, adjusted (3.38, 3.58) (0.22, 0.28) (2.36, 2.73) (5.53, 6.14) (8.59, 9.38) (8.58, 9.31) (7.78, 8.44) (7.21, 7.83) (6.38, 6.97) (5.44, 5.97) (4.82, 5.3) (3.6, 4.02) (2.7, 3.07) (2.21, 2.56) (1.84, 2.19) (1.52, 1.91) (1.32, 1.67) Female Incidence Rate (per 1,000 person-years), adjusted 5.14 0.25 2.98 8.12 12.51 12.33 11.57 11.27 11.07 10.06 8.6 6.56 5.12 4.13 3.07 2.52 1.88 Female 95% CI, adjusted (4.99, 5.3) (0.21, 0.3) (2.72, 3.27) (7.68, 8.58) (11.97, 13.08) (11.83, 12.86) (11.09, 12.06) (10.81, 11.76) (10.61, 11.55) (9.63, 10.51) (8.21, 9.02) (6.22, 6.92) (4.82, 5.44) (3.84, 4.44) (2.8, 3.37) (2.24, 2.85) (1.65, 2.14) Male Incidence Rate (per 1,000 person-years), adjusted 2.35 0.24 2.17 4.18 6.44 6.47 5.68 5.01 4.02 3.23 2.97 2.2 1.62 1.37 1.32 1.14 1.17 Male 95% CI, adjusted (2.27, 2.44) (0.21, 0.29) (1.94, 2.41) (3.85, 4.54) (6.02, 6.88) (6.08, 6.89) (5.34, 6.04) (4.7, 5.34) (3.75, 4.31) (3.0, 3.49) (2.75, 3.21) (2.01, 2.41) (1.45, 1.8) (1.2, 1.55) (1.14, 1.52) (0.95, 1.39) (0.96, 1.43) Any cell with data from < 11 patients or coming from < 3 practices will be hidden CI confidence interval, Incident Cases  Is any patient with a migraine diagnosis recorded between 1 January 2010 and 28 February 2024 for the Consented MD Clinical ® GPs, Incidence rate (per 1,000 person-years) Represents the proportion of incidence cases per 1,000 patient years at risk in the population (Incident cases * 1000/Years at Risk) The estimated incidence of diagnosed migraine in the Australian population Any cell with data from < 11 patients or coming from < 3 practices will be hidden CI confidence interval, Incident Cases  Is any patient with a migraine diagnosis recorded between 1 January 2010 and 28 February 2024 for the Consented MD Clinical ® GPs, Incidence rate (per 1,000 person-years) Represents the proportion of incidence cases per 1,000 patient years at risk in the population (Incident cases * 1000/Years at Risk) The incidence in all older age groups was significantly higher ( p  < 0.01) than the 0–9-year-old age group for both the unadjusted and adjusted analyses (Table  5 ). The IRRs in males was significantly lower than in females (approximately 60% lower in the unadjusted analysis (IRR of 0.405, 95% CI 0.382 to 0.428; p  < 0.01) and 54% lower in the adjusted analysis (IRR 0.457, 95% CI 0.442 to 0.474; p  < 0.01)). The IRRs were significantly lower for Victoria and the Other States group compared to NSW in both the adjusted and unadjusted analyses ( p  < 0.01); however, for Queensland and WA the IRRs were not significantly different to NSW. The IRRs were significantly higher in all other IRSD quintiles compared to the IRSD5 quintile (i.e. the least disadvantaged) in both adjusted and unadjusted analyses. For remoteness, in the unadjusted analyses the IRRs in both outer regional Australia and remote Australia were significantly higher compared to the major cities; however, in the adjusted analysis where age, sex, state and IRSD were included in the model, the IRR for remote Australia was no longer significantly different compared to major cities. IRRs in both inner and outer regional Australia were significantly lower compared to major cities in the adjusted analysis. Table 5 Incidence rate ratios (total eligible population) Parameter Unadjusted analysis Adjusted analysis Estimate (SE) 95% CI P -value Estimate (SE) 95% CI P  - value Age (years) 0–9 Reference Reference 10–14 10.48 (0.071) 10.341, 10.619 < 0.01 10.238 (0.072) 8.896, 11.782 < 0.01 15–19 26.12 (0.066) 25.991, 26.249 < 0.01 23.473 (0.067) 20.593, 26.755 < 0.01 20–24 40.68 (0.064) 40.554, 40.806 < 0.01 36.165 (0.065) 31.836, 41.084 < 0.01 25–29 40 (0.064) 39.875, 40.125 < 0.01 36 (0.065) 31.719, 40.859 < 0.01 30–34 36.16 (0.064) 36.035, 36.285 < 0.01 32.648 (0.065) 28.766, 37.055 < 0.01 35–39 33.88 (0.064) 33.755, 34.005 < 0.01 30.286 (0.065) 26.677, 34.382 < 0.01 40–44 31.36 (0.064) 31.234, 31.486 < 0.01 26.869 (0.065) 23.649, 30.528 < 0.01 45–49 27.8 (0.064) 27.674, 27.926 < 0.01 22.975 (0.066) 20.201, 26.131 < 0.01 50–54 23.92 (0.065) 23.793, 24.047 < 0.01 20.368 (0.066) 17.895, 23.182 < 0.01 55–59 18.12 (0.066) 17.991, 18.249 < 0.01 15.319 (0.067) 13.422, 17.484 < 0.01 60–64 13.84 (0.067) 13.709, 13.971 < 0.01 11.6 (0.069) 10.124, 13.292 < 0.01 65–69 11.2 (0.069) 11.065, 11.335 < 0.01 9.57 (0.072) 8.31, 11.021 < 0.01 70–74 8.88 (0.073) 8.736, 9.024 < 0.01 8.101 (0.076) 6.981, 9.4 < 0.01 75–79 7.52 (0.081) 7.362, 7.678 < 0.01 6.85 (0.085) 5.803, 8.087 < 0.01 80+ 6.32 (0.082) 6.159, 6.481 < 0.01 5.982 (0.086) 5.05, 7.086 < 0.01 Sex Female Reference Reference Male 0.405 (0.012) 0.382, 0.428 < 0.01 0.457 (0.018) 0.442, 0.474 < 0.01 State NSW Reference Reference QLD 1 (0.014) 0.972, 1.028 1.00 1.008 (0.015) 0.979, 1.037 0.59 VIC 0.864 (0.013) 0.838, 0.889 < 0.01 0.862 (0.013) 0.839, 0.884 < 0.01 WA 0.975 (0.032) 0.912, 1.037 0.42 1.051 (0.032) 0.986, 1.12 0.13 Other States 0.945 (0.02) 0.907, 0.984 < 0.01 0.83 (0.021) 0.796, 0.865 < 0.01 IRSD IRSD5 (least disadvantaged) Reference Reference IRSD1 (most disadvantaged) 1.413 (0.016) 1.381, 1.444 < 0.01 1.494 (0.018) 1.444, 1.547 < 0.01 IRSD2 1.289 (0.019) 1.252, 1.325 < 0.01 1.319 (0.02) 1.268, 1.371 < 0.01 IRSD3 1.265 (0.017) 1.233, 1.297 < 0.01 1.237 (0.018) 1.195, 1.28 < 0.01 IRSD4 1.289 (0.016) 1.257, 1.32 < 0.01 1.198 (0.016) 1.16, 1.237 < 0.01 Remoteness Major Cities of Australia Reference Reference Inner Regional Australia 0.994 (0.014) 0.967, 1.021 0.67 0.952 (0.015) 0.924, 0.982 < 0.01 Outer Regional Australia 1.066 (0.019) 1.028, 1.103 < 0.01 0.91 (0.022) 0.871, 0.951 < 0.01 Remote Australia 1.193 (0.039) 1.117, 1.269 < 0.01 1.037 (0.04) 0.959, 1.122 0.36 , CI confidence interval, SE standard error, IRSD Index of Relative Socioeconomic Disadvantage, Each age, sex, state, IRSD and remoteness group was compared to the respective reference group. Geography of practice This is determined using the postcode of the practice. The practice postcode is mapped against postcode data from the 2016 census. For the 22 postcodes with multiple states (Example: 872, sitting in NT, WA and SA), the state with the most "localities" was selected. Other States Is the sum of ACT, NT, SA and TAS, due to low sample sizes. Remoteness Is determined using census data from 2016. Socioeconomic status Is determined using census data from 2016. The scores are split into quintiles as per https://www.abs.gov.au/ausstats/[email protected]/Lookup/by Subject/2033.0.55.001~2016~Main Features~SEIFA Measures~14 Incidence rate ratios (total eligible population) , CI confidence interval, SE standard error, IRSD Index of Relative Socioeconomic Disadvantage, Each age, sex, state, IRSD and remoteness group was compared to the respective reference group. Geography of practice This is determined using the postcode of the practice. The practice postcode is mapped against postcode data from the 2016 census. For the 22 postcodes with multiple states (Example: 872, sitting in NT, WA and SA), the state with the most "localities" was selected. Other States Is the sum of ACT, NT, SA and TAS, due to low sample sizes. Remoteness Is determined using census data from 2016. Socioeconomic status Is determined using census data from 2016. The scores are split into quintiles as per https://www.abs.gov.au/ausstats/[email protected]/Lookup/by Subject/2033.0.55.001~2016~Main Features~SEIFA Measures~14 Of patients in the migraine set with follow-up, 32.03% had been referred to a physiotherapist and 18.64% had been referred to a neurologist; however, the reason for referral was not available (Table  6 ). Referrals to pain specialists and dentists were around 4%. Table 6 Referral pathways Category n % Physiotherapist 9,742 32.03% Neurologist 5,671 18.64% Pain Specialist 1,343 4.42% Dental 1,314 4.32% n number of people Referral pathways Overall, 81.08% of patients were prescribed an acute migraine medication. The top five most common acute medications were: metoclopramide (28.81%), sumatriptan (25.81%), rizatriptan (23.37%), meloxicam (20.83%) and prochlorperazine (19.46%) (Supplementary Table 3). In the migraine set with follow-up, 51.06% of patients had a triptan prescribed, with sumatriptan (25.62%), rizatriptan (23.37%) and eletriptan (9.63%) being the most prescribed triptans (Supplementary Table 4). Opioids or opioid combination medications were prescribed to 54.75% of patients (Supplementary Table 5). Codeine plus paracetamol was the most prescribed opioid combination medication (prescribed to 47.30%). The most prescribed opioids were oxycodone (14.71%) and tramadol (9.15%). Preventive medications were prescribed to 46.59% of patients (Supplementary Table 6). The top 5 most prescribed preventive medications were amitriptyline (17.23%), propranolol (12.82%), pregabalin (12.42%) and pizotifen (10.99%). Specified antidepressant medications were prescribed to 22.4% of patients (Supplementary Table 7).

Migraine is a common neurological condition, typically characterised by recurrent attacks of moderate or severe throbbing headache, accompanied by photophobia and phonophobia, and may be associated with nausea, vomiting, and the inability to move due to worsening of pain [ 1 , 2 ]. Migraine ranks second in the ten most disabling disorders, and accounts for more disability than all other neurologic conditions combined [ 3 ]. It may also be associated with a higher risk of other conditions such as stroke, heart disease and mental health conditions (including anxiety and depression), adding to the symptom burden in patients with migraine [ 4 , 5 ]. Data suggest that more than 40% of people with migraine do not seek medical care [ 6 – 8 ]. Of those who do seek medical care, only one-quarter receive an accurate diagnosis; of those with a correct diagnosis, less than half (44.4%) receive both acute and preventive pharmacologic treatments [ 7 ]. If untreated or sub optimally treated, migraine attacks can result in significant disability and reduced quality of life [ 1 , 2 , 6 ]. Effective migraine treatment strategies are now available [ 9 , 10 ]. Newer agents for acute management include: calcitonin gene related peptide (CGRP) receptor antagonists and 5-hydroxytryptamine 1 F (5-HT1F) receptor agonists [ 11 ]. Treatments for preventive management include: injectable anti-CGRP ligand or receptor monoclonal antibodies (mAbs) and oral CGRP receptor antagonists [ 10 , 12 ]. Rimegepant, a CGRP receptor antagonist, is the only treatment that has dual indications for acute treatment and prevention of episodic migraine [ 11 , 13 ]. However, the uptake of these newer agents may be lower than expected, as general practitioners do not routinely initiate them due to reimbursement requirements. The current global migraine prevalence is estimated to be between 14 and 15% [ 2 , 14 , 15 ]. The global incidence of migraine has increased by 40.1% from 1990 to 2019, with a reported incidence of 87.6 million people in 2019 [ 16 ]. Sex differences in migraine have been observed, with migraine being three times more prevalent in females than males [ 2 , 6 , 9 , 15 ]. Migraine onset occurs before the age of 35 years in 75% of people and peak prevalence has been reported in people aged between 35 and 39 years [ 14 ]. Currently, there is limited epidemiological data available on the prevalence and incidence of migraine in the Australian population. Published migraine epidemiological data vary significantly. The Global Burden of Disease Study (2016) reported an Australian migraine prevalence of 20.6% and the Blue Mountains Eye Study (1998) estimated a 17% lifetime prevalence in all people over 49 years [ 3 , 17 , 18 ]. However, a migraine sub study within Australian general practice (2007) (Bettering the Evaluation and Care of Health (BEACH)) reported a diagnosed migraine prevalence of 11.5% [ 19 ], and the 2021-22 Australian Bureau of Statistics (ABS) National Health Survey found that the prevalence of self-reported migraine in Australia was only 6.6% [ 20 ]. To help address this data gap, we conducted a retrospective longitudinal descriptive study of patients with medically diagnosed migraine within Australian general practice. The primary objective of this study was to describe the epidemiology (prevalence and incidence) of diagnosed migraine within the Australian general population. Secondary objectives included investigating the proportion of diagnosed migraine patients who were referred to a neurologist, and understanding medication treatment patterns.

Migraine prevalence is variable in the literature. The current study found the prevalence of diagnosed migraine in the Australian primary care setting to be 7.02 per 1000 persons, or 0.7%. This is markedly lower than rates previously reported, which ranged from 11.3 to 20.6% [ 3 , 17 – 19 ]. The reported point prevalence of 0.7% in this study indicates the proportion of individuals diagnosed with migraine across the time period of the study, adjusted for age and sex. Although the index period covered over 20 years, on average patients in the total eligible population were followed for about 2 years. This contrasts with prevalence estimates such as the BEACH study (11.5%), which represents lifetime prevalence. Consequently, the prevalence observed in this study might underestimate the burden compared to broader prevalence measures, contributing to the lower figures observed. The main difference between outcomes seen in this study and other research may be due to differences in migraine case definition, i.e. diagnosed migraine versus self-reported migraine. Additionally, literature indicates that the diagnostic accuracy of chronic migraine among people consulting a healthcare professional is approximately 24.6% [ 7 ]. A migraine diagnosis could also be listed under a different code, such as headache, even if it is diagnosed migraine. Another factor which may contribute to differences is variation in treatment seeking behaviour across regions, resulting in under-recognition and under-reporting [ 6 – 8 ]. This under-recognition and failure to seek treatment by people with migraine, as well as under-recognition of the condition by clinicians, can lead to delayed diagnosis, and ineffective or inappropriate treatment before assessment by a neurologist. Furthermore, methodological differences in data collection sources (e.g. population-based epidemiological studies versus surveys versus medical records), and migraine diagnosis criteria may account in part for differences seen in outcomes. Consistent with global data [ 2 , 6 , 9 , 15 ], this study found that migraine was approximately three times more prevalent in females than males. Migraine prevalence in this study was highest in the 40-to-44-year age range, which is slightly higher than the global estimates which show the highest prevalence in people aged 35-to-39 years [ 14 ]; suggesting that there may be delayed recognition and treatment in Australia. The present study is the first to estimate diagnosed migraine IR in Australia. The incidence of diagnosed migraine was found to be 0.348% (3.48 per 1,000 person-years). This IR is comparable to a UK general practice research database analysis reporting a migraine incidence of 3.69 per 1,000 person-years [95% CI 3.66 to 3.73][ 21 ], and a German claims database IR of 0.267% [ 22 ]. However, the IR found in this study was lower than the Global Burden of Disease 2016 study (1.7019%), which includes the only other Australian incidence estimate of migraine we are aware of [ 17 , 18 ]. Incidence rates in both inner and outer regional Australia were significantly lower compared to major cities in the adjusted analysis, which could suggest an access to services issue. The most common migraine comorbidities identified in this study (URTI, GORD, headache, UTI, depression, and anxiety) were reported more frequently in migraine patients than in the total eligible population; this is similar to previously reported findings on comorbid and co-occurring conditions in migraine [ 23 ]. In this study, depression and anxiety were reported over four times more frequently in migraine patients than the total eligible population (12.70% and 12.42% vs. 2.34% and 2.10%), which is similar to findings from previous research [ 23 ]. High use of opioids and opioid combinations (54.75%) was reported in the migraine group. Although opioid use may be due to other comorbidities, opioids are generally not indicated for use in migraine management [ 24 ], and triptans should be limited to < 10 days/month according to migraine management guidelines [ 25 ]. The high rate of opioid and opioid combination use is a critical finding; however, the absence of data regarding prescription indications (migraine vs. other conditions) constrains its interpretability. Triptan use (51.06%) was expected to be higher since these patients visited a general practitioner specifically to discuss migraine at some point and could be seeking treatment beyond over-the-counter (OTC) medicines alone. General practitioners may potentially be recommending NSAIDs or aspirin as first-line treatment before triptans, in accordance with guidelines, due to perceived lower risk [ 24 , 25 ]. The low rate of triptan use and high rate of opiate prescription may reflect a need for further education on acute migraine treatment. For instance, the reluctance to use triptans could be due to disproportionate concern around the risk of serotonin syndrome, a rare but potentially serious event with concomitant triptan and selective serotonin reuptake inhibitor (SSRI) or selective norepinephrine reuptake inhibitor antidepressants [ 26 ]. This prescribing pattern could suggest general practitioners are more comfortable prescribing opioids than triptans, given depression is a common comorbidity in this group. Nevertheless, almost half of patients (46.59%) in this study were prescribed a migraine prevention therapy, suggestive of a group of patients who did not respond to acute therapies or those with more frequent migraines. Alternatively, increased prevention therapy prescribing may relate to more choice of preventive therapies, for instance beta-adrenergic receptor antagonists may be used in patients concurrently receiving SSRIs. Among those with migraine, 32.03% had been referred to a physiotherapist and 18.64% had been referred to a neurologist. Medical data for the reason for referral was not extracted, however this referral pattern could suggest symptoms were viewed as a musculoskeletal condition rather than neurological, indicating a lack of awareness that neck pain can represent migraine. This may explain the use of opioids rather than migraine specific therapies such as triptans and the lower diagnosis rates observed, as the condition is not appreciated as migraine. Observational data indicate that while migraine patients frequently report experiencing neck pain, the causal link remains unclear [ 27 ]. This analysis shares the inherent limitations of database analyses utilising secondary use of data (data collected for a different purpose). However, a comparison of the characteristics of the patient cohort represented in the MedicalDirector clinician-consented dataset to data available from the ABS, found similar proportions of patients by location, SEIFA index, rurality index (Rural, Remote and Metropolitan Area) and chronic disease prevalence. In our dataset, few cases were obtained from the smaller states (NT, ACT, SA, TAS or WA) and were combined for analysis. Furthermore, the data governance framework for this dataset precludes the identification and linkage of patients across practices, as well as patients within the same practice who consult different clinicians. When a patient transfers from one clinic to another, the patient is added as a new record to the dataset, meaning that a single patient may have multiple records. Based on a cross-sectional survey conducted in 2013 involving 2477 Australian adults, it was revealed that more than 25% of participants had visited multiple practices in the previous year. It is unclear whether this behaviour changed in 2025. It is possible that the shorter longitudinality of the Australian dataset leads to underestimation of point prevalence and incidence estimates. Patients may be diagnosed outside of the index period, and patients may not return to the general practitioner for follow-up treatment (i.e. self-managed with OTC medications). It is also possible that missed cases could result as general practitioners may not record a patient visit as migraine if there are two reasons for the visit. The finding that 32.03% of patients were referred to physiotherapists could suggest misclassification of migraine as a musculoskeletal issue, or a correct migraine diagnosis with the aim to address co-morbid neck pain through physiotherapy [ 27 ]. The lack of data on referral reasons limits the ability to draw firm conclusions. There is also uncertainty regarding the reason for medication prescription, as neither are captured in the patient record. Consequently, there is also ambiguity regarding the reason for the use of migraine preventive medications, where they are indicated for multiple conditions. There is a possibility of under-reporting of OTC medications, such as paracetamol and aspirin. Furthermore, the precision will be lower in smaller subgroups, such as in older age groups or those in remote areas.

This study describes the epidemiology (incidence and prevalence) and management of diagnosed migraine in the Australian primary health care population through large-scale database analysis. It is the first study to describe the IR, accounting for life-years, in the Australian general practice population. The point prevalence of diagnosed migraine (< 1%) and incidence (0.348%) were lower than population-based estimates from other regions. The findings indicate that migraines are frequently underdiagnosed in the primary care setting, highlighting the need for further patient and clinician education. A lack of utilisation of acute migraine treatment options is seen, underscoring the need for further education in this area. Additionally, there appears to be a lack of awareness of potential presentations for migraine, including neck pain, with musculoskeletal referral and management pathways favoured over a diagnosis of a neurological condition and migraine specific treatment offered. The study also supports the common comorbidity among migraine patients of depression and/or anxiety and SSRI use — with subsequent low triptan use — suggesting a lack of awareness regarding the very low risk of serotonin syndrome which may influence prescribing decisions. Enhanced education of primary care prescribers and pharmacists regarding these issues, such the low risk of serotonin syndrome associated with triptans and SSRIs, migraine diagnosis, and alternative migraine presentations (e.g. neck pain), could potentially improve access to neurological care and disease-specific treatments for migraine sufferers. Increasing awareness of diagnostic tools for migraine, such as ID Migraine [ 28 ] as well as red flag indicators like SN(N)OOP4 (or 10) [ 29 , 30 ] for referral to specialists, and awareness of the latest treatment guidelines [ 11 ] could also assist primary care clinicians. Lastly, incorporating a migraine specific question into the Australian National Health Survey could be valuable for collecting more precise epidemiological data on this condition. These findings may inform future research, including prospective studies on diagnostic accuracy. The findings of this study further our understanding of Australian migraine epidemiology, how migraine is clinically managed, the associated burden of migraine, and suggest a need for a dedicated epidemiological study in Australia. These insights can be used to improve public health awareness, inform clinical care, and decrease suffering in people with migraine.

Supplementary Material 1. Supplementary Material 1.