Portal de Periódicos da CAPES

Dabigatran adherence in atrial fibrillation patients during the first year after diagnosis: a nationwide cohort study

2015; Elsevier BV; Volume: 13; Issue: 4 Linguagem: Inglês

10.1111/jth.12845

1538-7933

Anders Gorst‐Rasmussen, Flemming Skjøth, Torben Bjerregaard Larsen, Lars Hvilsted Rasmussen, G. Y. H. Lip, Deirdre A. Lane,

Antiplatelet Therapy and Cardiovascular Diseases

Journal of Thrombosis and HaemostasisVolume 13, Issue 4 p. 495-504 Original ArticleFree Access Dabigatran adherence in atrial fibrillation patients during the first year after diagnosis: a nationwide cohort study A. Gorst-Rasmussen, A. Gorst-Rasmussen Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorF. Skjøth, F. Skjøth Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorT. B. Larsen, T. B. Larsen Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorL. H. Rasmussen, L. H. Rasmussen Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorG. Y. H. Lip, G. Y. H. Lip Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, Denmark University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Birmingham, UKSearch for more papers by this authorD. A. Lane, Corresponding Author D. A. Lane Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, Denmark University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Birmingham, UK Correspondence: Deirdre A. Lane, University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Dudley Road, Birmingham B18 7QH, UK. Tel.: +44 121 507 5080; fax: +44 121 554 4083. E-mail: deirdrelane@nhs.netSearch for more papers by this author A. Gorst-Rasmussen, A. Gorst-Rasmussen Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorF. Skjøth, F. Skjøth Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorT. B. Larsen, T. B. Larsen Aalborg AF Study Group, Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorL. H. Rasmussen, L. H. Rasmussen Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, DenmarkSearch for more papers by this authorG. Y. H. Lip, G. Y. H. Lip Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, Denmark University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Birmingham, UKSearch for more papers by this authorD. A. Lane, Corresponding Author D. A. Lane Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, Denmark University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Birmingham, UK Correspondence: Deirdre A. Lane, University of Birmingham, Centre for Cardiovascular Sciences, City Hospital, Dudley Road, Birmingham B18 7QH, UK. Tel.: +44 121 507 5080; fax: +44 121 554 4083. E-mail: deirdrelane@nhs.netSearch for more papers by this author First published: 16 January 2015 https://doi.org/10.1111/jth.12845Citations: 108 Manuscript handled by: I. Pabinger Final decision: F. R. Rosendaal, 8 January 2014 AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary Background There is a perception among physicians that lack of routine monitoring with non-vitamin K antagonist oral anticoagulants (NOACs) may lead to poor adherence to medication. We studied adherence during the first year of usage in a cohort of patients with newly diagnosed non-valvular atrial fibrillation (AF) started on the NOAC, dabigatran etexilate. Methods and results Nationwide Danish patient and prescription purchase registries were used to identify newly diagnosed AF patients taking dabigatran, comorbidities, and refill patterns under a twice-daily, one pill regimen. Adherence was characterized among remaining users (N = 2960) after 1 year using the proportion of days covered (PDC), gap rates and restart rates. The overall 1-year PDC was 83.9%, with 76.8% of patients having a 1-year PDC in excess of 80%. Patients with a CHA2DS2-VASc score ≥ 2 were more adherent to medication regimes than patients with a CHA2DS2-VASc score of 1 (PDC ratio, 1.12; 95% confidence interval [CI], 1.08–1.17) and generally patients with higher morbidity showed more adherence. Patients with prior bleeding were not less adherent to medication regimes than patients with no prior bleeding (PDC ratio, 1.02; 95% CI, 0.98–1.06). The overall gap rate was 1.4 gaps per year. There were no clear tendencies in gap rates among subgroups, although patients with higher morbidity tended to have slightly more, but shorter, gap periods. Conclusions More than 75% of patients were showed > 80% adherence to medication regimes during the first year. Patients with higher morbidity, including patients with a higher risk of stroke or bleeding, exhibited better adherence. This improvement may be attributable to more regular contact with the healthcare system. Introduction Adherence to medication is essential for treatment efficacy and to reduce adverse safety outcomes and is particularly important for oral anticoagulation (OAC) therapy 1. Lifelong OAC therapy is the preferred treatment for thromboprophylaxis in the majority of patients with non-valvular atrial fibrillation (AF) and those with ≥ 1 risk factor for stroke 2, 3. Conventionally, OAC therapy involved administration of vitamin-K antagonists, commonly warfarin, a drug characterized by a variable-dose regimen, a narrow therapeutic range and a strong potential for drug, food and alcohol interactions 4. Adherence to warfarin therapy can be low, and effectiveness, even among adherent patients, is often poor due to inadequate time within the therapeutic range 5. Non-vitamin K antagonist oral anticoagulants (NOACs, previously referred to as new or novel OACs 6) have been demonstrated to be at least as effective as warfarin for prevention of stroke in AF but with the convenience of a fixed-dose regimen 7. However, the convenience of NOACs has been tempered by speculation that the general lack of routine monitoring of dabigatran patients may lead to an overall poorer adherence to, and persistence with, medication 5, 8. This is a concern because even minor deviations from strict adherence to NOACs can significantly decrease effectiveness, due to the shorter half-lives 9, and this perception may affect the prescription of NOACs. Research indicates that adherence to medication in chronic diseases decreases considerably after the first 3 months of treatment 10, 11 but that simplified dosing regimens may lead to improved adherence for patients undergoing chronic therapy 12 but not necessarily translate into better persistence rates 13. Non-vitamin K antagonist oral anticoagulants for prevention of stroke in AF lend themselves well to adherence studies based on prescription purchase data because all patients share the same on-label dosing frequency. To date, there is limited information on persistence and adherence to NOACs in real-life clinical practice, with only two recent studies reporting on dabigatran use in AF patients 14, 15. One study reported 1-year persistence with dabigatran therapy of 63% (N = 1775), compared with 39% (N = 3370) for warfarin 14. Another small study (N = 99) reported that 88% were adherent (≥ 80% of the time) to dabigatran therapy over a varying follow-up period 15. Thus, large-scale studies are needed in order to critically evaluate adherence to NOACs among AF patients in settings representative of routine care. Given the less stringent requirements for physician monitoring with dabigatran, it is of special interest to assess longer-term adherence in the population of AF patients who apparently remain on dabigatran therapy and do not experience bleeding issues. This population of relatively unproblematic patients is presumably less closely monitored, and adherence may suffer as a result 16. In the present study, we used nationwide Danish registries to describe adherence to dabigatran therapy (in the sense of regular dose-taking according to recommendations 17) in the population of patients who began dabigatran therapy shortly after diagnosis of AF, and remained bleeding-free and plausible dabigatran users throughout the first year of therapy. Rather than trying to correlate adherence with outcomes, a complex and bias-prone exercise in the context of observational data 18, the aim of the present study was to provide a detailed, descriptive characterization of the extent and form of non-adherence in this particular patient population. Methods Data sources and study population Study data were obtained by linking three nationwide databases: (i) the Danish National Prescription Registry, which records purchase date, Anatomical Therapeutic Chemical (ATC) classification code and package size for every prescription purchase in Denmark 19; (ii) the Danish National Patient Register, which contains discharge diagnoses for all hospital admissions in Denmark 20; and (iii) the Danish Civil Registration System 21, which holds information on sex, date of birth, vital status and emigration status. All registry data were available up to 31 June 2014. We prioritized specificity over sensitivity for indirectly identifying a preliminary population of treatment-naïve patients beginning dabigatran therapy for non-valvular AF. First, we identified patients who had received a hospital diagnosis of non-valvular AF during the period 1 August 2011 to 30 June 2013. Both inpatient and outpatient diagnoses were included. We then restricted the cohort to those purchasing dabigatran (110 or 150 mg) within 30 days after diagnosis. We finally excluded those who had purchased other anticoagulants within 2 years of the first dabigatran purchase date (switchers) and patients with knee or hip arthroplasty in a 10-day window around the date of the first dabigatran purchase. The final study population was plausible dabigatran users at 1 year after the first purchase. To obtain this population, we excluded from the preliminary population patients who within the first year experienced a switch in treatment (warfarin, rivaroxaban or heparin purchase), bleeding event, death, emigration or end of follow-up (30 June 2014). Note that exclusion due to emigration or end of follow-up is unlikely to bias results, because it can be reasonably regarded as ‘independent censoring’, that is, the non-excluded study population is representative of the hypothetical complete population. The remaining patients were included in the final study population, irrespective of their actual purchase history in the year following the first dabigatran purchase. This entailed a ‘presumption of persistence’, in accordance with the recommendation of lifelong anticoagulant therapy for the majority of our study population 2. Note that this final patient population is a selected population by definition and does not reflect all dabigatran users. Instead, it reflects the large subgroup of patients who are relatively ‘unproblematic’, in the sense that they neither switch therapy nor experience bleeding issues during the first year of therapy. Using the registries, we addressed the following clinical question: what was the extent of adherence to medication among AF patients who ought to have been receiving dabigatran therapy for 1 year, and if non-adherence was evident, what pattern did this follow? Specifically, we analyzed non-adherence and treatment gap durations during the first year of therapy in a combined fashion, in order to identify subgroups with decreased adherence and to structurally characterize non-adherence, for example, assess whether it resulted from a few long or many short gaps. Patient baseline characteristics Information on comorbidities and concomitant medications at date of first dabigatran purchase was obtained from the registries. Past-year cardiovascular prescriptions were determined from prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. History of bleeding was defined in terms of International Classification of Diseases [ICD] revision 10 (ICD-10) codes as a composite of: intracranial (I60, I61, I62, S063C, S064, S065, S066), major (D62, J942, H113, H356, H431, N02, N95, R04, R31, R58) or gastrointestinal bleeding (K250, K260, K270, K280, K290). History of stroke was defined as the composite of stroke (I63, I649), systemic embolic event (SE) (I74) or transient ischemic attach (TIA) (G45). Hospitalization was defined as any-cause hospitalization for at least one night. Lastly, we calculated a registry-based CHA2DS2-VASc score following the approach previously described 22. Definition of outcome measures Adherence to dabigatran was defined as one pill twice daily. Adequate adherence to dabigatran was categorized as adherence of > 80% 23. A gap day was defined as any day during follow-up where dabigatran would be unavailable under a twice-daily, one pill regimen. A gap period was a sequence of consecutive gap days. We adopted a conservative default grace period of zero days, reflecting the short half-life of dabigatran 9. More importantly, the short grace period, alongside the ‘presumption of persistence’ implicit in the definition of the 1-year study population, implied that all gap periods, long and short, contributed towards increased non-possession of medication. These conventions ensured a high degree of sensitivity to gaps, a preferable starting point when gap periods are an independent target of analysis. To account for medication provided during hospitalization, days with hospitalization were excluded from calculations of dabigatran availability. For each patient, we used the proportion of days covered (PDC) to describe 1-year adherence, that is, the relative number of non-gap days during the first year. The gap count was defined as the number of gaps occurring during the first year of therapy, as exemplified in Fig. 1. Figure 1Open in figure viewerPowerPoint Dabigatran therapy histories for five example patients with 1 year of follow-up. Solid lines indicate periods during which dabigatran would be available under a twice-daily one-pill regimen. Gap counts refer to the number of periods when dabigatran would be unavailable; PDC refers to the relative period during which dabigatran would be available, that is, proportion of days covered. Within each gap period, we measured time from the first gap day until the next dabigatran purchase (restart time). Restart rates were classified as: rapid, restart < 1 week after stop; intermediate, restart 1–8 weeks after stop; and late, more than 8 weeks after stop. To characterize the distribution of restart times among patients at risk of reacquisition, we used the hazard rate of restarting. Statistical analysis Patient characteristics at baseline (time of first dabigatran purchase) were summarized descriptively. We calculated 1-year PDCs according to clinical subgroups and used Poisson regression with robust variance estimation to obtain PDC ratios and confidence intervals according to subgroups. We also calculated the average number of gaps (1-year gap rate) within subgroups. To characterize variation in gap patterns between subgroups (i.e. whether non-adherence was attributable to many small versus few large gap periods), we estimated 1-year gap rate ratios by using Poisson regression, adjusting for the 1-year PDC. The resulting effect measure could be interpreted as the relative gap count when comparing a typical subgroup patient with a patient with identical PDC from a reference group. Accordingly, a gap rate ratio < 1 (respectively, > 1) implied that the non-reference patient on average arrived at their PDC through comparatively less numerous but longer (respectively, more but shorter) gaps. To analyze gap periods, we used weekly running averages to smooth the crude restart rates, while restart probabilities were estimated using the Kaplan–Meier method. Cox proportional hazards regression was used to calculate restart rate ratios according to subgroups, stratified according to the following three time periods: week 1 after gap start (rapid restart), weeks 1–8 after gap start (moderate restart) and more than 8 weeks after gap start (slow restart). Confidence intervals (CIs) were estimated using cluster-robust methods to account for patients contributing multiple gap periods. Statistical analyses were performed using R version 3.2.0 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided P-value < 0.05 was considered statistically significant. Results The final study population of plausible dabigatran users at 1 year (N = 2960) was obtained from the preliminary study population of patients started on dabigatran due to AF (N = 3761) by excluding all those who, during the first year: died (N = 232; 6.2%), experienced a bleeding event (N = 99; 2.6%) or switched to warfarin, rivaroxaban or heparin (N = 470; 12.5%) (see Fig. 2). Figure 2Open in figure viewerPowerPoint Flowchart of study population selection process. Baseline patient characteristics are shown in Table 1. Patients started on the 110 mg dose were substantially older (mean age 81.1 years) than patients started on the 150 mg dose (mean age 66.4 years). Baseline characteristics for the excluded patients in the preliminary study population were similar (data not shown). Female patients were generally older than men (mean age 75.3 vs. 69.4 years) and had more comorbidities (data not shown). Table 1. Baseline demographic and clinical characteristics Number of patients, N 2960 Female sex, % (N) 45.8 (1356) Age, years, mean (SD) 72.1 (10.8) Age, years, % (N) ≤ 54 6.4 (189) 55–64 17.2 (510) 65–74 34.9 (1034) ≥ 75 41.5 (1227) Initial dose 150 mg, % (N) 61.1 (1812) CHA2DS2-VASc score, mean (SD) 2.8 (1.6) CHA2DS2-VASc score, % (N) 0 7.0 (206) 1 14.2 (419) ≥ 2 78.9 (2335) History of bleeding, % (N) 11.7 (345) History of stroke/SE/TIA, % (N) 18.6 (550) Cardiovascular prescriptions, past year* SD, standard deviation; CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. , % (N) 68.5 (2028) Hospitalization, past year, % (N) 80.8 (2392) SD, standard deviation; CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. Adherence 1 year after dabigatran initiation, PDC The 1-year mean PDC was 83.9% (standard deviation [SD] 27.7%), which increased to 84.9% when incorporating a 3-day grace period. Figure 3 shows a graph of the proportion of patients above a given PDC when the PDC varies from 0 to 100%. The proportion of patients with a 1-year PDC above 80% was 76.8%, and 37.2% of patients had a PDC of 100%. Figure 3Open in figure viewerPowerPoint Proportion of patients above a given level of proportion of days covered (PDC). Solid line represents the result for the main analysis; dotted line represents the analysis incorporating a 3-day grace period. The dashed line marks the proportion of patients with a PDC above 80%. One-year PDCs according to subgroups are shown in Fig. 4. Female patients were significantly more adherent than male patients (PDC ratio, 1.06; 95% CI, 1.03–1.08). Patients who were regular users of cardiovascular drugs were significantly more adherent to medication than non-regular medication users (PDC ratio, 1.12; 95% CI, 1.09–1.15), as were patients with a history of stroke/SE/TIA (PDC ratio, 1.09; 95% CI, 1.06–1.12). Patients with CHA2DS2-VASc score ≥ 2 were also more adherent to medication than the reference group with one CHA2DS2-VASc risk factor (PDC ratio, 1.12; 95% CI, 1.08–1.17). Figure 4Open in figure viewerPowerPoint Proportion of days covered (PDC) and PDC ratios (calculated using robust Poisson regression) according to clinical subgroups during the first year of dabigatran therapy. Overall PDC was 83.9%. CI, confidence interval; PDC, proportion of days covered; CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *PDC incorporating a 3-day grace period. †Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. Adherence among patients with a history of bleeding was not decreased compared with patients with no history of bleeding (PDC ratio, 1.02; 95% CI, 0.98–1.06). Younger patients (aged < 55 years) and those with a CHA2DS2-VASc score of 0 were substantially less adherent to dabigatran than their respective reference groups, displaying significant reductions in PDC in excess of 18%. Treatment gap periods The annual gap rate in the overall population was 1.4 gaps per year (SD 1.6), which decreased to 1.0 when incorporating a 3-day grace period. Overall, the gap rate ratios indicated little difference in how patients arrived at their PDC in terms of short or long gaps. Patients originally started on the 150 mg dose had less numerous but longer gap periods than patients started on a 110 mg dose, as reflected by a gap rate ratio of 0.94 (95% CI, 0.87–1.01) (see Fig. 5). A similar tendency was seen among patients who were regular users of cardiovascular drugs compared with non-users (gap rate ratio, 0.91; 95% CI, 0.84–0.98), and patients recently hospitalized (gap rate ratio, 0.89; 95% CI, 0.81–0.97). Elderly patients (≥ 75 years) had a tendency for shorter but more numerous gaps (gap rate ratio, 1.08; 95% CI, 0.97–1.20) although this was not significant. Figure 5Open in figure viewerPowerPoint Gap rates and gap rate ratios (calculated using robust Poisson regression) according to clinical subgroups during the first year of dabigatran therapy (assuming a one pill twice daily regimen). Adjusted for PDC (categorical: 0–50%, 50–60%, 60–70%, 70–80%, 80–90%, 90–100%). Overall gap rate was 1.4. CI, confidence interval; CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *Gap rate incorporating a 3-day grace period. †Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. Restart rates for patients entering a gap period The smoothed restart rate is shown in Fig. 6, alongside the restart probability (dashed line). The restart rate was high immediately after the start of the gap period but decreased rapidly. After around 60 days, the restart rate was effectively zero, as indicated by the flattening of the restart probability curve. Hence, if therapy was not restarted within 60 days, it was unlikely to restart at all. Figure 6Open in figure viewerPowerPoint Dabigatran therapy restart rates (smooth line) and Kaplan–Meier restart probability (dashed line) during gap periods occurring in the first year of dabigatran therapy. Relative restart rates stratified by restart timing are shown in Table 2 (rapid, restart < 1 week after stop; intermediate, restart 1–8 weeks after stop; late, more than 8 weeks after stop). Patients with a CHA2DS2-VASc score of 0 were significantly less likely to restart in any of the time strata. Patients who had been hospitalized in the past year and patients started on the 150 mg dose were significantly less likely to be rapid restarters than their respective reference groups. On the other hand, patients with a history of stroke/SE/TIA, patients with CHA2DS2-VASc ≥ 2, and patients who were regular users of cardiovascular drugs, were generally more likely to restart therapy. Finally, the gap period index was strongly associated with the timing of restarts, with the rate of intermediate (restart 1–8 weeks after stop) or late (more than 8 weeks after stop) restart increasing with the gap period index. Patients with ≥ 3 gap periods had an almost 3-fold increased rate of late restarts compared with patients in their first gap period. Table 2. Cox model restart rate ratios, stratified by time since start of gap period Restart rate ratio (95% confidence interval) Rapid restart < 1 week Intermediate restart 1–8 weeks Late restart ≥ 8 weeks Female sex 1.15 (1.03–1.29) 1.12 (0.99–1.26) 0.97 (0.69–1.35) Age, years ≤ 54 (vs. 55–64) 0.69 (0.51–0.93) 0.72 (0.53–0.99) 0.73 (0.39–1.36) 65–74 (vs. 55–64) 1.24 (1.06–1.47) 1.39 (1.16–1.66) 1.74 (1.10–2.75) ≥ 75 (vs. 55–64) 1.24 (1.06–1.45) 1.45 (1.22–1.72) 1.40 (0.88–2.22) Original dose 150 mg 0.90 (0.81–1.01) 0.81 (0.72–0.91) 0.83 (0.59–1.17) CHA2DS2-VASc score 0 (vs. 1) 0.62 (0.47–0.81) 0.65 (0.49–0.88) 0.49 (0.25–0.94) ≥ 2 (vs. 1) 1.12 (0.95–1.31) 1.39 (1.16–1.68) 1.46 (0.95–2.23) History of bleeding 0.98 (0.83–1.17) 0.98 (0.81–1.19) 1.49 (0.96–2.33) History of stroke/SE/TIA 1.14 (0.99–1.32) 1.34 (1.15–1.57) 0.97 (0.59–1.59) Cardiovascular prescriptions, past year* CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. 1.14 (1.01–1.28) 1.29 (1.14–1.47) 1.32 (0.94–1.84) Hospitalization, past year 0.91 (0.80–1.04) 1.00 (0.86–1.17) 1.10 (0.74–1.64) Gap period index 2nd (vs. 1st) 1.05 (0.95–1.17) 1.88 (1.63-2.16) 2.23 (1.51–3.30) 3rd or later (vs. 1st) 0.86 (0.76–0.96) 1.97 (1.73–2.23) 4.12 (2.71–6.27) CHA2DS2-VASc, Congestive heart failure or left ventricular systolic dysfunction, Hypertension, Age ≥ 75 (Doubled), Diabetes, Stroke/TIA (Doubled)-Vascular disease, Age 65–74, and Sex category (female) stroke risk score, previously described 22; SE, systemic embolic event; TIA, transient ischemic attack. *Prescription purchases in the code C section of the Anatomical Therapeutic Chemical Classification System. Sensitivity analyses We conducted several sensitivity analyses: using less stringent inclusion criteria (defining the preliminary population in terms of dabigatran purchase within 60 days after AF diagnosis, rather than 30 days after); using more detailed categorization of continuous covariates; and considering the population of plausible dabigatran users after 180 days rather than 360 days (see Supporting Information). Conclusions from these analyses were similar to those of the main analysis. Discussion In this large nationwide cohort study of patterns of adherence to dabigatran therapy in AF patients during the first year after diagnosis, our principal findings were as follows: first, more than 75% of patients adhered to dabigatran therapy for > 80% of the time during the first year of usage. Second, patients at high risk of stroke (CHA2DS2-VASc score ≥ 2) were significantly more adherent than those with a CHA2DS2-VASc score of 1; similar observations applied for other patient groups with relatively higher morbidity (e.g. patients with recent hospitalizations or patients who took several cardiovascular medications), including patients with prior bleeding. Third, women were significantly more adherent to therapy than men. The overall PDC of 83.9% in the present study was higher than in a re