Received: 25 October 2019 | Revised: 4 February 2020 | Accepted: 6 February 2020 DOI: 10.1111/pai.13237 ORIGINAL ARTICLE Incidence, risk factors and re-exacerbation rate of severe asthma exacerbations in a multinational, multidatabase pediatric cohort study Marjolein Engelkes1 | Esme J. Baan1 | Maria A. J. de Ridder1 | Elisabeth Svensson2 | Daniel Prieto-Alhambra3,4 | Francesco Lapi5 | Carlo Giaquinto6 | Gino Picelli7 | Nada Boudiaf8 | Frank Albers9 | Lee A. Evitt10 | Sarah Cockle10 | Eric Bradford10 | Melissa K. Van Dyke11 | Robert Suruki12 | Peter Rijnbeek1 | Miriam C. J. M. Sturkenboom13 | Hettie M. Janssens14 | Katia M. C. Verhamme1,15 1 Medical Informatics, ErasmusMC, Rotterdam, The Netherlands 2 Department of Clinical Epidemiology, Aarhus University, Aarhus, Denmark 3 Jordi Gol Primary Care Research Institute, Universitat Autonoma de Barcelona, Barcelona, Spain 4 Centre for Statistics in Medicine, NDORMS, University of Oxford, Oxford, UK 5 SIMG, Florence, Italy 6 University of Padova, Padova, Italy 7 , Pedianet, Padova, Italy 8 Research and Development, GlaxoSmithKline, Middlesex, UK 9 Research and Development, GlaxoSmithKline, Research Triangle Park, NC, USA 10 Research and Development, GlaxoSmithKline, Brentford, UK 11 Research and Development, GlaxoSmithKline, Collegeville, PA, USA 12 UCB Biosciences, Raleigh, NC, USA 13 Julius Global Health, Utrecht University Medical Center, Utrecht, The Netherlands 14 Pediatrics div Respiratory Medicine and Allergology, ErasmusMC- /Sophia Children's Hospital, University Medical Center Rotterdam, Rotterdam, The Netherlands 15 Department of Infection Control & Epidemiology, OLV Hospital, Aalst, Belgium Correspondence Katia Verhamme, Department of Medical Abstract Informatics, ErasmusMC, Dr Molewaterplein Background: There are sparse real-world data on severe asthma exacerbations (SAE) 50, 3000 CA Rotterdam, The Netherlands. Email: k.verhamme@erasmusmc.nl in children. This multinational cohort study assessed the incidence of and risk factors for SAE and the incidence of asthma-related rehospitalization in children with asthma. Present address Frank Albers, Avillion US Inc., Northbrook, Methods: Asthma patients 5-17 years old with ≥1 year of follow-up were identified IL, USA in six European electronic databases from the Netherlands, Italy, the UK, Denmark Funding information and Spain in 2008-2013. Asthma was defined as ≥1 asthma-specific disease code GSK (PRJ2284) and ZonMw (113201006). within 3 months of prescriptions/dispensing of asthma medication. Severe asthma Editor: Ömer Kalaycı was defined as high-dosed inhaled corticosteroids plus a second controller. SAE was The peer review history for this article is available at https://publons.com/publon/10.1111/pai.13237 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2020 The Authors. Pediatric Allergy and Immunology published by John Wiley & Sons Ltd 496 |  wileyonlinelibrary.com/journal/pai Pediatr Allergy Immunol. 2020;31:496–505. ENGELKES et al. | 497 defined by systemic corticosteroids, emergency department visit and/or hospitaliza- tion all for reason of asthma. Risk factors for SAE were estimated by Poisson regres- sion analyses. Results: The cohort consisted of 212 060 paediatric asthma patients contributing to 678 625 patient-years (PY). SAE rates ranged between 17 and 198/1000 PY and were higher in severe asthma and highest in severe asthma patients with a history of ex- acerbations. Prior SAE (incidence rate ratio 3-45) and younger age increased the SAE risk in all countries, whereas obesity, atopy and GERD were a risk factor in some but not all countries. Rehospitalization rates were up to 79% within 1 year. Conclusions: In a real-world setting, SAE rates were highest in children with severe asthma with a history of exacerbations. Many severe asthma patients were rehospi- talized within 1 year. Asthma management focusing on prevention of SAE is impor- tant to reduce the burden of asthma. KEYWORDS asthma: epidemiology, asthma: risk factors, epidemiology: prevalence 1 | I NTRO D U C TI O N Key Message Asthma is an important public health problem and is responsible for There are few studies on the real-life incidence of severe a high number of missed school and work days; moreover, it can be asthma exacerbations in children and the few available associated with significant limitations on physical, social and profes- strongly vary in study design. In this large international sional/student aspects of life when it is not controlled.1,2 Asthma retrospective cohort study we investigated the incidence exacerbations represent a decline in lung function with worsening of of and risk factors for severe asthma exacerbations in 6 respiratory symptoms such as shortness of breath, cough and/or chest European databases using one single protocol and har- tightness.2 According to a joint statement of the European Respiratory monized methods for data extraction and data analysis. Society and the American Thoracic Society, a severe asthma exacerba- We observed exacerbation rates ranging between 17 tion (SAE) is defined as the use of systemic corticosteroids, emergency and 198/1000 person-years in the different databases department (ED) visit or hospitalization all for reason of asthma.3 with high rehospitalization rates. This underscores the There are scarce international data on the incidence of SAE in importance of enhanced patient care in children with asthma children in the general population. The Asthma Insights and asthma to achieve the ultimate goal of preventing asthma Reality in Europe (AIRE) study was conducted in 1999, in which exacerbations. asthma patients from 7 different European countries were ques- tioned during a telephone interview. It showed differences in SAE rates between countries, as 7%-40% of children with asthma re- ported an ED visit for asthma and 3%-12% were hospitalized in the severe asthma, using one protocol and harmonized methods with previous year.4 As this study used interview data, it was susceptible regard to data extraction and data analysis, across multiple data to recall bias. Randomized controlled trials can also provide data on sources from five different European countries. the incidence of severe asthma exacerbations, but these numbers tend to be much lower than in real life due to strict inclusion and exclusion criteria and short follow-up.5 Risk factors for SAEs have 2 | M E TH O DS been studied primarily in adults, and thus, data in children are scarce. Currently, we still lack reliable instruments or biomarkers to pre- A retrospective cohort study was conducted using data from six dict SAE in children. Several studies found that the best predictor of European electronic healthcare databases, namely the Integrated future SAE was a history of a previous SAE.6,7 This implies that the Primary Care Information Project (IPCI) from the Netherlands, the risk of rehospitalization is probably high. Health Search Database (HSD) and Pedianet from Italy, Clinical In this study, we aimed to estimate real-world incidence rates of Practice Research Datalink (CPRD) from the UK, the Sistema and risk factors for SAEs, as well as to study the rate and cumulative d'Informació per al Desenvolupament de la Investigació en Atenció incidence of asthma rehospitalization in children with asthma and Primària (SIDIAP) from Spain and the Aarhus University Prescription 498 | ENGELKES et al. Database (Aarhus) from Denmark. Detailed descriptions of these leaving the database, reaching the age of 18, death or end of the databases have been published before and are available in the study period, whichever came first. Appendix S1.8-13 All databases comprise detailed information on drug prescriptions or dispensing, outpatient diagnoses and hospitali- zations, comorbidity and measurement data (eg laboratory results, 2.3 | Outcomes spirometry, BMI). While in most databases children of all ages are captured, in Italy for children 0-6 years old all primary care is pro- An SAE was defined as any of the following: use of systemic corti- vided by paediatricians, and parents can choose between a paedia- costeroids, ED visit or hospitalization, all for worsening of asthma. 3 trician and a GP for children 6-16 years of age. Therefore, Pedianet The indication of corticosteroid use was retrieved from the prescrip- (primary care paediatrician data) tends to contain younger children, tion/dispensing file or through an automated search on asthma (ex- while HSD (GP data) tends to capture older children. acerbation) disease codes in a 7-day window around the prescription All participating databases comply with EU guidelines on date. Continuous use of systemic corticosteroids, defined as con- the use of medical data for research and are registered in the secutive use of at least 30 days, was not considered as an SAE. If the European Network of Centres for Pharmacoepidemiology and time between exacerbations, of whatever type, was <7 days, it was Pharmacovigilance (ENCePP) database.14 In all 5 countries, the study considered to be one single exacerbation. protocol was approved by their local institutional review board. Rehospitalization was defined as an asthma-related hospi- talization following a previous asthma-related hospitalization. Hospitalizations were based on primary discharge diagnosis codes 2.1 | Cohort definition in SIDIAP, Aarhus and CPRD and were based on manual validation of discharge letters in the other databases. A cohort of patients with asthma was determined in each data- base within the study period (1 January 2008-31 December 2013). To enter the cohort, patients needed to be 5-17 years old, with a 2.4 | Covariates minimum of 1-year database history and diagnosed with asthma. Asthma was defined as physician-diagnosed asthma based on We investigated the prevalence of the following comorbidities at the presence of at least one asthma-specific disease code (see the start of follow-up and at each exacerbation, using the complete Appendix S1) in combination with prescriptions/dispensing of medical history: atopy (allergic rhinitis and/or atopic eczema/der- asthma medications within 3 months of the asthma disease code. matitis), chronic rhinosinusitis, nasal polyposis, gastro-oesophageal Asthma medications consisted of the following: inhaled corti- reflux disease (GERD), obesity and diabetes mellitus. In addition, we costeroids (ICS), short-acting β2 agonists (SABA), long-acting β2 investigated whether or not patients had a history of exacerbations. agonists (LABA), fixed combination of ICS + LABA, leukotriene History of exacerbations was defined as ≥2 SAEs in the 1 year prior receptor antagonist, short-acting muscarinic antagonist (SAMA), to entering the (severe) asthma cohort. If patients had <12 months fixed combination of SABA + SAMA, xanthines, systemic corti- prior to asthma diagnosis or the start of severe asthma, 1 exacerba- costeroids for the treatment of asthma, and anti-IgE treatment. tion or more in the 6 months prior fulfilled the criterion. We also Information on drug use was retrieved by an ATC-specific search investigated age and gender as covariates. from the drug prescription and/or drug dispensing records. Based on the first date of an asthma disease code, patients were catego- rized into prevalent or incident asthma. 2.5 | Analysis Within the cohort of patients with asthma, a subcohort of pa- tients with severe asthma was nested. According to international The exacerbation rates were calculated by dividing the number of guidelines, severe asthma was defined as asthma requiring treat- exacerbations by the respective number of person-years of follow- ment with high-dose ICS (as defined in the Global Initiative of up and additionally stratified by gender and age in years. Asthma (GINA); see Appendix S1) plus a second controller and/or Asthma rehospitalization rates in predefined follow-up time systemic corticosteroids for at least 120 consecutive days.2,15 To windows (30, 180 and 365 days following a previous hospitaliza- account for immortal time bias, the start date of severe asthma was tion) were calculated. If a patient had more than one exacerbation day 120 of high-dose ICS use. during that time window, each exacerbation contributed to the nom- inator. Furthermore, from these rates cumulative incidences were calculated. 2.2 | Follow-up Heterogeneity was quantified by the I2 statistics to assess if it was appropriate to perform a meta-analysis on the overall SAE rates Follow-up started at the latest date of the following: the start of estimated in the databases. the study period, diagnosis of (severe) asthma, age of 5 years or Risk factors for exacerbations were assessed by Poisson re- after reaching 365 days of database history. Follow-up ended when gression analyses, adjusting for age, gender and their interaction. ENGELKES et al. | 499 A multivariate model was fitted using the following covariates: age, was 57%, 25% and 77%, respectively. Large differences between gender, comorbidity (atopy, obesity, GERD), history of exacerba- databases in incidence rates (IR) of SAE were observed especially tions and the interaction between age and gender and between age for the younger age. The overall SAE rates ranged between 17 and and history of exacerbations. These analyses were repeated after 198/1000 PY and were higher in severe asthma patients (range 46- stratification for age: for children 5-11 years old and for children 375/1000 PY). SAE rates decreased with age, from age 5 onwards, ® 12-17 years old. Analyses were conducted using SAS version 9.4. and this was significant for all databases. (Figure 1 and Table 3) From adolescence on, higher IRs of SAE were observed for women com- pared to men in CPRD, SIDIAP, IPCI and Aarhus. (Figure 1) The IR 3 | R E S U LT S ranged between 11 and 131/1000 PY in patients without a history of exacerbations and was higher in children with severe asthma espe- The cohort comprised 212 060 children with asthma contributing cially with a history of exacerbations (283-1000/1000 PY). (Figure 2 678 625 patient-years (PY). In total, 14 283 children (6.7%, ranging from and Table S1) As the heterogeneity of exacerbation rates was sub- 1.6% to 18.3% across databases) with severe asthma were identified. stantial (I2 > 99%), it was not appropriate to pool the data. The mean age at the start of follow-up was 10.4 years (ranging between 7.2 and 14.8 years across databases), and a male prepon- derance was observed. The prevalence of atopy was 9.6% in Aarhus 3.1 | Risk factors for severe asthma exacerbations and ranged between 24.1% and 43.6% in the other databases. In all databases, the prevalence of atopy, GERD and obesity was higher in After adjustment for age, gender and their interaction, having atopy, patients with severe asthma. (Tables 1 and 2). GERD and obesity increased the risk of SAE in some databases. Of all asthma patients, depending on the database, 2.8%-29.3% (Table S2) In the multivariate model, the same risk factors remained, had at least one exacerbation during the follow-up. In HSD, SIDIAP but risk estimates were lower. (Table 3) Additionally, the interaction and PEDIANET, SAE mostly comprised use of systemic corticoste- between age and a history of exacerbations showed that in some da- roids (89%-100%), while in IPCI, Aarhus and CPRD, this proportion tabases, the risk associated with a history of exacerbations increased TA B L E 1 Baseline characteristics of total asthma cohorts CPRD (n, %) SIDIAP (n, %) IPCI (n, %) AARHUS (n, %) PEDIANET (n, %) HSD (n, %) Total 124 554 30 749 31 705 12 521 8264 4267 Female 51 667 (41.5) 12 357 (40.2) 13 262 (41.8) 4681 (37.4) 2980 (36.1) 1665 (39.0) Male 72 887 (58.5) 18 392 (59.8) 18 443 (58.2) 7840 (62.6) 5284 (63.9) 2602 (61.0) Age (mean, min-max) 10.9 (5.0-17.9) 8.9 (5.0-17.9) 10.2 (5.0-17.9) 9.3 (5.0-17.9) 7.2 (5.0-14.0) 14.8 (6.6-17.9) Severe asthma 9162 (7.4) 1139 (3.7) 1858 (5.9) 1207 (9.6) 137 (1.7) 780 (18.3) Incident asthma 23 743 (19.1) 13 062 (42.5) 4604 (14.5) 1392 (11.1) 1892 (22.9) 1302 (30.5) Prevalent asthma 100 811 (80.9) 17 687 (57.5) 27 101 (85.5) 11 129 (88.9) 6372 (77.1) 2965 (69.5) History of exacerbations Yes 5276 (4.2) 2625 (8.5) 241 (0.8) 408 (3.3) 196 (2.4) 321 (7.5) No 119 278 (95.7) 28 124 (91.5) 31 464 (99.2) 12 113 (96.7) 8068 (97.6) 3946 (92.5) Atopy 54 356 (43.6) 9767 (31.8) 11 602 (36.6) 1200 (9.6) 1990 (24.1) 961 (22.5) GERD 3247 (2.6) 248 (0.8) 510 (1.6) 103 (0.8) 146 (1.8) 55 (1.3) Diabetes mellitus 294 (0.2) 42 (0.1) 81 (0.3) 43 (0.3) 0 7 (0.2) Obesity 24 742 (19.9) 12 828 (41.7) 1518 (5.0) 205 (1.6) 3977 (48.1) 265 (6.2) Chronic rhinosinusitis 4917 (3.9) 33 (0.1) 112 (0.4) 0 0 38 (0.9) Nasal polyposis 228 (0.2) 0 13 (0.0) 8 (0.1) 0 0 Patients with ≥1 12 638 (10.2) 8968 (29.3) 871 (2.8) 756 (6.0) 724 (8.8) 855 (20.0) exacerbation Asthma 19 530 18 808 1290 1020 1031 1461 exacerbations (n) Systemic 15 049 (77.0) 17 323 (92.1) 739 (57.3) 252 (24.7) 918 (89.0) 1460 (99.9) corticosteroids (n, %) ED/ hospitalization 4481 (23.0) 1485 (7.9) 551 (42.7) 768 (75.3) 113 (11.0) 1 (0.1) (n, %) Abbreviations: ED, emergency department; GERD, gastro-oesophageal reflux disease. 500 | ENGELKES et al. TA B L E 2 Baseline characteristics of severe asthma cohorts CPRD (n, %) SIDIAP (n, %) IPCI (n, %) AARHUS (n, %) PEDIANET (n, %) HSD (n, %) Total 9162 1139 1858 1207 137 780 Female 3705 (40.4) 481 (42.2) 806 (43.4) 406 (33.6) 34 (24.8) 292 (37.4) Male 5457 (59.6) 658 (57.8) 1052 (56.6) 801 (66.4) 103 (75.2) 488 (62.6) Age (mean, 11.0 (5.0-17.9) 8.60 (5.0-17.9) 13.1 (5.0-17.9) 10.4 (5.0-17.9) 9.3 (5.0-12.9) 14.5 (2.4) min-max) Incident asthma 436 (4.8) 164 (14.4) 122 (6.6) 102 (8.5) 6 (4.4) 79 (10.1) Prevalent asthma 8726 (95.2) 975 (85.6) 1736 (93.4) 1105 (91.5) 131 (95.6) 701 (89.9) History of exacerbations Yes 562 (6.1) 215 (18.9) 49 (2.6) 40 (1.7) 1 (0.8) 55 (7.1) No 8600 (93.9) 924 (81.1) 1809 (97.4) 1167 (98.3) 136 (99.3) 725 (92.9) Atopy 4724 (51.6) 407 (35.7) 789 (42.5) 194 (16.1) 54 (39.4) 190 (24.4) GERD 375 (4.1) 23 (2.0) 34 (1.8) 14 (1.2) 3 (2.2) 16 (2.1) Diabetes mellitus 25 (0.3) 2 (0.2) 4 (0.2) 6(0.5) 0 2 (0.3) Obesity 2541 (27.7) 522 (45.8) 174 (9.4) 45 (3.7) 60 (43.8) 57 (7.3) Chronic 475 (5.2%) 3 (0.3%) 12 (0.6%) 0 0 7 (0.9%) rhinosinusitis Nasal polyposis 31 (0.3%) 0 3 (0.2%) 1 (0.1%) 0 0 Abbreviation: GERD, gastro-oesophageal reflux disease FIGURE 1 SAE rate (per 1000 PY) with 95% CI by age and gender in the total asthma cohort per database with increasing age, and in others, this risk decreased. After strati- 3.2 | Rehospitalization following previous fication for age category, for female gender in 5- to 11-year-olds a hospitalization for asthma exacerbation decreased risk was found, while in 12- to 17-year-olds there was an increased risk. These findings were less prominent in Pedianet Rehospitalization rates decreased with time following index hospi- and HSD, possibly due to the limited age range in these databases. talization. The asthma rehospitalization rate was highest in SIDIAP No other distinct differences between the age categories were ob- (within 1 month: 13.7/10 PY in children with asthma and 88.3/10 PY served (see Appendix S1). in children with severe asthma), whereas in the other databases the ENGELKES et al. | 501 rehospitalization rate within the first month ranged between 5.4 and (4.56, 22.20) 7.7/10 PY in children with asthma and between 14.3 and 15.8/10 PY (0.21, 1.02) (0.81, 1.05) (0.92, 1.02) (0.67, 1.80) (1.00, 1.11) (0.91, 0.99) (0.73 1.16) in children with severe asthma. 95% CI The cumulative incidence of rehospitalization within 1 year was 17%-29% in children with asthma, whereas this was much higher, namely 33%-79%, in patients with severe asthma. (Table 4) No data 1.06 0.46 10.07 0.92 0.92 0.97 0.95 HSD are presented for HSD and Pedianet as the number of children being 1.1 IRR hospitalized for asthma exacerbation was too low. (16.25, 53.05) (1.01, 1.34) (0.85, 0.98) (0.62, 1.61) (0.95, 1.06) (0.90, 1.15) (0.91, 0.97) (0.55, 1.41) 4 | D I S CU S S I O N 95% CI In this study, the overall exacerbation rate in children with asthma PEDIANET ranged between 17 and 198/1000 PY across databases, increased in children with severe asthma (range 46-375/1000 PY) and was 0.88 29.36 1.02 1.01 0.91 0.94 1.17 IRR 1 the highest in those with severe asthma with a history of exacer- bations (283-1000/1000 PY). Rehospitalization following a previ- ous hospitalization was high, both in the overall asthma cohort, (31.2, 66.92) (0.26,0.57) (1.75,2.44) (0.92,0.97) (0.91,0.97) (0.57,2.31) (1.55,2.96) (1.07,1.15) namely 17%-29% being rehospitalized in a year, and especially in 95% CI children with severe asthma, as up to 79% were rehospitalized within a year. We observed large differences in incidence rate of SAE, espe- AARHUS cially for the youngest age categories, and also large differences in 0.39 2.07 1.15 45.71 0.94 0.95 1.11 2.14 IRR the type of SAE. Large differences in exacerbation rates between European countries have previously been observed in other stud- ies that used questionnaires or interviews to collect data instead of (12.91, 31.10) (0.30, 0.61) (1.08, 2.17) (1.06, 1.12) (1.36, 1.69) (0.92, 0.96) (1.00,1.08) (0.94,1.51) healthcare databases. In line with our results, these studies showed 95% CI higher severe exacerbation rates in Italy and Spain than in the UK, and higher rates of emergency care in Spain than in the UK. In the Asthma Insights and Reality in Europe (AIR) survey, hospital- Relative rates for SAE in the total asthma cohort (multivariate analysis) ization rates in children were higher in Spain and Italy than in The 20.04 1.04 1.09 1.53 0.43 1.51 0.94 1.19 IPCI IRR Netherlands and the UK. 16-18 In these studies, many reasons for these differences were suggested: differences in disease manage- ment such as asthma prevention plans, lung function measurement, (1.00, 1.03) (0.88, 0.93) (1.30, 1.67) (1.01, 1.03) (1.06, 1.12) (0.78, 0.79) (2.27, 2.81) (0.76, 0.91) and difference in drug prescriptions, compliance with therapy, envi- 95% CI ronmental differences and cultural differences such as perception of disease burden and healthcare-seeking behaviour. Our exacerbation rates are in line with a recent study by Suruki SIDIAP et al investigating SAE in adolescent and adult asthma patients.19 0.83 1.09 2.53 1.48 1.02 1.01 0.78 Abbreviation: GERD, gastro-oesophageal reflux disease IRR 0.9 They reported a mean exacerbation rate in asthma patients 12-17 years old of 156/1000 PY based on US claims data and of (0.88, 0.89) (5.25, 6.33) (1.03, 1.05) (1.05, 1.07) (1.21, 1.29) (1.34, 1.41) (0.62, 0.74) (1.09, 1.28) 77/1000 PY using CPRD data. Comparison with other studies is dif- ficult, as most of these only investigated hospitalizations for asthma, 95% CI included systemic corticosteroids irrespective of the indication of use or investigated a selective population.17,19-24 Our observation on frequent rehospitalization is in line with data CPRD 1.04 1.25 1.06 0.68 0.89 1.37 1.18 5.76 IRR from two cohort studies on hospitalizations for children with asthma in the United States which observed that 17% and 22% of hospital- ized children were readmitted within 365 days.25,26 The risk fac- exacerbations exacerbations Age*History of Characteristic tors for SAE were a previous SAE, male gender, youngest age and History of TA B L E 3 atopy. In agreement with published literature, in our study a previous Age*Sex Obesity Female GERD Atopy exacerbation was the best predictor of a subsequent asthma exacer- Age bation.27-29 Other published risk factors for asthma exacerbations 502 | ENGELKES et al. F I G U R E 2 SAE rate (per 1000 PY) with 95% CI by history of exacerbations and severity (severe asthma in Pedianet not presented as low numbers) [Colour figure can be viewed at wileyonlinelibrary.com] such as obesity and GERD could not be confirmed as risk factor in all differences between the databases complicate comparison between participating databases.30,31 countries, we believe that it is better to study six different databases As for all observational studies, our study has strengths and in one identical way than having six different single database studies limitations. Major strengths include the large number of children using different methods because that would make comparison even with asthma included in our cohort, from 6 different European more difficult. databases in 5 countries, using a single protocol and harmonized For Aarhus and SIDIAP, dispensing data instead of prescrip- methods for data extraction and data analysis. Detailed informa- tion data were used, reducing misclassification of exposure. tion on important covariates such as drug exposure and underly- Unfortunately, dosing information is missing in Aarhus and SIDIAP ing comorbidity and on lifestyle factors was collected. The use of and dosing was derived from the strength per device and the time real-life data in a heterogeneous asthmatic population increases window between prescriptions. Of course, this method is prone to the external validity of our findings as our data are highly rep- misclassification, which might explain the relative low proportion of resentative of actual practice conditions and likely to minimize severe asthma patients in SIDIAP. Hawthorne and similar effects.32 Because of large numbers, manual validation of asthma was not However, as this is an observational study, using data from elec- feasible and asthma was defined based on the presence of disease tronic healthcare databases, there is always the potential of bias codes, and thus, potential misclassification of asthma cannot be and/or confounding. excluded. However, to ensure good quality of health care, GPs are First, we did report large differences in the incidence rates of trained and requested to enter disease codes fitting the actual pa- SAE between countries. tient's diagnosis. For those databases where automatic linkage with hospital data Furthermore, for each of the comorbidities of interest, diseases was available (CPRD, SIDIAP and Aarhus), misclassification of ED were mapped through the Unified Medical Language System (UMLS) visits/hospitalization is unlikely. For the other databases, hospital- generating a list of disease codes (see Appendix S1) which were ization or ED visit was retrieved either via disease-specific codes in verified by the databases prior to extraction.33 Still we cannot ex- combination with codes for hospitalization (HSD and Pedianet) or via clude potential misclassification in case certain disease codes were review of the discharge letters (IPCI). For prescriptions of systemic omitted or in case of country-specific differences in disease coding. steroids, only those with an indication for asthma registered in the Despite these potential limitations, we were able to give a good re- database were included. If practices did not report the indication of flection of exacerbations in the general population and found similar use or did not report disease codes around the prescription date, we trends of risk factors across the databases. might have underreported the incidence rate. Similar underreporting might have occurred if patients used previously collected systemic corticosteroids. 5 | CO N C LU S I O N Also, the large differences in SAE rates might be due to inter- national differences in the use of systemic steroids. For example, in In this study, the incidence rates of severe asthma exacerbation Spain (SIDIAP), a large incidence of SAE based on use of systemic in children ranged between 17 and 198/1000 PY and increased steroids was observed in young children. Despite the fact that these in children with severe asthma. In keeping with previous studies, ENGELKES et al. TA B L E 4 Asthma rehospitalization per 10 person-years following discharge from previous asthma hospitalization in different time periods per database (HSD and PEDIANET not presented as low numbers) <30 d after hospitalization <180 d after hospitalization <365 d after hospitalization Cumulative Cumulative Rate/10 PY incidence Rate/10 PY incidence Rate/10 PY Cumulative Rehospitalizations (n) (95% CI) (%) Rehospitalizations (n) (95% CI) (%) Rehospitalizations (n) (95% CI) incidence (%) CPRD Asthma 121 5.4 (4.5,6.5) 4.3 257 2.1 (1.9,2.4) 9.8 421 1.9 (1.7,2.1) 17.3 Severe asthma 58 15.7 (12.2, 12.1 100 5.3 (4.4,6.4) 23.0 137 4.0 (3.4,4.8) 33.0 20.4) SIDIAP Asthma 87 13.7 (11.1, 16.9) 10.6 164 5.1 (4.3,5.9) 22.2 205 3.4 (3.0,3.9) 28.8 Severe asthma 58 88.3 (68.3, 51.6 62 25.1 (19.5, 71.0 67 15.8 (12.4, 79.4 114.2) 32.2) 20.0) IPCI Asthma 12 6.4 (3.6, 11.2) 5.1 37 3.9 (2.9,5.4) 17.5 48 3.0 (2.2,3.9) 25.9 Severe asthma 6 14.3 (6.4, 31.8) 11.1 20 10.8 (7.0, 16.8) 41.3 22 7.3 (4.8, 11.1) 51.8 AARHUS Asthma 44 7.7 (5.7, 10.3) 6.1 98 3.1 (2.6,3.8) 14.2 133 2.3 (1.9,2.7) 20.0 Severe asthma 16 15.8 (9.7, 25.7) 12.2 32 6.3 (4.4,8.9) 26.7 46 5.1 (3.8,6.8) 40.0 Abbreviations: 95% CI, 95% confidence interval; PY, person-years. | 503 504 | ENGELKES et al. the most important risk factor for hospitalization was a previous REFERENCES exacerbation. 1. Blakey JD, Wardlaw AJ. What is severe asthma? Clin Exp Allergy. Rehospitalization for asthma following a previous hospitalization 2012;42(5):617-624. 2. GINA. Pocket guide for asthma management and prevention. 2019. was high, especially in children with severe asthma as up to 50% 3. Reddel HK, Taylor DR, Bateman ED, et al. An official American were rehospitalized within 1 month and up to 79% within 1 year thoracic society/European respiratory society statement: asthma following discharge. This underscores the importance of enhanced control and exacerbations: standardizing endpoints for clin- patient care in children with asthma to achieve the ultimate goal of ical asthma trials and clinical practice. Am J Respir Crit Care Med. 2009;180(1):59-99. preventing asthma exacerbations. 4. Vermeire PA, Rabe KF, Soriano JB, Maier WC. Asthma control and differences in management practices across seven European coun- AC K N OW L E D G M E N T S tries. Respir Med. 2002;96(3):142-149. Funding for this study was provided by GSK (PRJ2284) and the 5. Jackson DJ, Sykes A, Mallia P, Johnston SL. Asthma exacer- bations: origin, effect, and prevention. J Allergy Clin Immunol. ZonMw (ME and KV) (113201006). All listed authors meet the cri- 2011;128(6):1165-1174. teria for authorship set forth by the International Committee for 6. Haselkorn T, Fish JE, Zeiger RS, et al. Consistently very poorly con- Medical Journal Editors. This work was carried out through the EU- trolled asthma, as defined by the impairment domain of the expert ADR Alliance. The authors wish to acknowledge Eva Molero and panel report 3 guidelines, increases risk for future severe asthma Natasha Yefimenko (Synapse Research Management Partners) for exacerbations in the epidemiology and natural history of asthma: outcomes and treatment regimens (TENOR) study. J Allergy Clin their contributions to managing the study protocol during the devel- Immunol. 2009;124(5):895-902.e4. opment of this manuscript. 7. Miller MK, Lee JH, Miller DP, Wenzel SE, Group TS. Recent asthma exacerbations: a key predictor of future exacerbations. Respir Med. C O N FL I C T O F I N T E R E S T 2007;101(3):481-489. 8. Vlug AE, van der Lei J, Mosseveld BM, et al. Postmarketing surveil- FA, SC, EB, LAE and MKVD are employees of GSK and own stocks/ lance based on electronic patient records: the IPCI project. Methods shares in GSK. NB and RS were employees of GSK at the time this Inf Med. 1999;38(4-5):339-344. research was conducted and own stocks/shares in GSK. EJB, GP, 9. Ehrenstein V, Antonsen S, Pedersen L. Existing data sources for CG and KB have no conflicts to declare. FL has received grants from clinical epidemiology: Aarhus university prescription database. Clin Epidemiol. 2010;2:273-279. Chiesi, GSK and Novartis. DPA has received research grants from 10. Cazzola M, Puxeddu E, Bettoncelli G, et al. The prevalence of Amgen, Bioiberica and GSK and speaker/advisory fees from Amgen asthma and COPD in Italy: a practice-based study. Respir Med. and Bioiberica, paid to his department. ME has received grants from 2011;105(3):386-391. GSK and ZonMw. The institutions of MDR, PR, MS and KV have re- 11. Cricelli C, Mazzaglia G, Samani F, et al. Prevalence estimates for chronic diseases in Italy: exploring the differences between ceived unconditional research grants from Boehringer-Ingelheim, self-report and primary care databases. J Public Health Med. Novartis, Pfizer, Yamanouchi, Servier and Johnson & Johnson, unre- 2003;25(3):254-257. lated to the current manuscript; MDR, MS and KV have received an 12. Gulliford MC, van Staa T, Dregan A, et al. Electronic health records unconditional grant from GSK to conduct research on the incidence for intervention research: a cluster randomized trial to reduce an- tibiotic prescribing in primary care (eCRT study). Ann Fam Med. of and risk factors for asthma exacerbations as part of the GSK/EU- 2014;12(4):344-351. ADR Alliance project. HMJ has received grants from NIH and NCFS, 13. Sorensen HT, Larsen BO. A population-based Danish data resource participates in studies funded by an unconditional research grant with possible high validity in pharmacoepidemiological research. J from Vectura, and received travel expenses from Vertex, all unrelated Med Syst. 1994;18(1):33-38. to the current manuscript. ES’s institution (Aarhus University) has re- 14. Blake KV, Devries CS, Arlett P, Kurz X, Fitt H. Increasing scientific standards, independence and transparency in post-authorisation ceived a grant from the GSK/EU-ADR Alliance, related to this study. studies: the role of the European network of centres for pharmaco- epidemiology and pharmacovigilance. Pharmacoepidemiol Drug Saf. AUTHOR CONTRIBUTIONS 2012;21(7):690-696. ME, EJB, MDR, ES, DPA, FL, GP, RS, PR, MS, NB and KV contributed 15. Chung KF, Wenzel SE, Brozek JL, et al. International ERS/ATS guide- lines on definition, evaluation and treatment of severe asthma. Eur to the study concept and design and acquisition of data and were Respir J. 2014;43(2):343-373. involved in data analysis and interpretation. FA, EB, LAE, CG, SC, 16. Vermeire PA, Rabe KF, Soriano JB, Maier WC. Asthma control and MKVD and EJB contributed to the study concept and design and differences in management practices across seven European coun- were involved in data analysis and interpretation. All authors were tries. Respir Med. 2002;96(3):142-149. 17. Henneberger PK, Mirabelli MC, Kogevinas M, et al. The occupa- involved in the preparation and review of the manuscript and ap- tional contribution to severe exacerbation of asthma. Eur Respir J. proved the submitted version. 2010;36(4):743-750. 18. Hyland ME, Sthl E. Asthma treatment needs: a comparison of ORCID patients' and health care professionals' perceptions. Clin Ther. 2004;26(12):2141-2152. Esme J. Baan https://orcid.org/0000-0002-0735-4296 19. Suruki RY, Daugherty JB, Boudiaf N, Albers FC. The frequency of Frank Albers https://orcid.org/0000-0001-7039-4621 asthma exacerbations and healthcare utilization in patients with Melissa K. Van Dyke https://orcid.org/0000-0001-7246-2093 asthma from the UK and USA. BMC Pulm Med. 2017;17(1):74. ENGELKES et al. | 505 20. Suruki RY, Boudiaf N, Ortega HG. Retrospective cohort analysis 30. Longo C, Bartlett G, Schuster T, Ducharme FM, MacGibbon B, of healthcare claims in the United States characterising asthma Barnett TA. The obese-asthma phenotype in children: an exac- exacerbations in paediatric patients. World Allergy Organ J. erbating situation? J Allergy Clin Immunol. 2018;141(4):1239-1249. 2016;9:18. e4. 21. Bisgaard H, Moller H. Changes in risk of hospital readmis- 31. Quinto KB, Zuraw BL, Poon KY, Chen W, Schatz M, Christiansen sion among asthmatic children in Denmark, 1978–93. BMJ. SC. The association of obesity and asthma severity and control in 1999;319(7204):229-230. children. J Allergy Clin Immunol. 2011;128(5):964-969. 22. Morrison DS, McLoone P. Changing patterns of hospital admission 32. Sedgwick P, Greenwood N. Understanding the Hawthorne effect. for asthma, 1981–97. Thorax. 2001;56(9):687-690. BMJ. 2015;351:h4672. 23. Kemp T, Pearce N. The decline in asthma hospitalisations in persons 33. Gu H, Chen Y, He Z, Halper M, Chen L. Quality assurance of UMLS aged 0–34 years in New Zealand. Aust N Z J Med. 1997;27(5):578-581. semantic type assignments using SNOMED CT hierarchies. Methods 24. Mitchell EA, Cutler DR. Paediatric admissions to Auckland hospital Inf Med. 2016;55(2):158-165. for asthma from 1970–1980. N Z Med J. 1984;97(749):67-70. 25. Kenyon CC, Melvin PR, Chiang VW, Elliott MN, Schuster MA, Berry JG. Rehospitalization for childhood asthma: timing, variation, and S U P P O R T I N G I N FO R M AT I O N opportunities for intervention. J Pediatr. 2014;164(2):300-305. Additional supporting information may be found online in the 26. Auger KA, Kahn RS, Davis MM, Simmons JM. Pediatric asthma Supporting Information section. readmission: asthma knowledge is not enough? J Pediatr. 2015;166(1):101-108. 27. Price D, Wilson AM, Chisholm A, et al. Predicting frequent asthma exacerbations using blood eosinophil count and other patient How to cite this article: Engelkes M, Baan EJ, de Ridder MAJ, data routinely available in clinical practice. J Asthma Allergy. et al. Incidence, risk factors and re-exacerbation rate of 2016;9:1-12. severe asthma exacerbations in a multinational, 28. Salamzadeh J, Wong IC, Hosker HS, Chrystyn H. A Cox regression multidatabase pediatric cohort study. Pediatr Allergy Immunol. analysis of covariates for asthma hospital readmissions. J Asthma. 2020;31:496–505. https://doi.org/10.1111/pai.13237 2003;40(6):645-652. 29. Arabkhazaeli A, Vijverberg SJH, van Erp FC, Raaijmakers JAM, van der Ent CK, Maitland van der Zee AH. Characteristics and se- verity of asthma in children with and without atopic conditions: a cross-sectional study. BMC Pediatr. 2015;15:172.