Received: 1 February 2019 Revised: 20 June 2019 Accepted: 5 July 2019 DOI: 10.1111/bcp.14071 ORIGINAL ARTICLE Incidence of direct oral anticoagulant use in patients with nonvalvular atrial fibrillation and characteristics of users in 6 European countries (2008–2015): A cross‐national drug utilization study Luisa Ibáñez1,2,3 | Mònica Sabaté1,2,3 | Xavier Vidal1,2,3 | Elena Ballarin1,2 | Marietta Rottenkolber4 | Sven Schmiedl5,6 | Andreas Heeke7 | Consuelo Huerta8 | Elisa Martin Merino8 | Dolores Montero8 | Luz María Leon‐Muñoz8 | Christiane Gasse9 | Nicholas Moore10 | Cécile Droz10 | Régis Lassalle10 | Mia Aakjær11 | Morten Andersen11 | Marie Louise De Bruin12 | Rolf Groenwold13 | Hendrika A. van den Ham14 | Patrick Souverein14 | Olaf Klungel14,15 | Helga Gardarsdottir15,16,17 1 Fundació Institut Català de Farmacologia (FICF), Hospital Vall d'Hebron, Barcelona, Spain 2 Department of Clinical Pharmacology, Hospital Universitari Vall d'Hebron, Barcelona, Spain 3 Department of Pharmacology, Toxicology and Therapeutics, Universitat Autònoma de Barcelona, Spain 4 Diabetes Research Group, Medizinische Klinik und Poliklinik IV, Klinikum der Universität München, Munich, Germany 5 Department of Clinical Pharmacology, School of Medicine, Faculty of Health, Witten/Herdecke University, Witten, Germany 6 Philipp Klee‐Institute for Clinical Pharmacology, Helios University Hospital Wuppertal, Wuppertal, Germany 7 AOK NORDWEST, Dortmund, Germany 8 Pharmacoepidemiology and Pharmacovigilance Division, Spanish Agency of Medicines and Medical Devices (AEMPS), Madrid, Spain 9 Aarhus University, Aarhus C, Denmark 10 Bordeaux PharmacoEpi, INSERM CIC1401, Université de Bordeaux, CHU de Bordeaux, Bordeaux, France 11 Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark 12 Copenhagen Centre for Regulatory Science (CORS), Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark 13 Department of Clinical Epidemiology, Leiden University Medical Center, Leiden, The Netherlands 14 Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Faculty of Science, Universiteit Utrecht, David de Wiedgebouw, Utrecht, The Netherlands 15 Julius Center, University Medical Center Utrecht, Utrecht, The Netherlands 16 Department of Clinical Pharmacy, University Medical Center Utrecht, Utrecht, The Netherlands 17 Faculty of Pharmaceutical Sciences, University of Iceland, Reykjavik, Iceland Correspondence Luisa Ibáñez, Fundació Institut Català de Farmacologia (FICF), Hospital Universitari Vall d'Hebron, Passeig de la Vall d'Hebron, 119–129, 08035 Barcelona, Spain. Email: li@icf.uab.cat Funding information European Medicines Agency, Grant/Award Number: EMA/2015/27/PH reopening of competition no3 The authors confirm that the Principal Investigator for this paper is Luisa Ibáñez and that she had direct clinical responsibility for patients. L. Ibáñez and M. Sabaté have contributed equally. -------------------------------------------------------------------------------------------------------------------------------- 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. © 2019 European Medicines Agency. British Journal of Clinical Pharmacology published by John Wiley & Sons Ltd on behalf of British Pharmacological Society. 2524 wileyonlinelibrary.com/journal/bcp Br J Clin Pharmacol. 2019;85:2524–2539. IBÁÑEZ ET AL. 2525 Aims: To estimate the incidence of direct oral anticoagulant drug (DOAC) use in patients with nonvalvular atrial fibrillation and to describe user and treatment characteristics in 8 European healthcare databases representing 6 European countries. Methods: Longitudinal drug utilization study from January 2008 to December 2015. A common protocol approach was applied. Annual period incidences and direct standardisation by age and sex were performed. Dose adjustment related to change in age and by renal function as well as concomitant use of potentially interacting drugs were assessed. Results: A total of 186 405 new DOAC users (age ≥18 years) were identified. Standardized incidences varied from 1.93– 2.60 and 0.11–8.71 users/10 000 (2011–2015) for dabigatran and rivaroxaban, respectively, and from 0.01–8.12 users/10 000 (2012–2015) for apixaban. In 2015, the DOAC incidence ranged from 9 to 28/10 000 inhabitants in SIDIAP (Spain) and DNR (Denmark) respectively. There were differences in population coverage among the databases. Only 1 data- base includes the total reference population (DNR) while others are considered a population representative sample (CPRD, BIFAP, SIDIAP, EGB, Mondriaan). They also varied in the type of drug data source (administrative, clinical). Dose adjustment ranged from 4.6% in BIFAP (Spain) to 15.6% in EGB (France). Concomitant use of interacting drugs varied between 16.4% (SIDIAP) and 70.5% (EGB). Cardiovascular comorbidities ranged from 25.4% in Mondriaan (The Netherlands) to 82.9% in AOK Nordwest (Germany). Conclusion: Overall, apixaban and rivaroxaban increased its use during the study period while dabigatran decreased. There was variability in patient characteristics such as comorbidities, potentially interacting drugs and dose adjustment. (EMA/2015/ 27/PH). K E Y W OR D S anticoagulants, arrhythmia, cardiovascular, drug utilization, pharmacoepidemiology 1 | I N T RO D U CT I O N addition, the very elderly, who may be at increased risk of adverse effects with inappropriate DOAC use, are poorly represented in Patients with atrial fibrillation have long been treated with vitamin K clinical trials.16,17 antagonists for the prevention of cerebral embolism. The newer direct This cross‐national comparison drug utilization study, using longi- oral anticoagulants (DOACs) have been approved by the European tudinal data from 8 electronic health care databases in 6 European Union since 2008. The first was dabigatran (2008), a thrombin inhibi- countries, uses a common protocol to characterize DOAC users in a tor; this was followed by rivaroxaban (2008), apixaban (2011) and real‐world setting in order to establish the effectiveness of existing edoxaban (2015), factor Xa inhibitors. The first approval for the pre- risk minimization measures and their appropriateness for the future. vention of stroke and systemic embolism in patients with nonvalvular Its objectives are to assess incidence of use and user characteristics, atrial fibrillation (NVAF) was for dabigatran in 2011 (Table S1). DOACs including concomitant exposure to potentially interacting medicines are currently recommended by the European Society of Cardiology as and rates of dose adjustment related to age or renal impairment. first‐line anticoagulant treatment in NVAF, without mitral stenosis or mechanical heart valves.1 The incidence of NVAF is estimated to be 3% in adults aged 20 or 2 | METHODS 2,3 older, increasing with age, with incidence of 12% in females and 14% in males aged over 75 years.3 Atrial fibrillation is associated with 2.1 | Data sources 4,5 increased cardiovascular morbidity, such as heart failure and stroke. Clinical trials have shown that both vitamin K antagonists and DOACs We conducted a longitudinal drug utilization study in 6 countries reduce stroke and mortality in AF patients.5,6 Utilization of DOACs for between January 2008 and December 2015. Data were retrieved stroke prevention in NVAF and their effectiveness and safety in clini- from the following 8 databases: (i) the Dutch Mondriaan project, 7-13 cal practice have been assessed in several European countries. which includes the Julius General Practitioner Network (JHN) data- However, little is known about their use beyond clinical trial condi- base18; (ii) the Danish National Registries (DNR), which includes the tions, especially in patients with hepatic or renal impairment. Findings Danish National Patient Register, Danish National Prescription Registry from 2 studies from USA and Australia showed that inappropriate and Danish Civil Registration System19-21; (iii) the AOK Nordwest data- dosing occurred among patients with renal failure, and there is still base22,23, Germany; (iv) the Bavarian Association of Statutory Health 14,15 uncertainty about appropriate dosing in these patients. In Insurance Physicians database, referred to here as Bavarian CD, 2526 IBÁÑEZ ET AL. Germany24; (v) the Base de datos para la Investigación Farmacoepidemiológica en Atención Primaria (BIFAP), Spain25; (vi) the What is already known about this subject Information System for the Development of Research in Primary Care • An increase in the number of direct oral anticoagulant (SIDIAP), Catalonia, Spain26; (vii) the Clinical Practice Research drug (DOAC) users with a diagnosis of nonvalvular atrial Datalink (CPRD), UK27,28; (viii) The Echantillon Généraliste de fibrillation (NVAF) has been reported since their Bénéficiaires (EGB), France.29 The databases characteristics are marketing in several national/regional studies in Europe, described in Table 1. but no cross‐national comparison is available. • The characteristics of DOAC users related to NVAF have 2.2 | Study population been described but concomitant use of potentially interacting drugs is rarely reported. DOAC users are The study population was defined as all new users (age ≥18 years) of usually younger than 75 years, male and 20% of DOAC the DOACs of interest (dabigatran, rivaroxaban, apixaban). Only those users with NVAF receive dose adjustment related to patients with a diagnosis of NVAF (see codes in supplementary mate- renal function. rial, Table S2) were included. A common protocol was applied for data extraction and analysis (EU PASS Register No: 16014) for each data- What this study adds base.30 Results were blinded to each database lead until the analysis • Overall DOAC incidences varied from 1.93 to 2.60 and was completed. 0.11 to 8.71 users/10 000 (2011–2015) for dabigatran New users were defined as patients initiating DOAC during the and rivaroxaban, respectively, and from 0.01 to 8.12 study period without any use of DOAC for at least 365 days prior to users/10 000 (2012–2015) for apixaban. the index date (date of first DOAC prescription). Flow charts for • In 2015, the new user DOAC incidence ranged from 9 to patient inclusion for each database are shown in Figure S1. 28/10 000 inhabitants 18 years and older in SIDIAP Patients registered in the database <1 year (365 days) before the (Catalonia, Spain) and DNR (Denmark) databases index date and patients with a history of valvular atrial fibrillation on respectively, this being higher in men than in women index date or any time prior to initiating DOACs were excluded. and in those older than 75 years. Follow‐up of each patient was until therapy switch, discontinuation or end of study, whichever came first. Switchers were defined as • Concomitant use of contraindicated or potentially patients with a subsequent prescription of another type of (D)OAC, interacting drugs varied between 16.4% (SIDIAP) and within the first treatment episode. Discontinuers were defined as 70.5% (EGB). Dose adjustment related to age or renal patients who did not receive subsequent DOACs within 30 days fol- function varied from 4.6% (BIFAP, Spain) to 15.6% lowing the theoretical end date of the prior DOAC. (EGB, France). As DOACs can be prescribed for indications other than NVAF, a sensitivity analysis was performed excluding patients with multiple potential DOAC indications in a ± 3 month period around the first DOAC prescription. (SmPC) during the first DOAC treatment episode of each patient. A list of the concomitant interacting drugs considered in the SmPC is given 2.3 | Ethical approval in eSupplementary Material (Table S3). A treatment episode was defined as series of prescriptions for a DOAC, independent of dose Participants in each country had study approval from the correspond- changes, considering a gap of up to 30 days, constructed according ing data owners. No other requirement was required since anonymized to the method of Gardarsdottir et al.31 data were used. Additionally, the study protocol was revised and Furthermore, the occurrence of dose adjustment, defined as approved by an internal European Medicines Agency expert panel. changing from 1 tablet strength to another strength of the same active substance, was assessed during the first treatment episode. 2.4 | Outcomes The main outcome was assessing incidence of DOAC use in patients 2.5 | Analysis with NVAF during the study period. Annual period incidences are estimated and defined as the number The analysis is descriptive and stratified by database, individual of new users during the year of interest, divided by the total number DOAC, age group (<75, 75–79, ≥80 years), sex and calendar year. of patients in the database at midyear (1 July). The baseline characteristics (demographics [sex, age]), comorbidi- Secondary aims were to assess concomitant use of interacting ties, chronic kidney disease (CKD), renal function, laboratory data drugs, defined as any prescription of a potentially interacting medicinal when available, hepatic impairment, previous major haemorrhagic epi- product as indicated in the Summary of Product Characteristics sodes, previous cardiovascular events (see codes in Table S4) and Database characteristics IBÁÑEZ TABLE 1 Danish National ET AL. Mondriaan Registries AOK Nordwest Bavarian Claims BIFAP SIDIAP CPRD EGB Source population 0.4 5.5 2.7 10.5 7.5 5.8 12.5 0.7 (million inhabitants) Population coverage 3% 100% 24% 84% 16.4%a 80% 8%a Representative sample of Système National d'Informations Inter‐ Régimes de l'Assurance Maladie [98.8%] Year(s) covered for this 2012–2015b 2008–2015 2008–2015 2008–2015 2008–2015 2009–2015 2008–2015 2013–2015c study Type of databased GP prescribing data Dispensing data from Claims database Claims database General practice General practice General practice Claims database including and pharmacy community including data including data prescribing prescribing data/ prescribing data. dispensed reimbursed dispensing data. pharmacies. for dispensed for dispensed data/ reimbursed data. data. reimbursed reimbursed reimbursed drugs drugs. data. Data available since 1991 1994 2007 2008 2001 2006 1987 2004 Demographic variables available Date of registration Yes Yes Yes Yes (first Yes Yes Yes Yes consultation) Date of transferring Yes Yes Yes Yes (last Yes Yes Yes Yes out consultation) Date of birth DD‐MM‐YY MM‐YY MM‐YY MM‐YY MM‐YY MM‐YY MM‐YY YY Sex Yes Yes Yes Yes Yes Yes Yes Yes Drug information available Active international ATC ATC ATC ATC ATC ATC BNF ATC coding Product coding HPK Nordic article PZN PZN CNF Yese Product code CIP‐13 number Date of prescribing/ Yes Yes Primary care Yes (for each Yes Yesf Yes Yes dispensing sector: Yes prescription the quarter is documented) Primary care sector GP Yes Yes Yes Yes Yes Yes Yes Yes Specialist No Yes Yes Yes No No No Yes (Continues) 2527 TABLE 1 (Continued) 2528 Danish National Mondriaan Registries AOK Nordwest Bavarian Claims BIFAP SIDIAP CPRD EGB Secondary care sector No No Yes—only for a No No No No Yes—only for a few selected few selected (expensive) compounds, (expensive) but no (D)OAC compounds, prescriptions but no (D)OAC prescriptions Quantity prescribed/ Yes Yes Yes (package size) Yes (package size) Yes Yesg Yes Yes dispensed Dosing regimen No No No No Yes Yesh Yes No Outcome information Outpatient primary ICPC No ICD‐10‐GM ICD‐10‐GM (for ICPC‐2, ICD‐9 ICD‐10 ICD‐9, ICD‐10 ICD‐10 care diagnosis (quarterly base) each diagnosis the quarter is documented) Hospital discharge No ICD‐8, ICD‐10 ICD‐10‐GM No No Noi ICD‐9, ICD‐10 ICD‐10 diagnosis Laboratory tests Yes No No No Yes (as requested Yes Yes No by GP) Mortality Yes Yes No (incompletely No Yes (no cause of Yes (no cause of Yes Yes (no cause of death) recorded, e.g. death) death) no follow‐up for patients leaving the AOK) a Representative of general population. b No data on DOAC) use available until 2012. DOACs were not prescribed by general practitioners in the Netherlands before 2012. c No data on DOAC use available until 2013. d Prescribing databases: collect information on prescribed drugs by GP; Reimbursement databases: collect information on dispensed drugs funded by Health Care Services; Dispensing databases: collect all dis- pensation of prescription drugs regardless of the drug's reimbursement status (Ref. Eriksson E and Ibañez L. Secondary data sources for drug utilization research. In Elseviers M et al. Drug utilization research. Methods and applications. John Willey and Sons, 2016). e Registered but not available for research due to confidentiality reasons; f For dispensing: MM/YY; g Number of reimbursed packages; h Only in prescribing data; i Available for 28% of included population. (D)OAC, (direct) oral anticoagulant; GP, general practitioner; ATC, Anatomical Therapeutic Classification; BNF, British National Formulary; HPK, Handels Product Code; PZN, Pharma Zentral Nummer; CNF, IBÁÑEZ Código Nacional Farmacéutico; ICD (International Classification of Diseases; ICD‐10‐GM: ICD ‐10‐German Modification), ICPC (International Classification of Primary Care). ET AL. IBÁÑEZ ET AL. 2529 concomitant exposure to potentially interacting medicines of DOAC The number of patients with acute or chronic kidney disease, iden- users are presented as absolute number and percentages for each tified through diagnosis codes, ranged from 0% in the Mondriaan to variable. 24.1% in the Bavarian CD databases. Assessment of laboratory values For all databases, annual period incidences, with direct with moderately reduced kidney function (Mondriaan, BIFAP, SIDIAP standardisation by age and sex was performed based on the European and CPRD) showed a range from 3.0% (Mondriaan) to 22.6% (CPRD). standard population corresponding to each year analysed.32 An inci- DOAC users with severely or very severely reduced renal function dence percentage change in DOAC users with NVAF is given in com- were very uncommon in these databases. However, the proportion parison to the first calendar year when NVAF became an approved of unregistered information was usually high (range: 3.4% in CPRD indication for use (2012 for dabigatran and rivaroxaban, 2013 for to 77.9% in BIFAP; Table S5). apixaban). However, in the EGB database, DOAC percentage change was assessed for 2013–2014 since these are the calendar years with complete information. In addition, a percentage change of the stan- 3.3 | Incidence dardized incidence, weighted by the database populations, was calcu- lated from the first to the last calendar year of use. During this period, overall DOAC incidence increased (Figure 1). Percentage of new DOAC users exposed to potentially interacting Standardized incidences varied from 1.93–2.60 and 0.11–8.71 medication is expressed as the absolute number and percentage of the users/10 000 (2011–2015) for dabigatran and rivaroxaban, respec- total DOAC users. Dose adjustment is expressed as the percentage of tively, and from 0.01–8.12 users/10 000 (2012–2015) for apixaban patients with adjusted dosage following the requirements of the (Table 3). Apixaban displayed the highest standardized incidence per- SmPC and presented either related to change in age or renal function, centage change from the first year of use following approval in as indicated in the SmPC. NVAF to the final year studied (543.2%), followed by rivaroxaban (100.2%; Figure 1). This was mainly driven by a sharp increase in EGB. The standardized percentage incidence change for this first cal- 2.6 | Nomenclature of targets and ligands endar year compared to the previous year was maximum in EGB (10 550.0%) and minimum in Bavarian CD (218.6%); values compar- Key protein targets and ligands in this article are hyperlinked to ing for 2014 and 2015 were maximum in Mondriaan (868.5%) and corresponding entries in http://www.guidetopharmacology.org, the minimum in SIDIAP (35.1%).The percentage for rivaroxaban common portal for data from the IUPHAR/BPS Guide to PHARMA- increased in all databases during the first calendar year of use, COLOGY,33 and are permanently archived in the Concise Guide to except in EGB, with the greatest increase in CPRD (120.9%); a PHARMACOLOGY 2017/18.34 decrease was seen in the 2 German databases and EGB in 2014– 2015 (Table 4). The standardized figures show that the incidence of dabigatran 3 | RESULTS clearly increased in most databases from 2010 to 2012, with the highest value in the DNR (15.5 new users per 10 000). The CPRD and Mondriaan figures increased slightly at the end of the study 3.1 | New users period. The apixaban standardized incidence increased across A total of 186 405 new DOAC users (age ≥18 years) with NVAF were 2013–2015 in all databases. The maximum value was observed in identified during the study period. the DNR database in 2015 (13.6 new users per 10 000). The Of all new DOAC users, 91 804 (49%), 52 495 (28%) and 42 106 rivaroxaban standardized incidence increased over time in all data- (22%) received rivaroxaban, dabigatran and apixaban, respectively. bases since its arrival on the market, except in EGB; however, it started to decrease in Bavarian CD in 2014 and AOK in 2015. The Bavarian CD database presented the highest rivaroxaban standard- 3.2 | Baseline characteristics ized incidence in 2013 (17.5 new users per 10 000). Rivaroxaban showed the highest annual incidence values in the 2012–2015 Most of the users were age 75 years or older (48.8% in Mondriaan period in all databases except in the DNR (Figure 2; Tables S6–S9). to 60.8% in BIFAP). The mean age ranged from 69.3; SD 11.3 The DOAC incidence increased in males and females across the (Mondriaan) to 75.7; SD 10.4 years (BIFAP) (Table 2). study period in most databases, especially in those over 75 years, Users were most frequently male (range 52.3–58.9%), except in whose incidences were higher than those, men and women, younger the AOK and Bavarian CD databases where the opposite was found than 75 years (Figures S2–S4). (range 46.4–47.7%). In 2015, the incidence of DOAC use ranged from 8.6 per 10 000 in The proportion of patients with comorbidities ranged from 31.1% SIDIAP to 27.6 per 10 000 in DNR (Table S6), with a higher incidence (Mondriaan) to 87.3% (AOK). Most frequent were previous cardiovas- in men than in women. cular events (25.4% in Mondriaan to 82.9 and 76.7% in AOK and Figure 3 shows the incidence data for all DOACs, per database and Bavarian CD, respectively). per year, for the whole study period. 2530 TABLE 2 Baseline characteristics for all direct oral anticoagulant new users Mondriaan Danish National Registries AOK Nordwest Bavarian Claims BIFAP SIDIAP CPRD EGB 460 44 876 21 718 84 276 14 161 11 962 6931 2021 n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) Female 189 41.1 20 331 45.3 11 650 53.6 44 070 52.3 6755 47.7 5598 46.8 3049 44.0 867 42.9 Male 271 58.9 24 545 54.7 10 068 46.4 40 206 47.7 7406 52.3 6364 53.2 3882 56.0 1154 57.1 Age (y), mean (SD) 69.3 11.3 73.9 10.9 74.5 11.0 74.5 11.0 75.7 10.4 74.6 10.9 74.7 11.0 72.8 12.0 18–74 310 67.4 22 977 51.2 8795 40.5 36 658 43.5 5558 39.2 5165 43.2 3110 44.9 995 49.2 75–79 61 13.3 7258 16.2 4785 22.0 17 979 21.3 2647 18.7 2249 18,8 1300 18.7 353 17.5 ≥80 89 19.3 14 641 32.6 8138 37.5 29 639 35.2 5956 42.1 4548 38,0 2521 36.4 673 33.3 Comorbidities 143 31.1 29 161 66.7 18 952 87.3 73 488 87.2 6865 48.5 6437 53.8 4337 62.6 1042 51.6 Hepatic impairment 0 0 969 2.1 3119 14.4 14 845 17.6 33 0.2 27 0.2 3 0.0 52 2.6 Chronic kidney disease (by diagnosis codes) 0 0 2055 4.6 4377 20.1 20 287 24.1 226 1.6 509 4.3 309 4.5 58 2.9 Previous major haemorrhagic episodes 28 6.1 1642 3.6 2634 12.1 13 745 16.3 532 3.8 500 4.2 177 2.6 60 3.0 Haemorrhagic stroke/intracranial bleeding 21 4.6 231 0.5 382 1.7 2087 2.5 37 0.3 66 0.5 15 0.2 14 0.7 Extracranial or unclassified major bleeding 0 0 833 1.8 1824 8.4 8182 9.7 326 2.3 300 2.5 114 1.6 34 1.7 Gastrointestinal bleeding 7 1.5 536 1.2 903 4.1 4036 4.8 168 1.2 116 1.0 49 0.7 14 0.7 Traumatic intracranial bleeding 0 0 136 0.3 40 0.2 671 0.8 10 0.1 30 0.2 0 0 4 0.2 Previous cardiovascular events 117 25.4 29 959 65.0 17 998 82.9 64 618 76.7 5330 37.6 5653 47.3 3107 44.8 996 49.3 Coronary heart disease: Angina/myocardial infarction 68 14.8 10 584 23.6 12 439 57.3 46 062 54.6 1932 13.6 2156 18.0 1286 18.6 376 18.6 Congestive heart failure 41 8.9 20 134 44.9 9708 44.7 39 363 46.7 1921 13.6 2278 19.0 755 10.9 690 34.1 Peripheral arterial disease 0 0 1768 3.9 4876 22.4 16 782 19.1 531 3.7 784 6.5 308 4.4 0 0 Stroke/transient ischaemic attack 26 5.7 9555 21.3 6098 28.1 22 789 27.0 2099 14.8 2418 20.2 1474 21.3 208 10.3 Deep vein thrombosis/pulmonary embolism 0 0 2121 4.7 1114 5.1 5247 6.2 58 0.4 68 0.6 236 3.4 21 1.0 Aortic plaque 4 0.9 140 0.3 961 4.4 4163 4.9 952 6.7 9 0.1 11 0.2 121 6.0 IBÁÑEZ ET AL. IBÁÑEZ ET AL. 2531 FIGURE 1 Distribution of the total standardized incidence by individual direct oral anticoagulant (DOAC; 2011–2015) TABLE 3 Standardized incidences by individual direct oral antico- between 0.4% (CPRD) and 24.3% (EGB). Nonsteroidal anti‐ agulant (DOAC) per year (new users per 10 000 people) inflammatory drugs (NSAIDs) varied from 4.3% (Mondriaan) to 26.0% (Bavarian CD) and antiplatelet drugs from 1% (SIDIAP) to DOAC 2010 2011 2012 2013 2014 2015 18.1% (EGB). The most frequent interacting drugs were heparins in Dabigatran 0.10 2.60 5.33 4.98 3.49 1.93 AOK, BIFAP and Bavarian CD (8.4, 10.4 and 12.1%, respectively), Rivaroxaban 0.06 0.11 4.35 8.77 8.85 8.71 amiodarone in SIDIAP, DNR and EGB databases (5.7, 6.2 and Apixaban 0.00 0.00 0.01 1.26 4.99 8.12 42.2%, respectively) and verapamil in Mondriaan (4.1%; Table 5). 3.4 | Concomitant use of potentially interacting 3.5 | Dose adjustment drugs The information on dose adjustment was available in BIFAP, SIDIAP, The proportion of patients who received an interacting drug during CPRD and EGB, varying from 4.6% in BIFAP to 15.6% in EGB. The the first treatment episode ranged from 16.4% (SIDIAP) to 70.5% proportion of dose adjustments related to changes in renal function (EGB). Concomitant use of contraindicated anticoagulants varied or age was <1% in the 3 databases where this information was TABLE 4 Standardized incidence percentage change in direct oral anticoagulant new users with a diagnosis of nonvalvular atrial fibrillation (NVAF) in the time periods for the first and last calendar year of use after NVAF indication approval (2012–2015) Incidence percentage change for the first Incidence percentage change from calendar year of use after NVAF approvala Incidence percentage change 2014–2015 the first calendar year of use to 2015 Dabigatran Apixaban Rivaroxaban Dabigatran Apixaban Rivaroxaban Dabigatran Apixaban Rivaroxaban (%) (%) (%) (%) (%) (%) (%) (%) (%) Mondriaan −22.93 833.97 2397.26 32.54 868.50 63.36 11.52 8945.51 21 048.75 NRD −4.58 268.27 266.94 −63.54 60.94 42.41 −73.56 492.68 418.75 AOK Nordwest −8.04 313.13 98.40 −35.54 80.03 −12.89 −66.09 643.75 74.60 Bavarian Claims −18.43 218.61 67.96 ‐ 41.56 47.54 −24.18 −74.38 370.09 14.99 BIFAP −15.02 712.82 227.42 −6.08 52.37 12.50 −32.81 1138.46 320.97 SIDIAP −7.13 462.60 247.88 −5.80 35.13 27.75 −26.43 660.23 480.62 CPRD 80.98 774.46 732.89 3.55 180.91 120.90 88.31 2356.46 4902.39 EGB −69.55 10 550.0 −32.33 −56.37 162.44 −16.21 −86.72 27850.0 −43.30 a Data for dabigatran and rivaroxaban are calculated for 2012–2013. Data for apixaban are calculated for 2013–2014 in all databases. Data for each DOAC are calculated for 2013–2014 in EGB. 2532 IBÁÑEZ ET AL. FIGURE 2 Standardized incidences by individual direct oral anticoagulant in each database IBÁÑEZ ET AL. 2533 4 | DISCUSSION Using a common protocol, we assessed the incidence of DOAC use during 2009–2015 and the characteristics of 186,405 users from 8 health care databases in 6 European countries (2008–2015). To our knowledge this is the first cross‐national drug utilization study provid- ing the incidence of DOAC use in NVAF patients at a national/regional level, across several European countries. Only a few, single country European studies with similar inclusion criteria have been published, which only provide number of DOAC users rather than incidence figures.10,12,35,36 During the study period, the overall incidence of DOAC user increased, except in the EGB database; the individual DOAC with the highest increase was apixaban followed by rivaroxaban. The largest incidence increase was for apixaban in the EGB. It was maximum for apixaban in the first calendar year of use in EGB (10 550.0) and minimum in Bavarian (218.6) while the values for 2014–2015 were maximum in Mondriaan (868.5) and minimum in SIDIAP (35.1). The striking increase for apixaban in EGB is due to the fact that its use was very low in 2013. In 2015, the incidence of DOAC use ranged from 8.7 per 10 000 in SIDIAP to 27.6 per 10 000 in DNR, with a higher incidence in men and in those older than 75 years. The differences in incidence across the databases might be explained by the high proportion of previous cardiovascular events in AOK, Bavarian CD and DNR databases (82.9, 76.7 and 65.0%, respectively), which may result in higher incidence of NVAF in these populations. These results correlate well with the distribution of hos- pital discharges for cardiovascular and cerebrovascular diseases in these countries.37 Furthermore, differences in incidence could arise from the charac- FIGURE 3 Standardized and crude incidence for all direct oral teristics of the databases such as: prescription vs dispensing databases anticoagulants per year and database (DOACs prescribed are not necessarily dispensed), inclusion of prescriptions from specialists (although we do not expect differences available (BIFAP, SIDIAP and CPRD). In the Mondriaan database in the management of NVAF between primary and specialist care there were no dose changes recorded. DNR, AOK and since the European guidelines do not differentiate), population Bavarian CD databases do not have this information registered coverage (DNR is the only national database), different prescribing (Table S5). patterns in different countries, health services characteristics and their reimbursement policies, publication of guidelines (local or European), media and marketing policies.9,38 3.6 | Sensitivity analysis Several national/regional studies have reported increases in the number of DOAC users, despite the NVAF population used differing Results from the main analysis were compared with results from from ours.9,10,39,40 The incidence increased in both the males and patients with only NVAF diagnosis. The total number of patients females across the study period in most databases. More specifically, treated with DOACs decreased slightly, but the proportions of the incidence was higher in males than in females, since NVAF is more individual DOAC were stable. common in men.10,35,41,42 Regarding renal function at baseline and dose adjustment, the numbers were quite similar. The percentage of concomitant interacting medications increased in SIDIAP database both for all 4.1 | Incidence of individual DOACs DOACs and individual DOACs. The percentage of previous cardiovas- cular events was lower for all DOACs across all databases and the Overall, rivaroxaban presented the highest incidence figures, except in overall percentage of previous haemorrhagic events was similar the DNR, followed by apixaban and dabigatran. The initial steep (Tables S10–S12). increase in rivaroxaban and apixaban incidence observed in most of 2534 TABLE 5 Concomitant treatment during the first treatment episode Danish National Mondriaan Registries AOK Nordwest Bavarian Claims BIFAP SIDIAP CPRD EGB 460 95 937 21718 84 459 14 161 11 962 6931 2021 n (%) n (%) n (%) n (%) n (%) n (%) n (%) n (%) CONCOMITANT 94 20.4 15 765 37.8 10 585 48.7 44 394 52.7 6781 47.9 1963 16.4 1785 25.8 1425 70.5 MEDICATION (potential interacting medicine products) Concomitant treatment 21 4.6 446 1.1 3730 17.2 17 153 20.4 1425 10.1 355 3.0 29 0.4 492 24.3 with any other anticoagulants (contraindicated) Most frequent drug Vitamin K 3.9 Vit K 0.6 Heparin 8.4 Heparin 12.0 Heparin 10.4 Heparin 2.9 Vitamin K antagonists 0.4 antagonists antagonists group group group group Heparin group 12.2 Platelet aggregation 23 5.0 6455 14.4 2140 9.9 9298 11 1859 13.1 123 1.0 849 12.2 365 18.1 inhibitors excl. Heparin Strong CYP3A4 and/or 0 0.0 0.5 0.3 329 1.5 1746 2.0 282 2.0 107 0.9 29 0.4 3 0.1 P‐gp inhibitor comedication Most frequent drug NA NA Dronedarone 0.3 Dronedarone 1.3 Dronedarone 1.9 Dronedarone 1.6 Dronedarone 0.8 Dronedarone 0.4 Cyclosporine 0.1 Not strong CYP3A4 30 6.5 5473 12.2 2203 10.1 8172 9.7 1592 11.2 712 5.9 623 9.0 900 44.5 and/or P‐gp inhibitors Most frequent drug Verapamil 4.1 Amiodarone 6.2 Amiodarone 6.7 Amiodarone 6.5 Amiodarone 8.8 Amiodarone 5.7 Amiodarone 6.4 Amiodarone 42.2 CYP3A4 and/or P‐gp 1 0.2 194 0.4 189 0.9 556 0.7 80 0.5 26 0.2 36 0.5 5 0.2 inducers Most frequent drug NA NA Carbamazepine 0.3 Carbamazepine 0.7 Carbamazepine 0.5 Carbamazepine 0.2 Rifampicin 0.1 Phenytoin 0.3 Rifampicin/ Carbamzepine 0.1% Transporter interaction: 8 1.7 456 1.0 123 0.6 160 0.2 139 0.1 58 0.5 10 0.1 21 1.0 CYP3A4 and/or P‐gp Most frequent drug Naproxen 1.3 Fluconazole 0.7 Naproxen 0.3 Naproxen 0.1 Naproxen 0.5 Naproxen 0.3 Naproxen 0.2 Fluconazole 0.6 Nonsteroidal anti‐ 20 4.3 5878 13.1 5319 24.5 21935 26.0 3337 23.6 696 5.8 313 4.5 308 15.2 inflammatory drug Other drugs (only for 9 2.0 2704 6.0 361 1.7 1417 1.7 507 3.6 257 2.1 700 10.1 47 2.3 dabigatran) IBÁÑEZ (Continues) ET AL. IBÁÑEZ ET AL. 2535 the databases appeared by the end of 2011 and 2012 respectively, 1.9 0.5 (%) after approval for stroke prevention.43 This sharp increase was not observed for dabigatran as clearly in any of the databases. Other stud- ies have shown a steep increase in users for each DOAC after their marketing.36,39 2021 EGB 39 10 n 4.2 | Demographic characteristics 9.1 1,2 (%) The mean age of our study population reflected other European studies on users of DOAC with NVAF, despite varying inclusion CPRD 6931 629 10,12,35,36,39,41,44 criteria. In line with other studies, we found that most 83 n of the users were older than 75 years, except in Mondriaan and DNR 1.8 0.5 (%) databases.10,12,40 A higher proportion of males was observed in all except the Ger- man databases, similarly to other published studies.10,12,35,36,41 The SIDIAP 11 962 larger proportion of women in the German databases could be related 210 64 n to the characteristics of the population or to differences in NVAF inci- 2.9 0.9 (%) dence. Other studies have reported that female patients are more prevalent than males in those older than 75 or 80 years.45,46 14 161 4.3 Baseline clinical characteristics BIFAP | 411 126 n The clinical characteristics observed are comparable to similar studies 1.4 0.4 (%) which used OAC naive or non‐naive NVAF patients.10,35,36 However, Bavarian Claims there is heterogeneity of the demographics and baseline characteris- tics across the databases. Different population coverage across the 84 459 databases, as well as differences in the prevalence of cardiovascular 1145 346 diseases, could explain the observed differences.37 n Previous cardiovascular events were the most frequent comorbid- 1.4 0.3 (%) ity across all databases. The observed high proportion of previous car- AOK Nordwest diovascular events in the German databases correlates with the high percentage of cardiovascular problems observed in Germany.37 It 21718 might also be related to the inclusion of data from medical specialists 304 63 in the German databases, whereas many of the other databases in this n study consider data from general practitioners only. In addition, 5.3 0.9 (%) resource allocation for German sickness funds is based on the so‐ Danish National called morbidity‐oriented risk structure compensation scheme. There- Registries fore, all primary and secondary diagnoses must be coded to enable 95 937 appropriate calculations.47 The low prevalence in Mondriaan database 2380 406 might be because only the least critical patients received DOAC from n primary care, as opposed to specialists. 0.7 1.3 (%) The unexpectedly high proportion of CKD in the AOK and Bavar- ian CD populations may be explained by a higher proportion of Mondriaan patients with cardiovascular comorbidities or the inclusion of nonspe- 460 cific codes. 3 6 n reuptake inhibitors reuptake inhibitors (Continued) 4.4 | Concomitant use of potentially interacting P‐gp, P‐glycoprotein. Selective serotonin norepinephrine drugs Serotonin‐ Concomitant use of drugs has been reported to be low and in very few TABLE 5 studies.35 Concomitant treatment with other anticoagulant drugs was present in variable proportions, around 20% in the 2 German 2536 IBÁÑEZ ET AL. databases, and EGB. This is of concern since these patients might have promoting independent results. The consented and broad definition a higher haemorrhagic risk. The presence of a high cardiovascular of the inclusion criteria ensures the generalisability of the results as comorbidity proportion in the German databases could partly explain all the new users with a diagnosis of NVAF were presumably included. a higher use of medications. Furthermore, these 3 databases include In addition, the performance of a sensitivity analysis excluding patients reimbursed drugs from specialist prescriptions. However, some of this with multiple potential DOAC indications gave similar results for most use could also be related to switching to anticoagulant therapy. of the main variables. This supports the results of the main analysis The concomitant use of cardiovascular drugs has been reported in with respect to closeness to real‐life DOAC use, since several indica- some studies, in particular antiplatelet drugs, between 11 and 30% of tions might be present in addition to NVAF. users12,36; similarly, antiplatelet drugs were used in 10–18% of the DOAC users in several databases in our study. Amiodarone, a strong P‐glycoprotein inhibitor, was the most frequent potentially interacting 4.7 | Limitations drug in several databases. Lower use has been described in the OAC naïve Danish study.36 There was large case mix in the databases, which made comparisons NSAIDs have been reported in similar or somewhat lower propor- between countries difficult. In addition, there are differences in popu- tions in other studies.12,35,41 Certain NSAIDs are available over the lation coverage among the databases. Only 1 database includes the counter in some countries, as well as on prescription. This, together total reference population (DNR), while others are considered repre- with differing prescriber behaviours treating pain or inflammatory con- sentative samples of the national populations when considering age, ditions, may account for some of the observed differences.48 sex and geographical distribution (UK/CPRD, Spain/BIFAP and SIDIAP, France/EGB, Mondriaan). The inclusion of nonrepresentative databases with population coverages under 90% (German databases) 4.5 | Dose adjustment related to age or changes in must be considered when extending the results to the whole popula- renal function tion. Since we used prescription, dispensing or reimbursement data that do not have complete information on actual drug intake, there The proportion of dose adjustment related to age or changes in renal might have been certain degree of drug use misclassification, common function was low (<1%). However, this result should be interpreted to all clinical studies, even randomized trials. In addition, drugs pre- with caution as data on renal function results were sparse (unregis- scribed by physicians, other than general practitioners, could be tered data for renal failure up to 77.9%). In fact, a study aiming to missed when using prescribing databases as these are commonly gen- report CKD prevalence and recognition in a Swedish healthcare cohort eral practice databases. showed that registration of CKD diagnosis was suboptimal, with only A drop in the number of patients was observed in some databases 12% of affected patients having an ICD‐10 related diagnostic code.49 (Figure S1). This may be related to the fact that DOACs can be used However, we suspect these unregistered values are more likely to for a variety of diagnoses. Although some patients with NVAF may reflect less severe impairment. Moreover, not all reasons for dose have been lost in this process, the criteria ensure that we can be certain changes are registered in the different databases and we only consid- all patients included in the analysis do indeed have NVAF. ered those during the treatment episode. Therefore, patients appropri- Information on the indication associated with the prescription ately dosed at the index date, without any further dose changes, are might also be incomplete. For example, a definite linkage between not considered as adjusted. The discrepancy between the proportion compound and indication is lacking in most of the databases hence, of patients with changes in tablet strengths and the low proportion we used the term potential indications. Additionally, some data of patients with age and CKD related adjustments might be due to sources include data from hospital admissions and contacts (DNR), patients with an increase in tablet strength not being accounted for. while others include exclusively general practice encounters. Codes Other studies have reported about 20% patients receiving inappropri- used in diagnosis were not specifically validated in this study but ate doses, most often too low (38%) or excessively high dose outcome validation has been performed in other studies showing (22%).14,15 high validity.19,50-57 Laboratory values were not documented in some databases so codes for renal impairment (including somewhat 4.6 | Strengths nonspecific renal impairment codes) were used. This broad definition may explain the high proportion of patients with CKD found in some The main strength of this study is the large number of patients provid- databases. Dose adjustment related to change in renal function and ing real‐world data about incidence, concomitant use of potentially age was only assessed if a change in dose was associated with a interacting drugs and dose adjustments. In addition, the use of a stan- change in renal function or in age; however, we did not assess if dardized protocol across the different databases supports the inter- changes in renal function were subsequently followed by changes pretation and comparability of results from selected countries. in dose. Moreover, a dose adjustment was defined as switching from Disease and drug codes were harmonized, the study results were 1 tablet strength to another of the same active substance. This def- blinded and only shared with the whole consortium after each centre inition precludes assessing posology changes that may have taken had completed their analysis, avoiding some information bias and place without changing tablet strengths. Regarding potential IBÁÑEZ ET AL. 2537 interactions, lacking documentation of over the counter medications O.K. has provided an educational lecture (nonproduct related) for might be relevant, in particular for some NSAIDS, low‐dose Roche in May 2017. acetylsalicylic acid (mainly pharmacodynamic interactions) and St M.An. reports research grants from AstraZeneca, H. Lundbeck & John's Wort (pharmacokinetic interaction). Mertz, Janssen, Merck Sharp & Dohme, Novartis, Nycomed, and Pfizer, In conclusion, this study shows an increased incidence of use of received by the institutions at which he has been employed. M.An. DOACs related to NVAF in the study period across 6 European coun- has received fees for organising and teaching pharmacoepidemiology tries. In 2015, the incidence of DOAC use ranged from 8.7 per 10 000 courses at Medicademy, the Danish Association for the Pharma- in Spain to 27.6 per 10 000 in Denmark, with a higher incidence in ceutical Industry. M.An.'s professorship is supported by a grant from men than in women, especially in patients ≥75 years. Potential use the Novo Nordisk Foundation to the University of Copenhagen of contraindicated drugs, such as other anticoagulants, in some coun- (NNF15SA0018404). tries raises concerns about potential haemorrhagic risk. Finally, the proportion of dose adjustment related to changes in renal function CONT RIB UT ORS or age deserves a more complete approach. The differences among the countries might be explained by different national or regional rec- All other authors participated in the conception and design, interpre- ommendations, prescription patterns and characteristics of the tation of data, and revising the manuscript critically for important selected databases. Drug utilization studies based on several data- intellectual content. All authors read and approved the final bases across different countries using a standard protocol may help manuscript. to compare drug use and identify sources of variation, enabling health L.I. participated in the conception and design, interpretation of care decisions and supporting the rational use of medicines. data, and writing the paper. M.S. participated in the conception and design, interpretation of data and writing the paper. X.V. participated in the design, data analysis of SIDIAP database. E.B. participated in ACKNOWLEDGEMEN TS the conception and design, data analysis of SIDIAP database. M.R. The research leading to these results was conducted as part of the participated in the design, data analysis of Bavarian database. S.S. activities of the PE‐PV Consortium (Pharmacoepidemiology and and A.H. participated in the design, data analysis of AOK Nordwest Pharmacovigilance Consortium), which is a public academic partner- database. C.H., E.M.M., D.M. and L.M. L.‐M. participated in the design, ship coordinated by the University of Utrecht. The project has data analysis of BIFAP database. C.G. participated in the conception received support from the European Medicines Agency under the and design, interpretation of data. N.M., C.D. and R.L. participated in Framework service contract (nr EMA/2015/27/PH) with regard to the design, data analysis of EGB database. M.Aa., M.An. and the reopening of competition no3. M.L.D.B. participated in the design, data analysis of DNR data. R.G., K. Janhsen (Witten/Herdecke University, Alfred‐Herrhausen‐ R.v.d.H. and P.S. participated in the design, data analysis of Mondriaan Straße 50, 58448 Witten, Germany (UW/H)) and B. Grave (AOK and CPRD databases. O.K. and H.G. coordinated the whole project. NORDWEST, Kopenhagener Straße 1, 44269 Dortmund, Germany). R. Gerlach and M. Tauscher (National Association of Statutory Health DATA AVAILABILITY STATEMENT Insurance Physicians of Bavaria, Elsenheimerstr. 39, D‐80687 Munich, Germany). The data that support the findings of this study are available from The authors of the BIFAP database would like to acknowledge the third party (data owners). Restrictions apply to the availability of these excellent collaboration of the primary care general practitioners and data, which were used under license for this study. Data are available pediatricians, and also the support of the regional governments to from the authors with the permission of third party. the database. This study is based in part on data from the BIFAP fully financed by the Spanish Agency on Medicines and Medical Devices ORCID (AEMPS). Luisa Ibáñez https://orcid.org/0000-0002-1175-0574 The views and opinions expressed in this article are those of the Mònica Sabaté https://orcid.org/0000-0001-6206-1085 authors and do not necessarily reflect the official policy or position Consuelo Huerta https://orcid.org/0000-0003-4012-7826 of the EMA (European Medicines Agency) or one of its committees Nicholas Moore https://orcid.org/0000-0003-1212-2817 or working parties, or AEMPS (Agencia Española de Medicamentos y Morten Andersen https://orcid.org/0000-0001-7029-2860 Productos Sanitarios). Patrick Souverein https://orcid.org/0000-0002-7452-0477 Helga Gardarsdottir https://orcid.org/0000-0001-5623-9684 COMPETING INTER ESTS N.M.: Bordeaux PharmacoEpi has done 2 postauthorisation studies of RE FE RE NC ES comparative effectiveness and safety of DOAC compared to VKA, 1. Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC guidelines for the requested by the French regulatory authorities and financed by the management of atrial fibrillation developed in collaboration with respective marketing authorisation holders. These studies do not EACTS. Eur Heart J. 2016;37:2893‐2962. https://doi.org/10.1093/ concern drug utilisation and do not represent a conflict of interest. eurheartj/ehw210 2538 IBÁÑEZ ET AL. 2. Björck S, Palaszewski B, Friberg L, Bergfeldt L. Atrial Pharmacol. 2016;72(9):1125‐1134. https://doi.org/10.1007/s00228‐ fibrillation, stroke risk, and warfarin therapy revisited: a population‐ 016‐2078‐1 based study. Stroke. 2013;44(11):3103‐3108. https://doi.org/10. 17. Ho P, Brooy BL, Hayes L, Lim WK. Direct oral anticoagulants in frail 1161/STROKEAHA.113.002329 older adults: a geriatric perspective. Semin Thromb Hemost. 3. Haim M, Hoshen M, Reges O, Rabi Y, Balicer R, Leibowitz M. Prospec- 2015;41(4):389‐394. https://doi.org/10.1055/s‐0035‐1550158 tive national study of the prevalence, incidence, management and 18. Smeets HM, Kortekaas MF, Rutten FH, et al. Routine primary care outcome of a large contemporary cohort of patients with incident data for scientific research, quality of care programs and educational non‐valvular atrial fibrillation. J am Heart Assoc. 2015;4(1):e001486. purposes: the Julius general Practitioners' network (JGPN). BMC https://doi.org/10.1161/JAHA.114.001486 Health Serv Res. 2018;18(1):735. https://doi.org/10.1186/s12913‐ 4. Stewart S, Hart CL, Hole DJ, McMurray JJ. A population‐based study of 018‐3528‐5 the long‐term risks associated with atrial fibrillation: 20‐year follow‐up of the Renfrew/Paisley study. Am J Med. 2002;113(5):359‐364. 19. Schmidt M, Schmidt SAJ, Sandegaard JL, Ehrenstein V, Pedersen L, https://doi.org/10.1016/s0002-9343(02)01236-6 Sørensen HT. The Danish National Patient Registry: a review of con- tent, data quality, and research potential. Clin Epidemiol. 5. Hart RG, Pearce LA, Aguilar MI. Meta‐analysis: antithrombotic therapy 2015;7:449‐490. https://doi.org/10.2147/CLEP.S91125 to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146(12):857‐867. https://doi.org/10.7326/ 20. Kildemoes HW, Sørensen HT, Hallas J. The Danish National Prescrip- 0003-4819-146-12-200706190-00007 tion Registry. Scand J Public Health. 2011;39(Suppl 7):38‐41. https:// doi.org/10.1177/1403494810394717 6. Ruff CT, Giugliano RP, Braunwald E, et al. Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with 21. Schmidt M, Pedersen L, Sørensen HT. The Danish civil registration sys- atrial fibrillation: a meta‐analysis of randomised trials. Lancet (London, tem as a tool in epidemiology. Eur J Epidemiol. 2014;29(8):541‐549. England). 2014;383(9921):955‐962. https://doi.org/10.1016/S0140‐ https://doi.org/10.1007/s10654‐014‐9930‐3 6736(13)62343‐0 22. Jaunzeme J, Eberhard S, Geyer S. [How “representative” are SHI (stat- 7. Larsen TB, Rasmussen LH, Skjøth F, et al. Efficacy and safety of utory health insurance) data? Demographic and social differences and dabigatran etexilate and warfarin in “real‐world” patients with atrial similarities between an SHI‐insured population, the population of fibrillation: a prospective nationwide cohort study. J am Coll Cardiol. Lower Saxony, and that of the Federal Republic of Germany using 2013;61(22):2264‐2273. https://doi.org/10.1016/j.jacc.2013.03.020 the example of the AOK in Lower Saxony] Wie “repräsentativ” Sind 8. Giner‐Soriano M, Roso‐Llorach A, Vedia Urgell C, et al. Effectiveness GKV‐Daten? Bundesgesundheitsblatt ‐ Gesundheitsforsch ‐ and safety of drugs used for stroke prevention in a cohort of non‐ Gesundheitsschutz. 2013;56(3):447‐454. https://doi.org/10.1007/ valvular atrial fibrillation patients from a primary care electronic data- s00103‐012‐1626‐9 base. Pharmacoepidemiol Drug Saf. 2017;26(1):97‐107. https://doi. 23. Hoffmann F, Koller D. Verschiedene Regionen, verschiedene org/10.1002/pds.4137 Versichertenpopulationen? Soziodemografische und gesundheitsbe- 9. Komen J, Forslund T, Hjemdahl P, Andersen M, Wettermark B. Effects zogene Unterschiede zwischen Krankenkassen. Das Gesundheitswes. of policy interventions on the introduction of novel oral anticoagulants 2015;79(1):e1‐e9. https://doi.org/10.1055/s‐0035‐1564074 in Stockholm: an interrupted time series analysis. Br J Clin Pharmacol. 24. Mehring M, Donnachie E, Schneider A, et al. Impact of regional 2017;83(3):642‐652. https://doi.org/10.1111/bcp.13150 socioeconomic variation on coordination and cost of ambulatory 10. Hellfritzsch M, Husted SE, Grove EL, et al. Treatment changes among care investigation of claims data from Bavaria, Germany. BMJ Open. users of non‐vitamin K antagonist Oral anticoagulants in atrial fibrilla- 2017;7(10):e016218. https://doi.org/10.1136/bmjopen‐2017‐016218 tion. Basic Clin Pharmacol Toxicol. 2017;120(2):187‐194. https://doi. 25. de Abajo FJ, Gil MJ, Bryant V, Timoner J, Oliva B, García‐Rodríguez LA. org/10.1111/bcpt.12664 Upper gastrointestinal bleeding associated with NSAIDs, other drugs 11. Becattini C, Franco L, Beyer‐Westendorf J, et al. Major bleeding and interactions: a nested case‐control study in a new general practice with vitamin K antagonists or direct oral anticoagulants in real‐life. database. Eur J Clin Pharmacol. 2013;69(3):691‐701. https://doi.org/ Int J Cardiol. 2017;227:261‐266. https://doi.org/10.1016/j.ijcard. 10.1007/s00228‐012‐1386‐3 2016.11.117 26. García‐Gil Mdel M, Hermosilla E, Prieto‐Alhambra D, et al. Construc- 12. Maura G, Pariente A, Alla F, Billionnet C. Adherence with direct oral tion and validation of a scoring system for the selection of high‐ anticoagulants in nonvalvular atrial fibrillation new users and quality data in a Spanish population primary care database (SIDIAP). associated factors: a French nationwide cohort study. Inform Prim Care. 2011;19(3):135‐145. http://www.ncbi.nlm.nih.gov/ Pharmacoepidemiol Drug Saf. 2017;26(11):1367‐1377. https://doi. pubmed/22688222 org/10.1002/pds.4268 27. Williams T, van Staa T, Puri S, Eaton S. Recent advances in the utility 13. Loo SY, Dell'Aniello S, Huiart L, Renoux C. Trends in the prescription of and use of the general practice research database as an example of a novel oral anticoagulants in UK primary care. Br J Clin Pharmacol. UK primary care data resource. Ther Adv Drug Saf. 2012;3(2):89‐99. 2017;83(9):2096‐2106. https://doi.org/10.1111/bcp.13299 https://doi.org/10.1177/2042098611435911 14. Shrestha S, Baser O, Kwong WJ. Effect of renal function on dosing of non‐vitamin K antagonist direct Oral anticoagulants among 28. Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource profile: clin- patients with Nonvalvular atrial fibrillation. Ann Pharmacother. ical practice research datalink (CPRD). Int J Epidemiol. 2017;52(2):147‐153. https://doi.org/10.1177/1060028017728295 2015;44(3):827‐836. https://doi.org/10.1093/ije/dyv098 15. Pattullo CS, Barras M, Tai B, McKean M, Donovan P. New oral antico- 29. Bezin J, Duong M, Lassalle R, et al. The national healthcare system agulants: appropriateness of prescribing in real‐world setting. Intern claims databases in France, SNIIRAM and EGB: powerful tools for Med J. 2016;46(7):812‐818. https://doi.org/10.1111/imj.13118 pharmacoepidemiology. Pharmacoepidemiol Drug Saf. September 2016;2017(8):1‐9. https://doi.org/10.1002/pds.4233 16. Desmaele S, Steurbaut S, Cornu P, Brouns R, Dupont AG. Clinical trials with direct oral anticoagulants for stroke prevention in atrial fibrilla- 30. EMA. Http://www.Encepp.eu/encepp/viewResource.htm?id=2441. in tion: how representative are they for real life patients? Eur J Clin Each Database. R. ENCePP Protocol Register. Accessed August 8, 2019. IBÁÑEZ ET AL. 2539 31. Gardarsdottir H, Souverein PC, Egberts TCG, Heerdink ER. Construc- Routinedaten. Das Gesundheitswes. 2018;80(11):1000‐1005. https:// tion of drug treatment episodes from drug‐dispensing histories is doi.org/10.1055/s‐0043‐125069 influenced by the gap length. J Clin Epidemiol. 2010;63(4):422‐427. 48. OTC. Http://www.aesgp.eu/facts‐Figures/otc‐ingredients/?result= 32. Eurostat. http://ec.europa.eu/eurostat/web/population‐demography‐ name&multiselect=none&country=15#by‐Parameter. migration‐projections/population‐data/database. Accessed May 16, 49. Gasparini A, Evans M, Coresh J, et al. Prevalence and recognition of 2018. chronic kidney disease in Stockholm healthcare. Nephrol Dial Trans- 33. Harding SD, Sharman JL, Faccenda E, et al. The IUPHAR/BPS guide to plant. 2016;31(12):2086‐2094. https://doi.org/10.1093/ndt/gfw354 PHARMACOLOGY in 2018: updates and expansion to encompass the 50. Hartmann J, Weidmann C, Biehle R. Validation of SHI claims data new guide to IMMUNOPHARMACOLOGY. Nucl Acids Res. 2018;46: exemplified by gender‐specific diagnoses. Gesundheitswesen. D1091‐D1106. 2016;78(10):e53‐e58. https://doi.org/10.1055/s‐0035‐1565072 34. Alexander SPH, Fabbro D, Kelly E, et al. The Concise Guide to PHAR- 51. García‐Poza P, de Abajo FJ, Gil MJ, Chacón A, Bryant V, García‐ MACOLOGY 2017/18: enzymes. Br J Pharmacol. 2017;174:S272‐ Rodríguez LA. Risk of ischemic stroke associated with non‐steroidal S359. anti‐inflammatory drugs and paracetamol: a population‐based case‐ 35. Mueller T, Alvarez‐Madrazo S, Robertson C, Bennie M. Use of control study. J Thromb Haemost. 2015;13(5):708‐718. https://doi. direct oral anticoagulants in patients with atrial fibrillation in Scotland: org/10.1111/jth.12855 applying a coherent framework to drug utilisation studies. 52. Ruigómez A, Brauer R, Rodríguez LAG, et al. Ascertainment of acute Pharmacoepidemiol Drug Saf. 2017;26(11):1378‐1386. https://doi.org/ liver injury in two European primary care databases. Eur J Clin 10.1002/pds.4272 Pharmacol. 2014;70(10):1227‐1235. https://doi.org/10.1007/s00228‐ 36. Olesen JB, Sørensen R, Hansen ML, et al. Non‐vitamin K antagonist 014‐1721‐y oral anticoagulation agents in anticoagulant naïve atrial fibrillation 53. Maier B, Wagner K, Behrens S, et al. Comparing routine administrative patients: Danish nationwide descriptive data 2011‐2013. Europace. data with registry data for assessing quality of hospital care in patients 2015;17(2):187‐193. https://doi.org/10.1093/europace/euu225 with myocardial infarction using deterministic record linkage. BMC 37. Townsend N, Wilson L, Bhatnagar P, Wickramasinghe K, Rayner M, Health Serv Res. 2016;16(1):605. https://doi.org/10.1186/s12913‐ Nichols M. Cardiovascular disease in Europe: epidemiological update 016‐1840‐5 2016. Eur Heart J. 2016;37(42):3232‐3245. https://doi.org/10.1093/ 54. GKV‐Spitzenverband and BARMER. Bewertung der Kodierqualität von eurheartj/ehw334 vertragsärztlichen Diagnosen. https://www.gkv‐spitzenverband.de/ 38. Vlahović‐Palcevski V, Wetermark B, Ibáñez L, Vander Stichele R. media/dokumente/krankenversicherung_1/aerztliche_versorgung/ver- Comparison of drug utilization across different geographical areas. guetung_und_leistungen/klassifikationsverfahren/9_Endbericht_Kodi- In: Elseviers M, ed. Drug Utilization Research. Methods and erqualitaet_Hauptstudie_2012_12‐19.pdf. Applicationsfirst. Chichester, West Susex, UK: John Wiley & Sons, 55. Giroud M, Hommel M, Benzenine E, Fauconnier J, Béjot Y, Quantin C. Ltd; 2016:153‐159. Positive predictive value of French hospitalization discharge codes for 39. Rodríguez‐Bernal CL, Hurtado I, García‐Sempere A, Peiró S, Sanfélix‐ stroke and transient ischemic attack. Eur Neurol. 2015;74(1–2):92‐99. Gimeno G. Oral anticoagulants initiation in patients with atrial fibrilla- https://doi.org/10.1159/000438859 tion: real‐world data from a population‐based cohort. Front 56. Blin P, Dureau‐Pournin C, Lassalle R, Abouelfath A, Droz‐Perroteau C, Pharmacol. 2017;8:63. https://doi.org/10.3389/fphar.2017.00063 Moore N. A population database study of outcomes associated with 40. Desai NR, Krumme AA, Schneeweiss S, et al. Patterns of initiation of vitamin K antagonists in atrial fibrillation before DOAC. Br J Clin Oral anticoagulants in patients with atrial fibrillation— quality and cost Pharmacol. 2016;81(3):569‐578. https://doi.org/10.1111/bcp.12807 implications. Am J Med. 2014;127(11):1075‐1082.e1. https://doi.org/ 57. Giner‐Soriano M, Vedia Urgell C, Roso‐Llorach A, et al. Effectiveness, 10.1016/J.AMJMED.2014.05.013 safety and costs of thromboembolic prevention in patients with non‐ 41. Reinhold T, Rosenfeld S, Müller‐Riemenschneider F, et al. Patienten valvular atrial fibrillation: phase I ESC‐FA protocol study and baseline mit Vorhofflimmern in Deutschland. Herz. 2012;37(5):534‐542. characteristics of a cohort from a primary care electronic database. https://doi.org/10.1007/s00059‐011‐3575‐8 BMJ Open. 2016;6(1):e010144. https://doi.org/10.1136/bmjopen‐ 42. Michelena HI, Powell BD, Brady PA, Friedman PA, Ezekowitz MD. 2015‐010144 Gender in atrial fibrillation: ten years later. Gend Med. 2010;7(3):206‐217. https://doi.org/10.1016/j.genm.2010.06.001 SUPPORTING INFORMATION 43. EMA. https://www.ema.europa.eu/. Additional supporting information may be found online in the 44. Hernandez I, Zhang Y, Saba S. Comparison of the effectiveness and Supporting Information section at the end of the article. safety of Apixaban, dabigatran, rivaroxaban, and warfarin in newly diagnosed atrial fibrillation. Am J Cardiol. 2017;120(10):1813‐1819. https://doi.org/10.1016/j.amjcard.2017.07.092 45. Anselmino M, Battaglia A, Gallo C, et al. Atrial fibrillation and female How to cite this article: Ibáñez L, Sabaté M, Vidal X, et al. sex. J Cardiovasc Med. 2015;16(12):795‐801. https://doi.org/ Incidence of direct oral anticoagulant use in patients with 10.2459/JCM.0000000000000239 nonvalvular atrial fibrillation and characteristics of users in 6 46. Gómez‐Doblas JJ, Muñiz J, Martin JJA, et al. Prevalence of atrial fibril- European countries (2008–2015): A cross‐national drug utiliza- lation in Spain. OFRECE study results. Rev Española Cardiol (English Ed). 2014;67(4):259‐269. https://doi.org/10.1016/j.rec.2013.07.014 tion study. Br J Clin Pharmacol. 2019;85:2524–2539. 47. Carnarius S, Heuer J, Stausberg J. Diagnosenkodierung in deutschen https://doi.org/10.1111/bcp.14071 Arztpraxen aus klassifikatorischer Sicht: Eine retrospektive Studie mit