Prevalence, incidence and healthcare burden of eosinophilic granulomatosis with polyangiitis in the UK

Background Eosinophilic granulomatosis with polyangiitis (EGPA) is a rare but serious disease characterised by the combination of small-to-medium vessel vasculitis, blood and tissue eosinophilia, and asthma and/or sinonasal disease. This study estimated the prevalence and incidence of diagnosed EGPA in the United Kingdom (UK), and described the demographics, clinical characteristics and healthcare resource utilisation (HCRU) of this population. Methods This retrospective longitudinal study of patients with newly diagnosed EGPA (index) (2005–2019) used the Clinical Practice Research Datalink AURUM and Hospital Episode Statistics databases. The primary outcomes were the annual prevalence (2005–2019) and incidence (2006–2019) of EGPA, and secondary outcomes included patient demographics and clinical characteristics, and HCRU in the year pre- and post-index (diagnosis). Results Populations of patients with EGPA comprised 940 prevalent cases and 502 incident cases, of which 377 were linked to Hospital Episode Statistics. EGPA prevalence increased from 22.7 to 45.6 cases per 1 000 000 (2005–2019), driven by patients aged ≥18 years. Incidence ranged from 2.3 to 4.0 per 1 000 000 person-years (2006–2019). Pre-index, the most common clinical symptoms were respiratory related, and the most common comorbidities were asthma (80.6%) and nasal polyps (32.1%). Post-index, 19.1% had an EGPA-related inpatient stay (median length of stay 11.0 days) and 38.7% had five or more oral corticosteroid (OCS) prescriptions with a mean OCS possession ratio per patient of 47.0%. Conclusions Although EGPA incidence in the UK remains relatively stable, prevalence is increasing, and HCRU and OCS use remain frequent, suggesting considerable healthcare burden for patients with EGPA.

The prevalence and incidence of EGPA varies globally, with estimated prevalence of 2.0-38.0 per 1 000 000 people [14][15][16][17], and incidence of 1.2 per 1 000 000 person-years [14].EGPA is a rare and challenging diagnosis, which is often misdiagnosed [18].Therefore, local differences in awareness and recognition among healthcare providers may contribute to regional variations, as has been observed in other AAVs [18,19].Another contributing factor may be the changing classification criteria over time [2,3,20,21].Given the rarity of and difficulty diagnosing EGPA, limited information is available on the prevalence, incidence and associated burden of disease in the United Kingdom (UK) [17,22].
This study aimed to estimate the prevalence and incidence of diagnosed EGPA in the UK, and to describe the demographics, clinical characteristics and HCRU of patients following EGPA diagnosis.

Study design and data sources
This was a retrospective, longitudinal study of patients newly diagnosed with EGPA (1 January 2005 to 31 December 2019) using the Clinical Practice Research Datalink (CPRD)-AURUM and Hospital Episode Statistics databases [23,24].The index date was the date of the first EGPA diagnosis during the study period based on the presence of a medcode identifier, Read code, Education Management Information System (EMIS) code or SNOMED identifier code for EGPA, allergic granulomatous angiitis or Churg-Strauss syndrome (supplementary table S1).The baseline and follow-up periods included the year pre-and post-index, respectively (supplementary figure S1).The CPRD-AURUM database (figure 1) consists of anonymised, longitudinal medical records of patients registered with contributing primary care practices across the UK (predominantly England and Northern Ireland) and contains data collected routinely from participating practices using the EMIS Web electronic patient record system software, including data on demographics, lifestyle factors, diagnoses, symptoms, prescriptions, referrals and medical tests [24].As of February 2019, CPRD-AURUM contained data of >22 million patients from 738 general practitioner (GP) practices in England, of whom 7 million were active (still alive and registered with a GP  practice), representing a coverage of ∼13% of the population of England.The CPRD-AURUM resource was launched in 2017, but the database includes a full historic collection of the coded part of each practice's electronic health records.Further characterisation of the data source has been published previously [24].
Anonymised data from CPRD-AURUM can be individually linked to secondary care and other health and area-based datasets, including the Hospital Episode Statistics database (figure 1).

Secondary outcomes included demographics at index, and clinical characteristics and Charlson
Comorbidity Index (CCI) score during baseline, and HCRU (including OCS use) during the follow-up period ( population 3).As Hospital Episode Statistics data are specific for England, the secondary outcomes data reflect an English rather than UK population.The CPRD-AURUM and Hospital Episode Statistics databases were used to identify clinical symptoms and comorbidity conditions using codes from a previous study (available upon request) [25].

Statistical analysis
This was a descriptive study, and neither hypothesis tests were conducted, nor formal power calculation performed.However, a feasibility assessment was performed, including widths calculations of the 95% confidence intervals for prevalence and incidence estimations, detailed in the supplementary methods and supplementary table S2.
EGPA prevalence was calculated as the number of patients with an EGPA diagnosis during and before a particular calendar year, divided by the number of patients with a calendar year of data in the CPRD-AURUM database on 31 December in each calendar year.EGPA incidence was calculated as the number of patients with an incident EGPA diagnosis from 1 January to 31 December in the calendar year of interest, divided by the total number of days at risk.For incidence rate, patients had at least 365 days after first registration in CPRD-AURUM prior to contributing to time at risk between 2006 and 2019.Time at risk started on day 366 after registration.This was to ensure that the incident cases were accurate and not an existing diagnosis that was recorded at time of registration.
All secondary outcomes were also analysed descriptively using mean±SD or median and interquartile range (IQR) for continuous variables and frequency (%) for categorical variables.CCI score was calculated using the Metcalfe adaptation [26].HCRU assessments included the proportion of patients with at least one event and mean number of events including EGPA-related and all-cause inpatient stays, all-cause A&E visits, specialist outpatient visits, all-cause outpatient visits, all-cause procedures and all-cause primary care visits.For inpatient stays, the cumulative and median length of stays was also reported.OCS use was measured according to the number of prescriptions throughout the year and split into quartiles, total prescriptions, and average days per year of use.The OCS medication possession ratio (MPR) was calculated based on the total number of days covered by OCS prescriptions (derived using quantity/daily dose variables) during the follow-up period divided by duration.

Patient populations
Population 1 and population 2 included 940 prevalent patients and 502 incident patients, respectively.There were 377 patients aged ⩾18 years who were successfully linked to CPRD-Hospital Episode Statistics and eligible for inclusion in population 3.  S3).The increase was driven by increased prevalence in patients aged ⩾18 years.The prevalence in the paediatric population aged ⩽17 years ranged between 0 and 0.51 (95% CI 0.01-2.84)per 1 000 000 people over the same period.

HCRU and OCS use in the 12 months following EGPA diagnosis in England
In the first 12 months following EGPA diagnosis, 49.9% of patients had all-cause inpatient stays and 19.1% had EGPA-related inpatient stays (table 2).The mean±SD number of annual EGPA-related inpatient stays was 1.2±0.6 per patient, with a median (IQR) length of stay of 11 (6.0-17.0)days.5% of patients required all-cause A&E visits, with a mean annual number of 1.8±1.7 visits per patient.Overall, 97.1% of patients had GP visits and 88.6% had outpatient visits (table 2).The most common specialist outpatient visits were with respiratory medicine (33.7% of patients, with an annual mean of 3.9±2.8visits per patient), followed by general medicine (32.9% of patients, with an annual mean of 3.5±3.6visits per patient) and rheumatology (31.8% of patients, with an annual mean of 2.8±2.6 visits per patient).The mean number of GP, nurse or allied health professional visits per patient per year was 16.0±11.1,3.4±3.9and 7.2±8.9,respectively.
OCS use was high, with 38.7% of patients having five or more prescriptions for OCS during the 12-month follow-up period (figure 3).

Discussion
EGPA is a rare disease and previous estimates of the incidence and prevalence are limited [17,22,28,29].To our knowledge, this is the first study assessing the prevalence and incidence of diagnosed EGPA exclusively in the UK, together with the associated disease burden.This study reported prevalence and incidence estimates of EGPA in the UK population of 22.7-45.6 per 1 000 000 people and 2.3-4.0 per 1 000 000 person-years, respectively, which is higher than estimates reported for other European countries between 1992 and 2017 [14].Furthermore, the annual EGPA prevalence increased over the study period, driven by increases in adult prevalence; whereas, overall, the incidence remained stable in all age-ranges.The results presented herein suggest a high healthcare burden for patients with EGPA in the UK, as well as a treatment burden suggested by the high OCS use in this population.This highlights an unmet clinical need that could potentially be addressed by optimised management and/or new optimised treatments.Recent published guidelines have highlighted the use of newer treatments such as biologics, including anti-interleukin (IL)-5 therapies, for the induction of remission or the maintenance of remission for patients with EGPA [18].Clinical benefits of anti-IL-5 therapies for patients with EGPA include OCS-sparing effects [30][31][32].
Patient demographics and clinical characteristics were similar to those reported in previous retrospective database studies assessing the prevalence, incidence and burden of EGPA in other countries [5,9,15,33].
Prior to diagnosis patients most commonly experienced respiratory-related symptoms and >80% had   comorbid asthma.This is consistent with the commonly reported pattern of disease development leading to EGPA where the development of asthma typically pre-dates the development of hypereosinophilia and vasculitis by several years [5,7].
Previously reported European and global pooled estimates from a meta-analysis of observational studies covering study periods from 1992 to 2017 indicated an EGPA prevalence of 12.1-15.3per 1 000 000 and an incidence of 1.1-1.2 per 1 000 000 person-years, although individual studies varied substantially.However, the results of that meta-analysis, and the underlying original studies, have limitations, including the change of criteria for identifying patients with EGPA over time and their inconsistency across studies, and the estimates for prevalence were heavily influenced by the high patient sample of one particular study from claims databases in the United States of America (USA) [14].In the UK, data from the early 2000s suggested a prevalence of 38.0 per 1 000 000 (2000) and an incidence of 4.2 per 1 000 000 person-years (2004), but with no clear explanation of the methods employed to obtain such estimates [22].Similarly, a previous England-based study, which analysed Hospital Episode Statistics data, indicated a prevalence of 31.8 cases per 1 000 000 in 2016 [17].By comparison, in the current study, the 2016 prevalence of EGPA was estimated to be 42.5 cases per 1 000 000.This discrepancy may be due to differences in data source, study methodology and reporting period.For example, the current study includes primary care data from across the UK (CPRD-AURUM database), whereas the England-based study only utilised Hospital Episode Statistics and therefore would not have captured patients seen in primary care but not treated in the hospital setting in that period, which may have not captured less severe cases of EGPA [17].Finally, the previous study estimated the point prevalence on a given day in 2016, whereas our study estimated annual prevalence from 2005 to 2019.
In the current study, prevalence of EGPA increased two-fold from 2005 to 2019, while EGPA incidence varied, but had no overall increase.Similarly, a retrospective study of administrative claims from the Japan Medical Data Centre (JMDC) claims database (132 patients) found a nine-fold increase in EGPA prevalence from 4.2 to 38.0 per 1 000 000 from 2005 to 2017, where EGPA cases were diagnosed via ICD-10 code for EGPA (M30.1),plus an additional ICD-10 code for allergic rhinitis, asthma or chronic sinusitis prior to their EGPA diagnosis [15].This trend of increasing EGPA prevalence in Japan has continued between 2017 and 2020 [34].This is consistent with previous studies in Australia and France, which showed two-to three-fold increases in EGPA prevalence over 8-10 years from the late-1990s to mid-2000s, but with little change in incidence, although both studies were small with only eight and 31 EGPA cases identified, respectively [35,36].However, the previously mentioned systematic review and meta-analysis study reported no strong trends for increasing EGPA prevalence over time [14].These apparent differences in prevalence highlight the difficulties in determining accurate prevalence estimates, and may reflect the impact of EGPA rarity, difficulty of diagnosis and disease under-recognition [37].Nonetheless, EGPA prevalence may have increased over time due to changes to the diagnostic criteria, increased disease awareness and/or the combination of a stable incidence rate and high long-term survival rates [2,14,20,21,38].The cumulative survival rate for patients with EGPA at 5 and 10 years from disease onset ranges between 89-97% and 79-89%, respectively [5,13,37,39,40].Conventional therapy for EGPA allows for high overall survival rates, and patients with EGPA are living longer, despite living with a high disease burden [9,41].HCRU in the year after a EGPA diagnosis was common, with half of patients having an inpatient stay for any reason and almost one-fifth of patients having an EGPA-related inpatient stay.On average, patients had one EGPA-related inpatient visit per year, staying for a median of 11 days per visit.Given the cost of inpatient treatment and the high demand for hospital beds [33], the extended length of hospital stays for EGPA-related treatment demonstrates the sizeable per-patient disease burden for the UK health system.The high discrepancy between all-cause and EGPA-related inpatient stays may reflect an underestimation of the latter due to the challenge in attributing the varied clinical manifestations to EGPA [5,12], and complications from OCS use [10,18].Additionally, many patients with EGPA experience asthma-related inpatients stays [9] and EGPA may therefore not be reported as the primary reason for such stays.The high HCRU burden of EGPA identified in this study is consistent with that demonstrated in other countries [9,14,15,33].For example, a previous systematic review and meta-analysis, which included studies from the USA, Europe, Australia and Japan, indicated that ∼42% of patients with EGPA required an unscheduled hospital visit [14].Furthermore, the study found that patients with EGPA required a median of one (range 0-6) hospital visit and one (range 0-12) A&E visit annually [14], consistent with the data reported here.
The high OCS use observed in this study is broadly consistent with the OCS dependence demonstrated in previous studies [9,15,33].Indeed, 38.7% of patients accumulated five or more OCS prescriptions over the year following diagnosis, although there is some evidence that these became less frequent with increasing time from diagnosis.Data on OCS dose were not available here, but previous studies have demonstrated a requirement for high-dose OCS among patients with EGPA.For example, a retrospective Japan-based study found that OCS dose reduced from baseline of 39.1 mg•day −1 ) in the year following an EGPA diagnosis, but remained high in absolute terms (mean of 9.8 mg•day −1 and most patients had daily dose ⩾15 mg•day −1 ) [15].Combined with this study's results, these observations suggest that patients with EGPA remain dependent on OCS, increasing the potential for OCS-related toxicity [8,42].
The burden of acute and chronic corticosteroid-related complications and associated HCRU in severe asthma and the risks increase with cumulative corticosteroid exposure are well documented [11,42].
Treatment guidelines for EGPA highlight the importance of minimising OCS exposure [8], and novel OCS-sparing therapies that control symptoms while reducing treatment-related side-effects are needed.
Given the role of eosinophils in the pathology of EGPA, biologics targeting IL-5, the major cytokine responsible for eosinophil differentiation, survival and activation [43][44][45], have been investigated for use in EGPA, and have shown benefit as OCS-sparing treatments [30,46].The anti-IL-5 monoclonal antibody mepolizumab is approved for the treatment of eosinophil-driven diseases including EGPA in multiple regions worldwide [47][48][49]; however, anti-IL-5 therapies are not currently approved by the National Institute for Health and Care Excellence in the UK for the treatment of EGPA.
A strength of this study was that it utilised the UK-wide CPRD-AURUM database to assess the prevalence and incidence of EGPA, as well as the England-specific Hospital Episode Statistics database which captures a patient's complete NHS HCRU profile.As of 2019, the CPRD-AURUM database included data from ∼13% of the population in England [24].In another study with a similar approach, the use of ICD codes in the Hospital Episode Statistics database for the diagnosis of AAV was validated, as these codes were found to have an 86% positive predictive value [17].Although diagnosis was obtained in CPRD-AURUM via different coding systems in the present study, this solidifies the Hospital Episode Statistics database as a promising data source for linkage to CPRD-AURUM for retrospective studies in EGPA.In terms of limitations, reasons for OCS use and the OCS dose were not captured in the Hospital Episode Statistics database, so it was not possible to distinguish whether OCS prescriptions were for EGPA or other comorbid conditions, or to calculate cumulative steroid exposure.EGPA diagnosis can be complicated by the heterogeneous nature of the disease, the need to exclude "vasculitis mimics" and other small/medium-vessel vasculitis, and overlap with other eosinophilic diseases, which can lead to delayed diagnosis or misdiagnosis [1][2][3].Consequently, prevalence and incidence could have been underestimated.Furthermore, it is possible that a patient may have had a previous EGPA diagnosis from a non-CPRD-AURUM practice, which could have resulted in previously diagnosed patients being incorrectly included in the first diagnosis/incidence population.Moreover, the number of all-cause hospitalisations being nearly three-fold that of EGPA-related hospitalisations in this study might suggest under-coding.Additionally, the findings of the study may not be generalisable to practices and patients not enrolled in CPRD-AURUM, although a previous assessment of the database found that it was representative of the English population [24].Finally, this study also shares limitations typical of retrospective database studies, such as potential inconsistencies and errors in the diagnostic codes used to identify EGPA and comorbidities.

Conclusion
In the UK, although the incidence of EGPA remains relatively stable, the prevalence of EGPA is increasing.This study adds to the currently limited UK-specific data on EGPA prevalence and incidence, and identifies for the first time the considerable healthcare burden for patients with EGPA in the UK, as indicated by frequent HCRU and OCS use.This study suggests a high level of remaining unmet need for patients with EGPA, and future studies are needed to understand the impact of new treatments on the patient and disease burden.

FIGURE 2 a
FIGURE 2 a) Prevalence and b) incidence of eosinophilic granulomatosis with polyangiitis in the United Kingdom over time.Patients from population 1 (940 prevalent patients) and population 2 (502 incident patients).

40 FIGURE 3
FIGURE 3 Prescriptions for oral corticosteroids (OCS) in the 12 months after index in England.Patients from population 3.

TABLE 1
Patient demographics # at index and clinical characteristics during the baseline period ¶ Data are presented as n or n (%), unless otherwise stated.IQR: interquartile range; CCI: Charlson Comorbidity Index; ENT: ear, nose and throat.# : patients from population 3; ¶ : baseline period defined as the year before index (inclusive); + : the maximum value was reported if multiple values were available; § : ⩾1 code for characteristic of interest; ƒ : patients aged 0-17 years were not included due to the small number (fewer than five) of patients included in this age group; ## : for clinical characteristics/conditions with fewer than five patients, Clinical Practice Research Datalink required data to be suppressed to minimise the risk of patient identification.https://doi.org/10.1183/23120541.00430-20235 ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | J. HWEE ET AL.

TABLE 2
Healthcare resource utilisation in the year following eosinophilic granulomatosis with polyangiitis (EGPA) diagnosis in England Data are presented as n or n (%), unless otherwise stated.IQR: interquartile range; A&E: accident and emergency; ENT: ear, nose and throat.# : number of patients with one or more event; ¶ : patients from population 3; + : top 10 most frequent specialty outpatient visits.https://doi.org/10.1183/23120541.00430-20236 ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | J. HWEE ET AL.

TABLE 3
Oral corticosteroid (OCS) use in the year following eosinophilic granulomatosis with polyangiitis diagnosis in England # # : patients from population 3; ¶ : patients with at least one event.https://doi.org/10.1183/23120541.00430-20238 ERJ OPEN RESEARCH ORIGINAL RESEARCH ARTICLE | J. HWEE ET AL.