Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Non-persistence to Oral Anticoagulation Treatment in Patients with Non-valvular Atrial Fibrillation in the USA

Non-persistence to Oral Anticoagulation Treatment in Patients with Non-valvular Atrial... Background Studies have shown that patients with non-valvular atrial fibrillation (NVAF) who discontinue oral anticoagu- lants (OACs) are at higher risk of complications such as stroke. Objective This analysis compared the risk of non-persistence with OACs among patients with NVAF. Methods Adult patients with NVAF who initiated apixaban, dabigatran, rivaroxaban, or warfarin were identified using 01JAN2013–30JUN2019 data from Centers for Medicare and Medicaid Services and four US commercial claims databases. Non-persistence was defined as discontinuation (no evidence of index OAC use for ≥  60 days from the last days’ supply) or switch to another OAC. Kaplan–Meier curves were generated to illustrate time to non-persistence along with cumulative incidences of non-persistence. Baseline and time-varying covariates were evaluated, and adjusted Cox proportional hazards models were used to evaluate non-persistence risk. Results In total, 363,823 patients receiving apixaban, 57,121 receiving dabigatran, 282,831 receiving rivaroxaban, and 317,337 receiving warfarin were included. Of these, 47–72% discontinued/switched OAC therapy within an average 9-month follow-up. Apixaban was associated with a lower risk of non-persistence than were dabigatran (hazard ratio [HR] 0.62; 95% confidence interval [CI] 0.61– 0.62), rivaroxaban (HR 0.76; 95% CI 0.75–0.76), and warfarin (HR 0.74; 95% CI 0.74–0.75). Dabigatran was associated with a higher risk of non-persistence than were warfarin (HR 1.21; 95% CI 1.19–1.22) and rivar- oxaban (HR 1.23; 95% CI 1.22–1.25), and rivaroxaban was associated with a lower risk of non-persistence than was warfarin (HR 0.98; 95% CI 0.97–0.98). Clinical events (stroke/systemic embolism and major bleeding [MB]) during follow-up were predictors of non-persistence (stroke HR 1.57; 95% CI 1.53–1.61; MB HR 2.96; 95% CI 2.92–3.00). Conclusion In over one million patients with NVAF, our results suggest differences in anticoagulation treatment persistence across OAC agents, even after accounting for clinical events after OAC initiation. It is important for clinicians and patients to take these differences into consideration, especially as non-persistence to OAC therapy is associated with thromboembolic complications. Key Points This study examined non-persistence with oral antico- * Allison Keshishian agulants (OACs; apixaban, dabigatran, rivaroxaban, and akeshishian@statinmed.com warfarin) among patients with non-valvular atrial fibril- Bristol Myers Squibb, Lawrenceville, NJ, USA lation (NVAF). OAC non-persistence among patients Department of Pharmacy and Therapeutics, School with NVAF has previously been associated with an of Pharmacy, University of Pittsburgh, Pittsburgh, PA, USA increased risk of adverse events. Pfizer, New York, NY, USA The risk of non-persistence varied among patients with STATinMED Research, 4110 Varsity Dr, Ann Arbor, NVAF; baseline characteristics and clinical events occur- MI 48108, USA ring after OAC initiation were significant predictors of New York City College of Technology, City University non-persistence. of New York, New York, NY, USA Vol.:(0123456789) A. D. Dhamane et al. and Medicaid Services data (1 January 2012–31 Decem- 1 Introduction ber 2017), Truven Mar ketScan Commercial Claims and Encounter (1 January 2012–30 June 2019), IMS PharMetrics Non-valvular atrial fibrillation (NVAF) is the most common Plus™ (1 January 2012–31 March 2019), Optum Clinfor- heart dysrhythmia diagnosed in the USA and an independ- matics™ Data Mart (1 January 2012–31 March 2019), and ent risk factor for stroke [1]. The prevalence of NVAF has the Humana Research Database (1 January 2012–31 March been increasing in the USA and is expected to continue to 2019). Of note, patients with Medicare supplemental plans increase substantially because of the aging of the popula- in Truven MarketScan and IMS PharMetrics Plus data were tion [2]. Atrial fibrillation (AF), in turn, increases the risk not included in the study to avoid potential duplicates with of mortality and morbidities such as major adverse cardiac Medicare Part A and Part B. More details on the datasets and events, arterial thromboembolism, and ischemic stroke [3, 4]. pooling method have been published [15, 16]. While vitamin K antagonists (VKAs), such as warfarin, have been the anticoagulant treatment of choice for several 2.2 Patient Selection years, they have unstable pharmacokinetic profiles, with wide interpatient variability and extensive interactions, We identified adult patients with NVAF between 1 Janu- requiring routine blood monitoring for dose adjustment ary 2012 and 30 June 2019 using inpatient and outpatient [5]. Clinical trials have demonstrated that the reduction in claims, and selected patients with a pharmacy claim for risk of stroke and bleeding with direct oral anticoagulants apixaban, dabigatran, edoxaban, rivaroxaban, or warfarin (DOACs: apixaban, dabigatran, edoxaban, rivaroxaban) is during the identification period between 1 January 2013 and similar or superior to that with warfarin, and DOACs have 30 June 2019. Each patient’s first OAC was labeled as their been approved in the USA to reduce the risk of stroke among “index therapy,” and the date of the first OAC prescription patients with NVAF [6–9]. claim was labeled as the “index date.” Patients prescribed Studies have also shown that patients who discontinue edoxaban were not included in the final analysis because of oral anticoagulants (OACs) are at increased risk of adverse the small sample size (N = 1629). Patients were required to cardiovascular effects, thromboembolic events, minor and have continuous medical and pharmacy health plan enroll- major bleeds, and all-cause mortality [10, 11]. Gradual ment for ≥ 12 months before the index date [17]. declines in persistence after the initial prescriptions among We excluded patients with any OAC treatment within patients with NVAF have been noted, with some studies 12 months before the index date to ensure they were treat- reporting a wide range of persistence, from 55 to 69%, after ment naïve. We also excluded patients with evidence of val- 12 months [12]. Although DOACs have fewer drug interac- vular heart disease, venous thromboembolism, transient AF tions than warfarin and do not require international normal- (pericarditis, hyperthyroidism, or thyrotoxicity), or heart valve ized ratio (INR) monitoring, real-world evidence rates of replacement or transplant during the baseline period; preg- discontinuation 12 months after treatment initiation have nancy during the study period; or hip/knee replacement sur- been estimated at 55% [13]. Results from a meta-analysis of gery within 6 weeks before the index date. To give all patients the DOAC randomized controlled trials indicated apixaban the possibility of discontinuing/switching index treatment, we had lower discontinuation rates than OACs, whereas war- also excluded patients with more than one OAC on the index farin had lower rates than dabigatran and rivaroxaban [14]. date and those with fewer than 60 days of follow-up, based on While studies have evaluated OAC persistence, real-world continuous health plan enrollment post-index. studies comparing persistence across DOACs and warfarin in large populations are lacking. Therefore, the objectives of this study were to compare the incidence and risk of non- 2.3 Outcome Measures persistence across DOACs and warfarin and identify pre- dictors of non-persistence among patients with NVAF who Patient data were assessed from the day after the index date initiated OAC therapy. until the earliest of the following: end of continuous medical and pharmacy enrollment, death (when available), or end of the study period. The primary outcome of interest was 2 Methods non-persistence, defined as either discontinuation or switch, with discontinuation defined as no evidence of the index 2.1 Data Sources OAC prescription for ≥ 60 days from the last day of supply of the last filled prescription and switch defined as filling a In this study, we pooled five large national claims databases prescription for an OAC other than the index drug during with the latest available data at the time of application. They the follow-up period within ± 60 days of last days’ supply included 100% fee-for-service US Centers for Medicare from the index OAC. Oral Anticoagulant Non-persistence in Patients with NVAF Baseline variables included demographics (age, sex, US 3 Results geographic region) and clinical characteristics such as clini- cal risk scores, prior stroke/systemic embolism (SE), prior 3.1 Baseline Characteristics bleeding, comorbidities, and baseline medication use. In total, 1,021,112 adults with NVAF were newly prescribed an OAC during the study period (Fig. 1). Of those patients, 2.4 Statistical Methodology 363,823 (35.6%) initiated apixaban with a mean ± standard deviation follow-up of 605 ± 410.5 days; 57,121 (5.6%) ini- Descriptive analysis of clinical and demographic variables was tiated dabigatran with a mean follow-up of 939.3 ± 566.4 conducted for patients prescribed DOACs or warfarin. Rates of days; 282,831 (27.7%) patients initiated rivaroxaban with non-persistence among patients with NVAF who initiated an a mean follow-up of 799.7  ±  524.4 days; and 317,337 OAC during the identification period were calculated during (31.1%) patients initiated warfarin with a mean follow-up the follow-up period overall and at 12 months post-index date. of 892  ±  543.7 days. Approximately 53% of patients in Unadjusted Kaplan–Meier survival curves were generated to all treatment groups combined were male. The mean age illustrate time to non-persistence in addition to cumulative of patients receiving apixaban, dabigatran, rivaroxaban, incidence of non-persistence. Cox proportional hazard mod- and warfarin was 75.1 ± 10.6, 73.4 ± 10.3, 73.5 ± 10.7, els with robust sandwich estimates were used to compare the and 76.7 ± 9.3 years, respectively. Patients initiating war- risk of non-persistence across treatment groups while adjust- farin were older and were at higher risk in terms of the ing for patient covariates. Baseline variables included in the CHA DS -VASc score (4.5 ± 1.8) and HAS-BLED score 2 2 model were patient demographics, index OAC prescription, (3.2  ±  1.4) (see Table  1 for CHA DS -VASc and HAS- 2 2 and clinical characteristics such as prior stroke/SE, prior bleed- BLED definitions) and had higher mean Deyo– Charlson ing, comorbidities, and baseline medication use. Comorbidity Index scores of 3.4 ± 3.0. For those receiving A secondary analysis was conducted taking into considera- apixaban, dabigatran, or rivaroxaban, 22.6% (2.5 mg), 15.6% tion clinical events occurring after the index date, including (15.2% on 75 mg; 0.4% on 110 mg), and 25.8% (5.4% on 10 stroke/SE hospitalizations, major bleeding (MB) hospitaliza- mg; 20.4% on 15 mg) had lower dosage regimens, respec- tions, new acute renal failure (considered new when it occurred tively (Table 1). for the first time within a timeframe of 6 months), new chronic renal failure (considered new when it occurred for the first 3.2 Unadjusted Cumulative Incidence time after the index date), new cancer diagnoses (considered of Non‑persistence new when it occurred for the first time within a timeframe of 6 months), and cardioversions and catheter ablations [18]. The cumulative incidence of non-persistence at 3 months The diagnosis codes used for stroke/SE and MB hospitaliza- was 21.2, 33.9, 28.8, and 26.7% for patients receiving tions were based on validated administrative claims-based apixaban, dabigatran, rivaroxaban, or warfarin, respectively algorithms and the International Society on Thrombosis and (Fig. 2). The cumulative incidence of non-persistence at 12 Haemostasis definition of MB [19, 20]. months was 42.7, 58.9, 52.2, and 51.3% for patients receiv- We conducted three sensitivity analyses: ing apixaban, dabigatran, rivaroxaban, or warfarin, respec- tively (Fig. 2). Most of the non-persistence was because of • Follow-up was limited to the first 12 months after the discontinuation of therapy, but 3–9% of patients switched index date. The shorter follow-up period allowed us to treatment, with dabigatran having the highest switch rate focus on short-term persistence. (Table 2). Non-persistence was reanalyzed using a ≥ 30-day dis- continuation gap instead of ≥ 60 days, which allowed us to observe how a shorter gap period would alter non- 3.3 Primary Analyses Controlling for Baseline persistence outcomes [21]. Characteristics Warfarin non-persistence was reanalyzed with the inclu- sion of INR records to extend warfarin treatment lines. Figure  3 and Table  3 show the results of primary analy- Patients receiving warfarin were considered discontinued ses with Cox models that adjusted for demographic and if the gap between two consecutive warfarin prescriptions clinical covariates defined at baseline. When compared or from the last prescription to end of study was longer with those receiving warfarin, patients initiating apixaban than 60 days and they did not have INR measurements (adjusted hazard ratio [aHR] 0.74; 95% confidence interval at least every 42 days [22, 23]. This sensitivity analysis [CI] 0.74–0.75; P < 0.001) or rivaroxaban (aHR 0.98; 95% allowed us to take into consideration dose adjustments CI 0.97–0.98; P < 0.001) were 26 and 2% less likely to be or possible additional prescriptions. A. D. Dhamane et al. Fig. 1 Patient selection criteria. The patient selection criteria yielded > 1,000,000 patients with non-valvular atrial fibrillation prescribed either apixaban, dabigatran, rivaroxa- ban, or warfarin. Patients were stratified into cohorts based on their index OAC. *Edoxaban was not included in the study because it received US FDA approval in 2015 so the sample size was small (N = 1629). AF atrial fibrillation, ICD-9/10-CM International Classification of Diseases, Ninth/Tenth Revision, Clinical Modification, OAC oral anticoagulant, VTE venous thromboembolism non-persistent, respectively. Patients receiving dabigatran results when time-varying covariates were included in Cox (aHR 1.21; 95% CI 1.19–1.22; P < 0.001) were 21% more models. The HRs for OACs remained consistent after the likely to be non-persistent than those receiving warfarin. inclusion of time-varying covariates. All time-varying When compared with patients receiving rivaroxaban, clinical characteristics were significant predictors of non- those initiating apixaban (aHR 0.76; 95% CI 0.75–0.76; persistence: stroke/SE aHR 1.57; 95% CI 1.53–1.61; MB P  <  0.001) were 24% less likely to be non-persistent. aHR 2.96; 95% CI 2.92–3.00; new acute renal failure Those receiving dabigatran (aHR 1.23; 95% CI 1.22–1.25; aHR 1.44; 95% CI 1.42–1.46; new chronic renal failure P < 0.001) were 23% more likely to be non-persistent than aHR 1.14; 95% CI 1.12–1.15, new cancer aHR 1.22; 95% those receiving rivaroxaban. When compared with patients CI 1.20–1.25; and cardioversions and catheter ablations receiving dabigatran, those initiating apixaban (aHR 0.62; aHR 1.17; 95% CI 1.15–1.19. All HRs were significant 95% CI 0.61–0.62; P < 0.001) were 38% less likely to be at P < 0.001. non-persistent. Older age and history of hypertension and stroke/SE were 3.5 Sensitivity Analyses predictors of a lower likelihood of non-persistence (Table 3). Patient use of statins was also a predictor of reduced risk of In the first sensitivity analysis, follow-up was limited to non-persistence. the first 12 months after index treatment initiation. The results are shown in Tables 2 and 4 in the ESM. The results 3.4 Secondary Analyses Controlling for Baseline were generally consistent with those of the main analysis, Characteristics and Time‑Dependent Variables except when comparing rivaroxaban and warfarin: when follow-up was limited to the first 12 months, rivaroxa- Table 1 in the electronic supplementary material (ESM) ban was associated with a 3% increase in non-persistence shows the proportion of patients with time-varying clinical compared with warfarin (aHR 1.03; 95% CI 1.02–1.04; events after the index date, and Table 4 shows the model P < 0.001). Oral Anticoagulant Non-persistence in Patients with NVAF Table 1 Baseline characteristics and treatment follow-up Characteristics Warfarin cohort Apixaban cohort P value Dabigatran cohort P value Rivaroxaban cohort P value (N = 317,337), reference (N = 363,823) (N = 57,121) (N = 282,831) Age, years 76.7 ± 9.3 75.1 ± 10.6 < 0.001 73.4 ± 10.3 < 0.001 73.5 ± 10.7 < 0.001  18–54 5630 (1.8) 13,780 (3.8) < 0.001 2735 (4.8) < 0.001 14,352 (5.1) < 0.001  55–64 17,296 (5.5) 35,915 (9.9) < 0.001 6169 (10.8) < 0.001 32,520 (11.5) < 0.001  65–74 104,121 (32.8) 117,039 (32.2) < 0.001 21,238 (37.2) < 0.001 100,054 (35.4) < 0.001  ≥ 75 190,290 (60.0) 197,089 (54.2) < 0.001 26,979 (47.2) < 0.001 135,905 (48.1) < 0.001 Sex  Male 165,668 (52.2) 188,252 (51.7) < 0.001 32,309 (56.6) < 0.001 156,327 (55.3) < 0.001  Female 151,669 (47.8) 175,570 (48.3) < 0.001 24,812 (43.4) < 0.001 126,503 (44.7) < 0.001 US geographic region  Northeast 60,061 (18.9) 60,877 (16.7) < 0.001 11,430 (20.0) < 0.001 50,509 (17.9) < 0.001  North central 98,273 (31.0) 81,283 (22.3) < 0.001 13,225 (23.2) < 0.001 68,069 (24.1) < 0.001  South 98,706 (31.1) 161,072 (44.3) < 0.001 22,184 (38.8) < 0.001 113,057 (40.0) < 0.001  West 59,688 (18.8) 60,182 (16.5) < 0.001 10,148 (17.8) < 0.001 50,634 (17.9) < 0.001  Other 609 (0.2) 409 (0.1) < 0.001 134 (0.2) 0.035 562 (0.2) 0.552 Baseline comorbidity  Deyo–Charlson Comor- 3.4 ± 3.0 2.9 ± 2.8 < 0.001 2.5 ± 2.5 < 0.001 2.6 ± 2.6 < 0.001 bidity Index   CHA DS -VASc score 4.5 ± 1.8 4.1 ± 1.9 < 0.001 3.8 ± 1.9 < 0.001 3.8 ± 1.9 < 0.001 2 2  HAS-BLED score 3.2 ± 1.4 3.0 ± 1.4 < 0.001 2.8 ± 1.3 < 0.001 2.9 ± 1.3 < 0.001  Bleeding history 71,975 (22.7) 63,850 (17.5) < 0.001 9339 (16.3) < 0.001 48,054 (17.0) < 0.001  Congestive heart failure 108,185 (34.1) 99,373 (27.3) < 0.001 13,521 (23.7) < 0.001 67,500 (23.9) < 0.001  Diabetes mellitus 125,920 (39.7) 125,450 (34.5) < 0.001 19,679 (34.5) < 0.001 94,611 (33.5) < 0.001  Hypertension 270,847 (85.3) 309,483 (85.1) 0.001 47,463 (83.1) < 0.001 235,415 (83.2) < 0.001  Renal disease 88,114 (27.8) 87,925 (24.2) < 0.001 9026 (15.8) < 0.001 49,075 (17.4) < 0.001  Liver disease 17,539 (5.5) 20,070 (5.5) 0.850 2708 (4.7) < 0.001 14,555 (5.1) < 0.001  Cancer 46,240 (14.6) 48,281 (13.3) < 0.001 7391 (12.9) < 0.001 37,374 (13.2) < 0.001  Myocardial infarction 48,548 (15.3) 46,891 (12.9) < 0.001 5762 (10.1) < 0.001 30,437 (10.8) < 0.001  Cardioversion and 7739 (2.4) 14,005 (3.8) < 0.001 1905 (3.3) < 0.001 9784 (3.5) < 0.001 catheter ablations  Dyspepsia or stomach 64,295 (20.3) 67,744 (18.6) < 0.001 9804 (17.2) < 0.001 51,007 (18.0) < 0.001 discomfort  Non-stroke/SE periph- 86,342 (27.2) 90,367 (24.8) < 0.001 11,594 (20.3) < 0.001 61,243 (21.7) < 0.001 eral vascular disease  Stroke/SE history 48,714 (15.4) 44,225 (12.2) < 0.001 6209 (10.9) < 0.001 27,976 (9.9) < 0.001  Transient ischemic 29,861 (9.4) 40,381 (11.1) < 0.001 4814 (8.4) < 0.001 22,912 (8.1) < 0.001 attack  Anemia and coagula- 108,463 (34.2) 101,299 (27.8) < 0.001 13,261 (23.2) < 0.001 69,496 (24.6) < 0.001 tion defects  Alcoholism 4333 (1.4) 7139 (2.0) < 0.001 872 (1.5) 0.003 5208 (1.8) < 0.001  Peripheral artery 84,900 (26.8) 83,434 (22.9) < 0.001 11,176 (19.6) < 0.001 58,338 (20.6) < 0.001 disease  Coronary artery disease 145,663 (45.9) 155,696 (42.8) < 0.001 22,774 (39.9) < 0.001 111,304 (39.4) < 0.001 Baseline medication use  ACEI/ARB 186,500 (58.8) 218,361 (60.0) < 0.001 33,813 (59.2) 0.057 165,980 (58.7) 0.504  Amiodarone 34,585 (10.9) 40,772 (11.2) < 0.001 5713 (10.0) < 0.001 27,606 (9.8) < 0.001  β-blocker 189,302 (59.7) 224,582 (61.7) < 0.001 33,939 (59.4) 0.287 169,755 (60.0) 0.004   H -receptor antagonist 24,156 (7.6) 26,202 (7.2) < 0.001 3,548 (6.2) < 0.001 18,177 (6.4) < 0.001  Proton pump inhibitor 99,092 (31.2) 112,992 (31.1) 0.132 16,342 (28.6) < 0.001 83,248 (29.4) < 0.001  Statin 190,060 (59.9) 221,106 (60.8) < 0.001 32,954 (57.7) < 0.001 162,231 (57.4) < 0.001  Antiplatelets 57,262 (18.0) 64,285 (17.7) < 0.001 8992 (15.7) < 0.001 45,541 (16.1) < 0.001 A. D. Dhamane et al. Table 1 (continued) Characteristics Warfarin cohort Apixaban cohort P value Dabigatran cohort P value Rivaroxaban cohort P value (N = 317,337), reference (N = 363,823) (N = 57,121) (N = 282,831)  NSAIDs 63,271 (19.9) 86,806 (23.9) < 0.001 13,453 (23.6) < 0.001 70,583 (25.0) < 0.001 Dose of the index prescription  Standard dose 281,476 (77.4) 48,219 (84.4) 209,777 (74.2)  Low dose 82,347 (22.6) 8902 (15.6) 73,054 (25.8) Data are presented as mean ± standard deviation or N (%) unless otherwise indicated ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin II receptor blocker, CHA DS -VASc congestive heart failure, hypertension, aged 2 2 ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65–74 years, sex category, HAS-BLED hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratios, elderly, drugs or alcohol, INR international normalized ratio, NSAIDs non-steroidal anti-inflammatory drugs, SE systemic embolism Sex was unknown for one patient in the apixaban cohort and another patient in the rivaroxaban cohort As the INR value was not available in the databases, a modified HAS-BLED score was calculated with a range of 0–8 Standard dose: apixaban 5 mg, dabigatran 150 mg, rivaroxaban 20 mg Lower dose: apixaban 2.5 mg, dabigatran 75 mg, dabigatran 110 mg, rivaroxaban 10 mg, rivaroxaban 15 mg; 200 patients treated with dabi- gatran were prescribed dabigatran 110 mg, and 15,362 patients treated with rivaroxaban were prescribed rivaroxaban 10 mg Fig. 2 Cumulative incidence of non-persistence to oral anticoagulants. The cumulative incidence of non-persistence to oral anticoagulants during the entire follow-up period was calculated using a 60-day gap. At 3 months, the cumulative incidences of non-persistence were 21.2, 33.9, 28.8, and 26.7% for apixaban, dabigatran, rivaroxaban, and warfarin patients, respectively. Apix apixaban, Dabi dabigatran, Riva rivaroxaban, Warf warfarin The effect of a change to the definition of the discontinua- 4 Discussion tion gap from 60 to 30 days is shown in Tables 2, 4 and Fig. 1 in the ESM. As expected, there was an increase in non-persis- This is the largest retrospective observational study to date tence for all OACs, but the results for the comparative risk of to examine the risk of non-persistence among patients with non-persistence across DOACs did not change significantly. NVAF initiating OAC treatment. Non-persistence was high Finally, the addition of INR records to extend warfarin treat- across all OAC treatment cohorts; however, patients initi- ment increased the likelihood of non-persistence with rivaroxa- ating apixaban or rivaroxaban were less likely than those ban, as compared with warfarin (aHR 1.02; 95% CI 1.01–1.03; receiving warfarin to be non-persistent. Patients initiating P < 0.001), as shown in Tables 2, 4 and Fig. 2 in the ESM. Oral Anticoagulant Non-persistence in Patients with NVAF Table 2 Descriptive outcomes for main analysis Outcomes Warfarin cohort Apixaban cohort Dabigatran cohort Rivaroxaban cohort (N = 317,337) (N = 363,823) (N = 57,121) (N = 282,831) Follow-up time (in days) 892.0 ± 543.7 605.0 ± 410.5 939.3 ± 566.4 799.7 ± 524.4  Minimum 62 62 62 62  Q1 412 269 445 335  Median 845 521 872 715  Q3 1336 867 1420 1208  Maximum 2371 2319 2371 2371 Number of prescriptions  Pts with one index OAC prescription 40,192 (12.7) 41,298 (11.4) 11,416 (20.0) 47,499 (16.8)  Pts with more than one index OAC prescription 277,145 (87.3) 322,525 88.6) 45,705 (80.0) 235,332 (83.2) Non-persistent patients 207,565 (65.4) 172,574 (47.4) 41,108 (72.0) 171,799 (60.7) Type of change in therapy  Discontinued 183,401 (57.8) 162,455 (44.7) 35,869 (62.8) 157,717 (55.8)  Time to discontinuation (days) 278 ± 317.1 214 ± 240.2 250 ± 317.7 223 ± 275.5  Switched 24,164 (7.6) 10,119 (2.8) 5239 (9.2) 14,082 (5.0)  Time to switch (days) 211 ± 301.3 154 ± 213.6 208 ± 299.6 198 ± 279.4 OAC switched to  Apixaban 10,627 (44.0) 0 (0.0) 2016 (38.5) 6642 (47.2)  Dabigatran 2392 (9.9) 959 (9.5) 0 (0.0) 1170 (8.3)  Edoxaban 55 (0.2) 60 (0.6) 8 (0.2) 58 (0.4)  Rivaroxaban 11,090 (45.9) 4552 (45.0) 1727 (33.0) 0 (0.0)  Warfarin 0 (0.0) 4548 (44.9) 1488 (28.4) 6212 (44.1) Data are presented as mean ± standard deviation or N (%) unless otherwise indicated OAC oral anticoagulant, pts patients, Q quarter Fig. 3 Risk of non-persistence among oral anticoagulants. Cox proportional hazard models were used to evaluate the risk of non-persistence. Apixaban and rivaroxaban were associ- ated with a lower risk of non- persistence than was warfarin. Dabigatran was associated with a higher risk of non- persistence than were warfarin and rivaroxaban. Apixaban was associated with a lower risk of non-persistence than were dabigatran and rivaroxaban. Large sample sizes and preci- sion mean the 95% confidence intervals (CIs) are narrow and difficult to observe in the figure A. D. Dhamane et al. apixaban were less likely to be non-persistent than those Table 3 Adjusted hazard ratios of non-persistence receiving rivaroxaban, whereas patients initiating rivaroxa- Variable HR (95% CI) ban were less likely to be non-persistent than those receiving Age, years dabigatran.  18–54 (reference) Our results were generally consistent with those of other  55–64 0.72 (0.71–0.73)* studies, including a recently published network meta-analy-  65–74 0.66 (0.65–0.66)* sis that pooled 36 real-world retrospective studies published  ≥ 75 0.64 (0.63–0.65)* between 2013 and 2018 and found persistence to be higher Sex among DOAC users than VKA users (odds ratio [OR] 1.44;  Male (reference) 95% CI 1.12–1.86; P = 0.005) [12, 24, 25]. Our results for  Female 0.96 (0.95–0.96)* the comparisons between rivaroxaban and apixaban and war- US geographic region farin were consistent with these observations. Our results Northeast (reference) were also in line with previous studies, which reported a  Midwest 1.02 (1.01–1.03)* higher risk of non-persistence among dabigatran users than  South 1.16 (1.15–1.17)* rivaroxaban and warfarin users [12]. The twice-daily dosing  West 1.13 (1.12–1.14)* of dabigatran was thought to be a possible explanation for  Other 1.42 (1.34–1.51)* these observations. However, we found that patients receiv- Comorbidities  Deyo–Charlson Comorbidity Index 1.02 (1.02–1.02)* ing apixaban, which also has a twice-daily dosing regimen,  Bleeding history 1.08 (1.07–1.09)* had a lower risk of non-persistence than those receiving  Congestive heart failure 0.99 (0.99–1.00) rivaroxaban and warfarin. This suggests that factors other  Diabetes mellitus 1.00 (0.99–1.01) than the dosing regimen play a major role in the lower per-  Hypertension 0.89 (0.88–0.90)* sistence associated with dabigatran [12]. In addition, a recent  Renal disease 1.02 (1.01–1.03)* meta-analysis found that apixaban had the best safety profile  Liver disease 0.99 (0.97–1.00) of the DOACs, another possible explanation for the lower  Cancer 0.97 (0.96–0.98)* risk of non-persistence among patients receiving apixaban  Myocardial infarction 1.04 (1.03–1.05)* [26].  Cardioversion and catheter ablations 1.02 (1.01–1.04) Our study adds to the literature on persistence with  Dyspepsia or stomach discomfort 1.07 (1.06–1.08)* OAC because we evaluated whether differences in the  Non-stroke/SE peripheral vascular disease 0.93 (0.91–0.95)* comparative risk of non-persistence across treatment  Stroke/SE 0.90 (0.89–0.91)* groups remained significant after adjusting for clinical  Transient ischemic attack 0.96 (0.95–0.97)* events during follow-up. This is relevant because a Dan-  Anemia and coagulation defects 1.11 (1.10–1.11)*  Alcoholism 1.11 (1.09–1.13)* ish study evaluating OAC switch and discontinuation  Peripheral artery disease 1.12 (1.09–1.14)* found that half of OAC treatment changes were preceded  Coronary artery disease 1.07 (1.06–1.08)* by a hospitalization, most frequently for stroke/SE, MB, Medication use new acute renal failure, new chronic renal failure, new  ACEI/ARB 0.94 (0.93–0.94)* cancer diagnoses, or cardioversions or catheter ablations  Amiodarone 1.18 (1.17–1.19)* [18, 27]. To build on this evidence, we tested whether  β-blockers 0.95 (0.95–0.96)* the differences in the risk of non-persistence across treat-   H -receptor antagonist 1.01 (1.00–1.02) ment groups remained significant after accounting for  Proton pump inhibitor 1.03 (1.02–1.03)* the differential risk of these events associated with treat-  Statins 0.88 (0.88–0.89)* ment changes. In doing so, we demonstrated that apixaban  Antiplatelets 0.99 (0.98–1.00) remained associated with a lower risk of non-persistence  NSAIDs 1.07 (1.06–1.07)* after accounting for ischemic and bleeding events and ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin II other relevant clinical factors. receptor blocker, CI confidence interval, HR hazard ratio, NSAIDs Our analyses, including both baseline patient charac- non-steroidal anti-inflammatory drugs, SE systemic embolism teristics and time-dependent variables for major clinical Models adjusted for age, sex, region, atrial fibrillation index year, events, demonstrated that the occurrence of MB events Deyo–Charlson Comorbidity Index, bleeding history, history of con- after treatment initiation was the strongest predictor of gestive heart failure, diabetes mellitus, hypertension, renal disease, liver disease, cancer, myocardial infarction, cardioversion and cathe- non-persistence. However, including the occurrence of ter ablations, dyspepsia or stomach discomfort, non-stroke/SE periph- major clinical events during follow-up did not significantly eral vascular disease, stroke/SE, transient ischemic attack, anemia and change the comparative risk of non-persistence across coagulation defects, alcoholism, peripheral artery disease, coronary groups in our model that included time-dependent vari- artery disease, and baseline medication use ables. This may be because of other factors related to the *HR significant at P < 0.001 Oral Anticoagulant Non-persistence in Patients with NVAF Table 4 Adjusted hazard ratios of non-persistence with time-varying covariates a a a HR (95% CI) HR (95% CI) HR (95% CI) Cohort  Warfarin Reference  Apixaban 0.76 (0.75–0.76) 0.76 (0.76–0.77) 0.62 (0.61–0.62)  Dabigatran 1.23 (1.22–1.24) 1.25 (1.23–1.26) Reference  Rivaroxaban 0.99 (0.98–0.99) Reference Time-varying covariates  Stroke/SE (primary discharge) 1.57 (1.53–1.61) 1.58 (1.53–1.64) 1.52 (1.45–1.60)  Major bleeding (primary discharge) 2.96 (2.92–3.00) 3.31 (3.25–3.37) 3.12 (3.04–3.20)  New acute renal failure 1.44 (1.42–1.46) 1.45 (1.43–1.47) 1.44 (1.41–1.47)  New chronic renal failure 1.14 (1.12–1.15) 1.17 (1.14–1.19) 1.13 (1.10–1.16)  New cancer 1.22 (1.20–1.25) 1.23 (1.21–1.26) 1.23 (1.19–1.27)  Cardioversions and catheter ablations 1.17 (1.15–1.19) 1.15 (1.13–1.18) 1.13 (1.10–1.16) Data are presented as hazard ratio (95% confidence interval). All hazard ratios were significant at P  <  0.001. Models adjusted for age, sex, region, atrial fibrillation index year, Deyo–Charlson Comorbidity Index, bleeding history, history of congestive heart failure, diabetes mellitus, hypertension, renal disease, liver disease, cancer, myocardial infarction, cardioversion and catheter ablations, dyspepsia or stomach discomfort, non-stroke/SE peripheral vascular disease, stroke/SE, transient ischemic attack, anemia and coagulation defects, alcoholism, peripheral artery disease, coronary artery disease, baseline medication use, and time-varying covariates during the follow-up HR hazard ratio, CI confidence interval, SE systemic embolism and access issues, which we were unable to capture but have differences in non-persistence risk among OACs, including been documented as possible indicators for treatment switch drug tolerability and meal requirements for certain DOACs or discontinuation [29]. [28], which we were unable to ascertain because of data Nevertheless, our study is a major contribution to the lit- limitations. Among baseline characteristics, increasing erature on real-world persistence with OAC, since it is by age, statin use, and history of hypertension or stroke were far the largest retrospective observational study, with over associated with a lower risk of non-persistence. 1 million patients, examining the comparative risk of non- persistence between OACs. By pooling five datasets and 4.1 Limitations including a comprehensive comparison of the OACs, this study adds supplemental information to the literature and As with many real-world studies, our study has several limi- may assist in decisions around treatment selection for stroke tations. This study was designed to examine non-persistence prevention among patients with NVAF. The robustness of among patients with NVAF initiated on OACs and predictors our findings is evidenced by the similar results observed of non-persistence, so we could not evaluate causal rela- after limiting follow-up to the initial 12 months after OAC tionships. As is the nature with retrospective observational initiation, shortening of the gap length from 60 to 30 days, studies, our study was subject to confounders. This study and the inclusion of INR results in lengthening warfarin was bound by the limitations of the claims data; variables treatment line. such as over-the-counter use of aspirin, serum creatinine/ creatinine clearance, and laboratory values were unavailable and thus were not controlled for in the model. Codes from the International Classification of Diseases, Ninth and Tenth 5 Conclusion Revisions, Clinical Modification were used to identify base- line characteristics and outcomes, which may lack clinical In this real-world study of over 1 million patients with NVAF, accuracy. Additionally, we were unable to determine time in our results suggest that noteworthy differences in anticoagu- therapeutic range for patients prescribed warfarin or jointly lation treatment persistence exist across OAC agents, even assess the relationship between persistence and adherence. after accounting for clinical events after OAC initiation. It is Furthermore, we were unable to identify predictors related important for clinicians and patients to take these differences to cost or access. Our results might in fact have been driven into consideration, especially as non-persistence to OAC by non-medical reasons, including out-of-pocket costs, for- therapy is associated with thromboembolic complications, mulary changes, insurance changes, physician preferences, including stroke, MB, and all-cause mortality. A. D. Dhamane et al. Supplementary Information The online version contains supplemen- for thromboprophylaxis. J Am Coll Cardiol. 2010;56(11):827– tary material available at https://doi. or g/10. 1007/ s40256- 021- 00501-w . 37. https:// doi. org/10. 1016/j. jacc. 2010. 05. 028. 3. Sankaranarayanan R, Kirkwood G, Visweswariah R, et al. How does chronic atrial fibrillation influence mortality in the modern Declarations treatment era? Curr Cardiol Rev. 2015;11(3):190–8. https://d oi. org/10. 2174/ 15734 03x10 66614 09021 43020. Funding This study was sponsored by Bristol Myers Squibb and Pfizer. 4. Pastori D, Menichelli D, Del Sole F, Pignatelli P, Violi F. Long-term risk of major adverse cardiac events in atrial fibril- lation patients on direct oral anticoagulants. Mayo Clin Proc. Conflict of interest Amol Dhamane and Mauricio Ferri are paid em- 2021;96(3):658–65. https://doi. or g/10. 1016/j. ma yocp.2020. 06. ployees of Bristol Myers Squib. Inmaculada Hernandez is an employee of the University of Pittsburgh. Manuela Di Fusco, Cristina Russ, and 5. Vranckx P, Valgimigli M, Heidbuchel H. The significance of Birol Emir are paid employees and shareholders of Pfizer. Allison Ke- drug-drug and drug-food interactions of oral anticoagulation. shishian, Cynthia Gutierrez, and Wan-Lun Tsai are paid employees Arrhythm Electrophysiol Rev. 2018;7(1):55–61. https://doi. or g/ of STATinMED Research, which is a paid consultant to Pfizer and 10. 15420/ aer.2017. 50.1. Bristol Myers Squibb. Huseyin Yuce has no conflicts of interest that 6. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban ver- are directly relevant to the content of this article. sus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365(11):981–92. https:// doi. org/ 10. 1056/ NEJMo a1107 Ethics approval Since this study did not involve the collection, use, or transmittal of individually identifiable data, it was exempt from Insti- 7. Connolly SJ, Ezekowitz MD, Yusuf S, et  al. Dabigatran ver- tutional Review Board review. Both the datasets and the security of sus warfarin in patients with atrial fibrillation. N Engl J Med. the offices where analysis was completed (and where the datasets are 2009;361(12):1139–51. https://doi. or g/10. 1056/ NEJMo a0905 561 . kept) meet the requirements of the Health Insurance Portability and 8. Patel MR, Mahaffey KW, Garg J, et  al. Rivaroxaban ver- Accountability Act of 1996. sus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365(10):883–91. https://doi. or g/10. 1056/ NEJMo a1009 638 . Consent to participate Not applicable. 9. Giugliano RP, Ruff CT, Braunwald E, et  al. Edoxaban ver- sus warfarin in patients with atrial fibrillation. N Engl J Med. Consent for publication Not applicable. 2013;369(22):2093–104. https:// doi. org/ 10. 1056/ NEJMo a1310 Data availability The data, analytical methods, and study materials will 10. Rivera-Caravaca JM, Esteve-Pastor MA, Roldán V, et al. Non-vita- not be made available to other researchers for purposes of reproducing min K antagonist oral anticoagulants: impact of non-adherence the results or replicating the procedure. and discontinuation. Expert Opin Drug Saf. 2017;16(9):1051–62. https:// doi. org/ 10. 1080/ 14740 338. 2017. 13515 42. Code availability Code not publicly available. 11. Lowres N, Giskes K, Hespe C, Freedman B. Reducing stroke risk in atrial fibrillation: adherence to guidelines has improved, but Author Contributions Conceptualization: all authors; Methodology: all patient persistence with anticoagulant therapy remains subop- authors; Formal analysis and investigation: Allison Keshishian, Cynthia timal. Korean Circ J. 2019;49(10):883–907. https:// doi. org/ 10. Gutierrez, and Wan-Lun Tsai; Writing - original draft preparation: Alli- 4070/ kcj. 2019. 0234. son Keshishian and Cynthia Gutierrez; Writing - review and editing: 12. Ozaki AF, Choi AS, Le QT, et al. Real-world adherence and per- all authors; Funding acquisition: Amol D Dhamane Manuela Di Fusco, sistence to direct oral anticoagulants in patients with atrial fibrilla- Mauricio Ferri, Cristina Russ, Birol Emir. tion: a systematic review and meta-analysis. Circ Cardiovasc Qual Outcomes. 2020;13(3):e005969. https://doi. org/ 10. 1161/ CIRCO Open Access This article is licensed under a Creative Commons Attri- UTCOMES. 119. 005969. bution-NonCommercial 4.0 International License, which permits any 13. Wiley V, Franchina-Elder J, Fu AC, et al. Treatment and persis- non-commercial use, sharing, adaptation, distribution and reproduction tence with oral anticoagulants among newly diagnosed patients in any medium or format, as long as you give appropriate credit to the with non-valvular atrial fibrillation: a retrospective observational original author(s) and the source, provide a link to the Creative Com- study in a US commercially insured and Medicare Advantage mons licence, and indicate if changes were made. The images or other population. BMJ Open. 2018;8:e020676. https:// doi.or g/10. 1136/ third party material in this article are included in the article's Creative bmjop en-2017- 020676. Commons licence, unless indicated otherwise in a credit line to the 14. Mitchell SA, Simon TA, Raza S, et al. The efficacy and safety of material. If material is not included in the article's Creative Commons oral anticoagulants in warfarin-suitable patients with nonvalvular licence and your intended use is not permitted by statutory regula- atrial fibrillation: systematic review and meta-analysis. Clin Appl tion or exceeds the permitted use, you will need to obtain permission Thromb Hemost. 2013;19(6):619–31. https:// doi. org/ 10. 1177/ directly from the copyright holder. To view a copy of this licence, visit 10760 29613 486539. http:// creat iveco mmons.or g/ licen ses/ by- nc/4. 0/. 15. Lip GY, Keshishian A, Li X, et al. Ee ff ctiveness and safety of oral anticoagulants among nonvalvular atrial fibrillation patients: the ARISTOPHANES study. Stroke. 2018;49(12):2933–44. https:// doi. org/ 10. 1161/ STROK EAHA. 118. 020232. 16. Li X, Deitelzweig S, Keshishian A, et  al. Effectiveness and References safety of apixaban versus warfarin in non-valvular atrial fibrilla- tion patients in “real-world” clinical practice. Thromb Haemost. 1. Shea JB, Sears SF. Cardiology patient pages. A patient’s guide to 2017;117(6):1072–82. https:// doi. org/ 10. 1160/ TH17- 01- 0068. living with atrial fibrillation. Circulation. 2008;117(20):340– 3. 17. Jensen PN, Johnson K, Floyd J, Heckbert SR, Carnahan R, Dublin https:// doi. org/ 10. 1161/ CIRCU LATION AHA. 108. 780577. S. A systematic review of validated methods for identifying atrial 2. Marinigh R, Lip GY, Fiotti N, Giansante C, Lane DA. Age as a fibrillation using administrative data. Pharmacoepidemiol Drug risk factor for stroke in atrial fibrillation patients: implications Saf. 2012;21(suppl 1):141–7. https:// doi. org/ 10. 1002/ pds. 2317. Oral Anticoagulant Non-persistence in Patients with NVAF 18. Hellfritzsch M, Grove EL, Husted SE, Rasmussen L, Poulsen 25. Bancroft T, Lim J, Wang C, Sander SD, Swindle JP. Health care BK, Johnsen SP, Hallas J, Pottegård A. Clinical events preceding resource utilization, costs, and persistence in patients newly diag- switching and discontinuation of oral anticoagulant treatment in nosed as having nonvalvular atrial fibrillation and newly treated patients with atrial fibrillation. Ep Europace. 2017;19(7):1091 –5. with dabigatran versus warfarin in the United States. Clin Ther. https:// doi. org/ 10. 1093/ europ ace/ euw241. 2016;38(3):545–56. https:// doi. org/ 10. 1016/j. clint hera. 2016. 01. 19. Cunningham A, Stein CM, Chung CP, Daugherty JR, Smalley 008. WE, Ray WA. An automated database case definition for serious 26. Menichelli D, Del Sole F, Di Rocco A, et al. Real-world safety and bleeding related to oral anticoagulant use. Pharmacoepidemiol efficacy of direct oral anticoagulants in atrial fibrillation: a sys- Drug Saf. 2011;20(6):560–6. https:// doi. org/ 10. 1002/ pds. 2109. tematic review and meta-analysis of 605 771 patients. Eur Heart 20. Thigpen JL, Dillon C, Forster KB, et  al. Validity of interna- J Cardiovasc Pharmacother. 2021;7(FI1):f11–9. https:// doi. org/ tional classification of disease codes to identify ischemic stroke 10. 1093/ ehjcvp/ pvab0 02. and intracranial hemorrhage among individuals with associated 27. Manzoor BS, Walton SM, Sharp LK, Galanter WL, Lee TA, diagnosis of atrial fibrillation. Circ Cardiovasc Qual Outcomes. Nutescu EA. High number of newly initiated direct oral antico- 2015;8(1):8–14. agulant users switch to alternate anticoagulant therapy. J Thromb 21. Lip GYH, Pan X, Kamble S, Kawabata H, Mardekian J, Mas- Thrombolysis. 2017;44(4):435–41. https:// doi. org/ 10. 1007/ seria C, Phatak H. Discontinuation risk comparison among ‘real- s11239- 017- 1565-2. world’ newly anticoagulated atrial fibrillation patients: Apixaban, 28. Xarelto. Package insert. Janssen Pharmaceuticals, Inc. Food and warfarin, dabigatran, or rivaroxaban. PLoS ONE. 2018;13(4): Drug Administration website. https:// www. acces sdata. fda. gov/ e0195950. https:// doi. org/ 10. 1371/ journ al. pone. 01959 50. drugs atfda_ docs/ label/ 2017/ 20243 9s021 lbl. pdf. Published May 22. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, et al. Anti- 2016. Accessed 10 Dec 2020. coagulation therapy for stroke prevention in atrial fibrillation: how 29. Jackson LR 2nd, Kim S, Shrader P, et al. Early therapeutic per- well do randomized trials translate into clinical practice? JAMA. sistence on dabigatran versus warfarin therapy in patients with 2003;290(20):2685–92. https:// doi. org/ 10. 1001/ jama. 290. 20. atrial fibrillation: results from the Outcomes Registry for Better 2685. Informed Treatment of Atrial Fibrillation (ORBIT-AF) registry. 23. Rosendaal FR, Cannegieter SC, Van der Meer FJ, Briet EA. J Thromb Thrombolysis. 2018;46(4):435–9. https:// doi. org/ 10. Method to determine the optimal intensity of oral anticoagulant 1007/ s11239- 018- 1715-1. therapy. Thromb Haemost. 1993;69(3):236–9. 24. Nelson WW, Song X, Coleman CI, Thomson E, Smith DM, Damaraju CV, Schein JR. Medication persistence and discontinu- ation of rivaroxaban versus warfarin among patients with non-val- vular atrial fibrillation. Curr Med Res Opin. 2014;30(12):2461– 9. https:// doi. org/ 10. 1185/ 03007 995. 2014. 933577. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Cardiovascular Drugs Springer Journals

Non-persistence to Oral Anticoagulation Treatment in Patients with Non-valvular Atrial Fibrillation in the USA

Loading next page...
 
/lp/springer-journals/non-persistence-to-oral-anticoagulation-treatment-in-patients-with-non-Mbj5nR0jGL
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
1175-3277
eISSN
1179-187X
DOI
10.1007/s40256-021-00501-w
Publisher site
See Article on Publisher Site

Abstract

Background Studies have shown that patients with non-valvular atrial fibrillation (NVAF) who discontinue oral anticoagu- lants (OACs) are at higher risk of complications such as stroke. Objective This analysis compared the risk of non-persistence with OACs among patients with NVAF. Methods Adult patients with NVAF who initiated apixaban, dabigatran, rivaroxaban, or warfarin were identified using 01JAN2013–30JUN2019 data from Centers for Medicare and Medicaid Services and four US commercial claims databases. Non-persistence was defined as discontinuation (no evidence of index OAC use for ≥  60 days from the last days’ supply) or switch to another OAC. Kaplan–Meier curves were generated to illustrate time to non-persistence along with cumulative incidences of non-persistence. Baseline and time-varying covariates were evaluated, and adjusted Cox proportional hazards models were used to evaluate non-persistence risk. Results In total, 363,823 patients receiving apixaban, 57,121 receiving dabigatran, 282,831 receiving rivaroxaban, and 317,337 receiving warfarin were included. Of these, 47–72% discontinued/switched OAC therapy within an average 9-month follow-up. Apixaban was associated with a lower risk of non-persistence than were dabigatran (hazard ratio [HR] 0.62; 95% confidence interval [CI] 0.61– 0.62), rivaroxaban (HR 0.76; 95% CI 0.75–0.76), and warfarin (HR 0.74; 95% CI 0.74–0.75). Dabigatran was associated with a higher risk of non-persistence than were warfarin (HR 1.21; 95% CI 1.19–1.22) and rivar- oxaban (HR 1.23; 95% CI 1.22–1.25), and rivaroxaban was associated with a lower risk of non-persistence than was warfarin (HR 0.98; 95% CI 0.97–0.98). Clinical events (stroke/systemic embolism and major bleeding [MB]) during follow-up were predictors of non-persistence (stroke HR 1.57; 95% CI 1.53–1.61; MB HR 2.96; 95% CI 2.92–3.00). Conclusion In over one million patients with NVAF, our results suggest differences in anticoagulation treatment persistence across OAC agents, even after accounting for clinical events after OAC initiation. It is important for clinicians and patients to take these differences into consideration, especially as non-persistence to OAC therapy is associated with thromboembolic complications. Key Points This study examined non-persistence with oral antico- * Allison Keshishian agulants (OACs; apixaban, dabigatran, rivaroxaban, and akeshishian@statinmed.com warfarin) among patients with non-valvular atrial fibril- Bristol Myers Squibb, Lawrenceville, NJ, USA lation (NVAF). OAC non-persistence among patients Department of Pharmacy and Therapeutics, School with NVAF has previously been associated with an of Pharmacy, University of Pittsburgh, Pittsburgh, PA, USA increased risk of adverse events. Pfizer, New York, NY, USA The risk of non-persistence varied among patients with STATinMED Research, 4110 Varsity Dr, Ann Arbor, NVAF; baseline characteristics and clinical events occur- MI 48108, USA ring after OAC initiation were significant predictors of New York City College of Technology, City University non-persistence. of New York, New York, NY, USA Vol.:(0123456789) A. D. Dhamane et al. and Medicaid Services data (1 January 2012–31 Decem- 1 Introduction ber 2017), Truven Mar ketScan Commercial Claims and Encounter (1 January 2012–30 June 2019), IMS PharMetrics Non-valvular atrial fibrillation (NVAF) is the most common Plus™ (1 January 2012–31 March 2019), Optum Clinfor- heart dysrhythmia diagnosed in the USA and an independ- matics™ Data Mart (1 January 2012–31 March 2019), and ent risk factor for stroke [1]. The prevalence of NVAF has the Humana Research Database (1 January 2012–31 March been increasing in the USA and is expected to continue to 2019). Of note, patients with Medicare supplemental plans increase substantially because of the aging of the popula- in Truven MarketScan and IMS PharMetrics Plus data were tion [2]. Atrial fibrillation (AF), in turn, increases the risk not included in the study to avoid potential duplicates with of mortality and morbidities such as major adverse cardiac Medicare Part A and Part B. More details on the datasets and events, arterial thromboembolism, and ischemic stroke [3, 4]. pooling method have been published [15, 16]. While vitamin K antagonists (VKAs), such as warfarin, have been the anticoagulant treatment of choice for several 2.2 Patient Selection years, they have unstable pharmacokinetic profiles, with wide interpatient variability and extensive interactions, We identified adult patients with NVAF between 1 Janu- requiring routine blood monitoring for dose adjustment ary 2012 and 30 June 2019 using inpatient and outpatient [5]. Clinical trials have demonstrated that the reduction in claims, and selected patients with a pharmacy claim for risk of stroke and bleeding with direct oral anticoagulants apixaban, dabigatran, edoxaban, rivaroxaban, or warfarin (DOACs: apixaban, dabigatran, edoxaban, rivaroxaban) is during the identification period between 1 January 2013 and similar or superior to that with warfarin, and DOACs have 30 June 2019. Each patient’s first OAC was labeled as their been approved in the USA to reduce the risk of stroke among “index therapy,” and the date of the first OAC prescription patients with NVAF [6–9]. claim was labeled as the “index date.” Patients prescribed Studies have also shown that patients who discontinue edoxaban were not included in the final analysis because of oral anticoagulants (OACs) are at increased risk of adverse the small sample size (N = 1629). Patients were required to cardiovascular effects, thromboembolic events, minor and have continuous medical and pharmacy health plan enroll- major bleeds, and all-cause mortality [10, 11]. Gradual ment for ≥ 12 months before the index date [17]. declines in persistence after the initial prescriptions among We excluded patients with any OAC treatment within patients with NVAF have been noted, with some studies 12 months before the index date to ensure they were treat- reporting a wide range of persistence, from 55 to 69%, after ment naïve. We also excluded patients with evidence of val- 12 months [12]. Although DOACs have fewer drug interac- vular heart disease, venous thromboembolism, transient AF tions than warfarin and do not require international normal- (pericarditis, hyperthyroidism, or thyrotoxicity), or heart valve ized ratio (INR) monitoring, real-world evidence rates of replacement or transplant during the baseline period; preg- discontinuation 12 months after treatment initiation have nancy during the study period; or hip/knee replacement sur- been estimated at 55% [13]. Results from a meta-analysis of gery within 6 weeks before the index date. To give all patients the DOAC randomized controlled trials indicated apixaban the possibility of discontinuing/switching index treatment, we had lower discontinuation rates than OACs, whereas war- also excluded patients with more than one OAC on the index farin had lower rates than dabigatran and rivaroxaban [14]. date and those with fewer than 60 days of follow-up, based on While studies have evaluated OAC persistence, real-world continuous health plan enrollment post-index. studies comparing persistence across DOACs and warfarin in large populations are lacking. Therefore, the objectives of this study were to compare the incidence and risk of non- 2.3 Outcome Measures persistence across DOACs and warfarin and identify pre- dictors of non-persistence among patients with NVAF who Patient data were assessed from the day after the index date initiated OAC therapy. until the earliest of the following: end of continuous medical and pharmacy enrollment, death (when available), or end of the study period. The primary outcome of interest was 2 Methods non-persistence, defined as either discontinuation or switch, with discontinuation defined as no evidence of the index 2.1 Data Sources OAC prescription for ≥ 60 days from the last day of supply of the last filled prescription and switch defined as filling a In this study, we pooled five large national claims databases prescription for an OAC other than the index drug during with the latest available data at the time of application. They the follow-up period within ± 60 days of last days’ supply included 100% fee-for-service US Centers for Medicare from the index OAC. Oral Anticoagulant Non-persistence in Patients with NVAF Baseline variables included demographics (age, sex, US 3 Results geographic region) and clinical characteristics such as clini- cal risk scores, prior stroke/systemic embolism (SE), prior 3.1 Baseline Characteristics bleeding, comorbidities, and baseline medication use. In total, 1,021,112 adults with NVAF were newly prescribed an OAC during the study period (Fig. 1). Of those patients, 2.4 Statistical Methodology 363,823 (35.6%) initiated apixaban with a mean ± standard deviation follow-up of 605 ± 410.5 days; 57,121 (5.6%) ini- Descriptive analysis of clinical and demographic variables was tiated dabigatran with a mean follow-up of 939.3 ± 566.4 conducted for patients prescribed DOACs or warfarin. Rates of days; 282,831 (27.7%) patients initiated rivaroxaban with non-persistence among patients with NVAF who initiated an a mean follow-up of 799.7  ±  524.4 days; and 317,337 OAC during the identification period were calculated during (31.1%) patients initiated warfarin with a mean follow-up the follow-up period overall and at 12 months post-index date. of 892  ±  543.7 days. Approximately 53% of patients in Unadjusted Kaplan–Meier survival curves were generated to all treatment groups combined were male. The mean age illustrate time to non-persistence in addition to cumulative of patients receiving apixaban, dabigatran, rivaroxaban, incidence of non-persistence. Cox proportional hazard mod- and warfarin was 75.1 ± 10.6, 73.4 ± 10.3, 73.5 ± 10.7, els with robust sandwich estimates were used to compare the and 76.7 ± 9.3 years, respectively. Patients initiating war- risk of non-persistence across treatment groups while adjust- farin were older and were at higher risk in terms of the ing for patient covariates. Baseline variables included in the CHA DS -VASc score (4.5 ± 1.8) and HAS-BLED score 2 2 model were patient demographics, index OAC prescription, (3.2  ±  1.4) (see Table  1 for CHA DS -VASc and HAS- 2 2 and clinical characteristics such as prior stroke/SE, prior bleed- BLED definitions) and had higher mean Deyo– Charlson ing, comorbidities, and baseline medication use. Comorbidity Index scores of 3.4 ± 3.0. For those receiving A secondary analysis was conducted taking into considera- apixaban, dabigatran, or rivaroxaban, 22.6% (2.5 mg), 15.6% tion clinical events occurring after the index date, including (15.2% on 75 mg; 0.4% on 110 mg), and 25.8% (5.4% on 10 stroke/SE hospitalizations, major bleeding (MB) hospitaliza- mg; 20.4% on 15 mg) had lower dosage regimens, respec- tions, new acute renal failure (considered new when it occurred tively (Table 1). for the first time within a timeframe of 6 months), new chronic renal failure (considered new when it occurred for the first 3.2 Unadjusted Cumulative Incidence time after the index date), new cancer diagnoses (considered of Non‑persistence new when it occurred for the first time within a timeframe of 6 months), and cardioversions and catheter ablations [18]. The cumulative incidence of non-persistence at 3 months The diagnosis codes used for stroke/SE and MB hospitaliza- was 21.2, 33.9, 28.8, and 26.7% for patients receiving tions were based on validated administrative claims-based apixaban, dabigatran, rivaroxaban, or warfarin, respectively algorithms and the International Society on Thrombosis and (Fig. 2). The cumulative incidence of non-persistence at 12 Haemostasis definition of MB [19, 20]. months was 42.7, 58.9, 52.2, and 51.3% for patients receiv- We conducted three sensitivity analyses: ing apixaban, dabigatran, rivaroxaban, or warfarin, respec- tively (Fig. 2). Most of the non-persistence was because of • Follow-up was limited to the first 12 months after the discontinuation of therapy, but 3–9% of patients switched index date. The shorter follow-up period allowed us to treatment, with dabigatran having the highest switch rate focus on short-term persistence. (Table 2). Non-persistence was reanalyzed using a ≥ 30-day dis- continuation gap instead of ≥ 60 days, which allowed us to observe how a shorter gap period would alter non- 3.3 Primary Analyses Controlling for Baseline persistence outcomes [21]. Characteristics Warfarin non-persistence was reanalyzed with the inclu- sion of INR records to extend warfarin treatment lines. Figure  3 and Table  3 show the results of primary analy- Patients receiving warfarin were considered discontinued ses with Cox models that adjusted for demographic and if the gap between two consecutive warfarin prescriptions clinical covariates defined at baseline. When compared or from the last prescription to end of study was longer with those receiving warfarin, patients initiating apixaban than 60 days and they did not have INR measurements (adjusted hazard ratio [aHR] 0.74; 95% confidence interval at least every 42 days [22, 23]. This sensitivity analysis [CI] 0.74–0.75; P < 0.001) or rivaroxaban (aHR 0.98; 95% allowed us to take into consideration dose adjustments CI 0.97–0.98; P < 0.001) were 26 and 2% less likely to be or possible additional prescriptions. A. D. Dhamane et al. Fig. 1 Patient selection criteria. The patient selection criteria yielded > 1,000,000 patients with non-valvular atrial fibrillation prescribed either apixaban, dabigatran, rivaroxa- ban, or warfarin. Patients were stratified into cohorts based on their index OAC. *Edoxaban was not included in the study because it received US FDA approval in 2015 so the sample size was small (N = 1629). AF atrial fibrillation, ICD-9/10-CM International Classification of Diseases, Ninth/Tenth Revision, Clinical Modification, OAC oral anticoagulant, VTE venous thromboembolism non-persistent, respectively. Patients receiving dabigatran results when time-varying covariates were included in Cox (aHR 1.21; 95% CI 1.19–1.22; P < 0.001) were 21% more models. The HRs for OACs remained consistent after the likely to be non-persistent than those receiving warfarin. inclusion of time-varying covariates. All time-varying When compared with patients receiving rivaroxaban, clinical characteristics were significant predictors of non- those initiating apixaban (aHR 0.76; 95% CI 0.75–0.76; persistence: stroke/SE aHR 1.57; 95% CI 1.53–1.61; MB P  <  0.001) were 24% less likely to be non-persistent. aHR 2.96; 95% CI 2.92–3.00; new acute renal failure Those receiving dabigatran (aHR 1.23; 95% CI 1.22–1.25; aHR 1.44; 95% CI 1.42–1.46; new chronic renal failure P < 0.001) were 23% more likely to be non-persistent than aHR 1.14; 95% CI 1.12–1.15, new cancer aHR 1.22; 95% those receiving rivaroxaban. When compared with patients CI 1.20–1.25; and cardioversions and catheter ablations receiving dabigatran, those initiating apixaban (aHR 0.62; aHR 1.17; 95% CI 1.15–1.19. All HRs were significant 95% CI 0.61–0.62; P < 0.001) were 38% less likely to be at P < 0.001. non-persistent. Older age and history of hypertension and stroke/SE were 3.5 Sensitivity Analyses predictors of a lower likelihood of non-persistence (Table 3). Patient use of statins was also a predictor of reduced risk of In the first sensitivity analysis, follow-up was limited to non-persistence. the first 12 months after index treatment initiation. The results are shown in Tables 2 and 4 in the ESM. The results 3.4 Secondary Analyses Controlling for Baseline were generally consistent with those of the main analysis, Characteristics and Time‑Dependent Variables except when comparing rivaroxaban and warfarin: when follow-up was limited to the first 12 months, rivaroxa- Table 1 in the electronic supplementary material (ESM) ban was associated with a 3% increase in non-persistence shows the proportion of patients with time-varying clinical compared with warfarin (aHR 1.03; 95% CI 1.02–1.04; events after the index date, and Table 4 shows the model P < 0.001). Oral Anticoagulant Non-persistence in Patients with NVAF Table 1 Baseline characteristics and treatment follow-up Characteristics Warfarin cohort Apixaban cohort P value Dabigatran cohort P value Rivaroxaban cohort P value (N = 317,337), reference (N = 363,823) (N = 57,121) (N = 282,831) Age, years 76.7 ± 9.3 75.1 ± 10.6 < 0.001 73.4 ± 10.3 < 0.001 73.5 ± 10.7 < 0.001  18–54 5630 (1.8) 13,780 (3.8) < 0.001 2735 (4.8) < 0.001 14,352 (5.1) < 0.001  55–64 17,296 (5.5) 35,915 (9.9) < 0.001 6169 (10.8) < 0.001 32,520 (11.5) < 0.001  65–74 104,121 (32.8) 117,039 (32.2) < 0.001 21,238 (37.2) < 0.001 100,054 (35.4) < 0.001  ≥ 75 190,290 (60.0) 197,089 (54.2) < 0.001 26,979 (47.2) < 0.001 135,905 (48.1) < 0.001 Sex  Male 165,668 (52.2) 188,252 (51.7) < 0.001 32,309 (56.6) < 0.001 156,327 (55.3) < 0.001  Female 151,669 (47.8) 175,570 (48.3) < 0.001 24,812 (43.4) < 0.001 126,503 (44.7) < 0.001 US geographic region  Northeast 60,061 (18.9) 60,877 (16.7) < 0.001 11,430 (20.0) < 0.001 50,509 (17.9) < 0.001  North central 98,273 (31.0) 81,283 (22.3) < 0.001 13,225 (23.2) < 0.001 68,069 (24.1) < 0.001  South 98,706 (31.1) 161,072 (44.3) < 0.001 22,184 (38.8) < 0.001 113,057 (40.0) < 0.001  West 59,688 (18.8) 60,182 (16.5) < 0.001 10,148 (17.8) < 0.001 50,634 (17.9) < 0.001  Other 609 (0.2) 409 (0.1) < 0.001 134 (0.2) 0.035 562 (0.2) 0.552 Baseline comorbidity  Deyo–Charlson Comor- 3.4 ± 3.0 2.9 ± 2.8 < 0.001 2.5 ± 2.5 < 0.001 2.6 ± 2.6 < 0.001 bidity Index   CHA DS -VASc score 4.5 ± 1.8 4.1 ± 1.9 < 0.001 3.8 ± 1.9 < 0.001 3.8 ± 1.9 < 0.001 2 2  HAS-BLED score 3.2 ± 1.4 3.0 ± 1.4 < 0.001 2.8 ± 1.3 < 0.001 2.9 ± 1.3 < 0.001  Bleeding history 71,975 (22.7) 63,850 (17.5) < 0.001 9339 (16.3) < 0.001 48,054 (17.0) < 0.001  Congestive heart failure 108,185 (34.1) 99,373 (27.3) < 0.001 13,521 (23.7) < 0.001 67,500 (23.9) < 0.001  Diabetes mellitus 125,920 (39.7) 125,450 (34.5) < 0.001 19,679 (34.5) < 0.001 94,611 (33.5) < 0.001  Hypertension 270,847 (85.3) 309,483 (85.1) 0.001 47,463 (83.1) < 0.001 235,415 (83.2) < 0.001  Renal disease 88,114 (27.8) 87,925 (24.2) < 0.001 9026 (15.8) < 0.001 49,075 (17.4) < 0.001  Liver disease 17,539 (5.5) 20,070 (5.5) 0.850 2708 (4.7) < 0.001 14,555 (5.1) < 0.001  Cancer 46,240 (14.6) 48,281 (13.3) < 0.001 7391 (12.9) < 0.001 37,374 (13.2) < 0.001  Myocardial infarction 48,548 (15.3) 46,891 (12.9) < 0.001 5762 (10.1) < 0.001 30,437 (10.8) < 0.001  Cardioversion and 7739 (2.4) 14,005 (3.8) < 0.001 1905 (3.3) < 0.001 9784 (3.5) < 0.001 catheter ablations  Dyspepsia or stomach 64,295 (20.3) 67,744 (18.6) < 0.001 9804 (17.2) < 0.001 51,007 (18.0) < 0.001 discomfort  Non-stroke/SE periph- 86,342 (27.2) 90,367 (24.8) < 0.001 11,594 (20.3) < 0.001 61,243 (21.7) < 0.001 eral vascular disease  Stroke/SE history 48,714 (15.4) 44,225 (12.2) < 0.001 6209 (10.9) < 0.001 27,976 (9.9) < 0.001  Transient ischemic 29,861 (9.4) 40,381 (11.1) < 0.001 4814 (8.4) < 0.001 22,912 (8.1) < 0.001 attack  Anemia and coagula- 108,463 (34.2) 101,299 (27.8) < 0.001 13,261 (23.2) < 0.001 69,496 (24.6) < 0.001 tion defects  Alcoholism 4333 (1.4) 7139 (2.0) < 0.001 872 (1.5) 0.003 5208 (1.8) < 0.001  Peripheral artery 84,900 (26.8) 83,434 (22.9) < 0.001 11,176 (19.6) < 0.001 58,338 (20.6) < 0.001 disease  Coronary artery disease 145,663 (45.9) 155,696 (42.8) < 0.001 22,774 (39.9) < 0.001 111,304 (39.4) < 0.001 Baseline medication use  ACEI/ARB 186,500 (58.8) 218,361 (60.0) < 0.001 33,813 (59.2) 0.057 165,980 (58.7) 0.504  Amiodarone 34,585 (10.9) 40,772 (11.2) < 0.001 5713 (10.0) < 0.001 27,606 (9.8) < 0.001  β-blocker 189,302 (59.7) 224,582 (61.7) < 0.001 33,939 (59.4) 0.287 169,755 (60.0) 0.004   H -receptor antagonist 24,156 (7.6) 26,202 (7.2) < 0.001 3,548 (6.2) < 0.001 18,177 (6.4) < 0.001  Proton pump inhibitor 99,092 (31.2) 112,992 (31.1) 0.132 16,342 (28.6) < 0.001 83,248 (29.4) < 0.001  Statin 190,060 (59.9) 221,106 (60.8) < 0.001 32,954 (57.7) < 0.001 162,231 (57.4) < 0.001  Antiplatelets 57,262 (18.0) 64,285 (17.7) < 0.001 8992 (15.7) < 0.001 45,541 (16.1) < 0.001 A. D. Dhamane et al. Table 1 (continued) Characteristics Warfarin cohort Apixaban cohort P value Dabigatran cohort P value Rivaroxaban cohort P value (N = 317,337), reference (N = 363,823) (N = 57,121) (N = 282,831)  NSAIDs 63,271 (19.9) 86,806 (23.9) < 0.001 13,453 (23.6) < 0.001 70,583 (25.0) < 0.001 Dose of the index prescription  Standard dose 281,476 (77.4) 48,219 (84.4) 209,777 (74.2)  Low dose 82,347 (22.6) 8902 (15.6) 73,054 (25.8) Data are presented as mean ± standard deviation or N (%) unless otherwise indicated ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin II receptor blocker, CHA DS -VASc congestive heart failure, hypertension, aged 2 2 ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65–74 years, sex category, HAS-BLED hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratios, elderly, drugs or alcohol, INR international normalized ratio, NSAIDs non-steroidal anti-inflammatory drugs, SE systemic embolism Sex was unknown for one patient in the apixaban cohort and another patient in the rivaroxaban cohort As the INR value was not available in the databases, a modified HAS-BLED score was calculated with a range of 0–8 Standard dose: apixaban 5 mg, dabigatran 150 mg, rivaroxaban 20 mg Lower dose: apixaban 2.5 mg, dabigatran 75 mg, dabigatran 110 mg, rivaroxaban 10 mg, rivaroxaban 15 mg; 200 patients treated with dabi- gatran were prescribed dabigatran 110 mg, and 15,362 patients treated with rivaroxaban were prescribed rivaroxaban 10 mg Fig. 2 Cumulative incidence of non-persistence to oral anticoagulants. The cumulative incidence of non-persistence to oral anticoagulants during the entire follow-up period was calculated using a 60-day gap. At 3 months, the cumulative incidences of non-persistence were 21.2, 33.9, 28.8, and 26.7% for apixaban, dabigatran, rivaroxaban, and warfarin patients, respectively. Apix apixaban, Dabi dabigatran, Riva rivaroxaban, Warf warfarin The effect of a change to the definition of the discontinua- 4 Discussion tion gap from 60 to 30 days is shown in Tables 2, 4 and Fig. 1 in the ESM. As expected, there was an increase in non-persis- This is the largest retrospective observational study to date tence for all OACs, but the results for the comparative risk of to examine the risk of non-persistence among patients with non-persistence across DOACs did not change significantly. NVAF initiating OAC treatment. Non-persistence was high Finally, the addition of INR records to extend warfarin treat- across all OAC treatment cohorts; however, patients initi- ment increased the likelihood of non-persistence with rivaroxa- ating apixaban or rivaroxaban were less likely than those ban, as compared with warfarin (aHR 1.02; 95% CI 1.01–1.03; receiving warfarin to be non-persistent. Patients initiating P < 0.001), as shown in Tables 2, 4 and Fig. 2 in the ESM. Oral Anticoagulant Non-persistence in Patients with NVAF Table 2 Descriptive outcomes for main analysis Outcomes Warfarin cohort Apixaban cohort Dabigatran cohort Rivaroxaban cohort (N = 317,337) (N = 363,823) (N = 57,121) (N = 282,831) Follow-up time (in days) 892.0 ± 543.7 605.0 ± 410.5 939.3 ± 566.4 799.7 ± 524.4  Minimum 62 62 62 62  Q1 412 269 445 335  Median 845 521 872 715  Q3 1336 867 1420 1208  Maximum 2371 2319 2371 2371 Number of prescriptions  Pts with one index OAC prescription 40,192 (12.7) 41,298 (11.4) 11,416 (20.0) 47,499 (16.8)  Pts with more than one index OAC prescription 277,145 (87.3) 322,525 88.6) 45,705 (80.0) 235,332 (83.2) Non-persistent patients 207,565 (65.4) 172,574 (47.4) 41,108 (72.0) 171,799 (60.7) Type of change in therapy  Discontinued 183,401 (57.8) 162,455 (44.7) 35,869 (62.8) 157,717 (55.8)  Time to discontinuation (days) 278 ± 317.1 214 ± 240.2 250 ± 317.7 223 ± 275.5  Switched 24,164 (7.6) 10,119 (2.8) 5239 (9.2) 14,082 (5.0)  Time to switch (days) 211 ± 301.3 154 ± 213.6 208 ± 299.6 198 ± 279.4 OAC switched to  Apixaban 10,627 (44.0) 0 (0.0) 2016 (38.5) 6642 (47.2)  Dabigatran 2392 (9.9) 959 (9.5) 0 (0.0) 1170 (8.3)  Edoxaban 55 (0.2) 60 (0.6) 8 (0.2) 58 (0.4)  Rivaroxaban 11,090 (45.9) 4552 (45.0) 1727 (33.0) 0 (0.0)  Warfarin 0 (0.0) 4548 (44.9) 1488 (28.4) 6212 (44.1) Data are presented as mean ± standard deviation or N (%) unless otherwise indicated OAC oral anticoagulant, pts patients, Q quarter Fig. 3 Risk of non-persistence among oral anticoagulants. Cox proportional hazard models were used to evaluate the risk of non-persistence. Apixaban and rivaroxaban were associ- ated with a lower risk of non- persistence than was warfarin. Dabigatran was associated with a higher risk of non- persistence than were warfarin and rivaroxaban. Apixaban was associated with a lower risk of non-persistence than were dabigatran and rivaroxaban. Large sample sizes and preci- sion mean the 95% confidence intervals (CIs) are narrow and difficult to observe in the figure A. D. Dhamane et al. apixaban were less likely to be non-persistent than those Table 3 Adjusted hazard ratios of non-persistence receiving rivaroxaban, whereas patients initiating rivaroxa- Variable HR (95% CI) ban were less likely to be non-persistent than those receiving Age, years dabigatran.  18–54 (reference) Our results were generally consistent with those of other  55–64 0.72 (0.71–0.73)* studies, including a recently published network meta-analy-  65–74 0.66 (0.65–0.66)* sis that pooled 36 real-world retrospective studies published  ≥ 75 0.64 (0.63–0.65)* between 2013 and 2018 and found persistence to be higher Sex among DOAC users than VKA users (odds ratio [OR] 1.44;  Male (reference) 95% CI 1.12–1.86; P = 0.005) [12, 24, 25]. Our results for  Female 0.96 (0.95–0.96)* the comparisons between rivaroxaban and apixaban and war- US geographic region farin were consistent with these observations. Our results Northeast (reference) were also in line with previous studies, which reported a  Midwest 1.02 (1.01–1.03)* higher risk of non-persistence among dabigatran users than  South 1.16 (1.15–1.17)* rivaroxaban and warfarin users [12]. The twice-daily dosing  West 1.13 (1.12–1.14)* of dabigatran was thought to be a possible explanation for  Other 1.42 (1.34–1.51)* these observations. However, we found that patients receiv- Comorbidities  Deyo–Charlson Comorbidity Index 1.02 (1.02–1.02)* ing apixaban, which also has a twice-daily dosing regimen,  Bleeding history 1.08 (1.07–1.09)* had a lower risk of non-persistence than those receiving  Congestive heart failure 0.99 (0.99–1.00) rivaroxaban and warfarin. This suggests that factors other  Diabetes mellitus 1.00 (0.99–1.01) than the dosing regimen play a major role in the lower per-  Hypertension 0.89 (0.88–0.90)* sistence associated with dabigatran [12]. In addition, a recent  Renal disease 1.02 (1.01–1.03)* meta-analysis found that apixaban had the best safety profile  Liver disease 0.99 (0.97–1.00) of the DOACs, another possible explanation for the lower  Cancer 0.97 (0.96–0.98)* risk of non-persistence among patients receiving apixaban  Myocardial infarction 1.04 (1.03–1.05)* [26].  Cardioversion and catheter ablations 1.02 (1.01–1.04) Our study adds to the literature on persistence with  Dyspepsia or stomach discomfort 1.07 (1.06–1.08)* OAC because we evaluated whether differences in the  Non-stroke/SE peripheral vascular disease 0.93 (0.91–0.95)* comparative risk of non-persistence across treatment  Stroke/SE 0.90 (0.89–0.91)* groups remained significant after adjusting for clinical  Transient ischemic attack 0.96 (0.95–0.97)* events during follow-up. This is relevant because a Dan-  Anemia and coagulation defects 1.11 (1.10–1.11)*  Alcoholism 1.11 (1.09–1.13)* ish study evaluating OAC switch and discontinuation  Peripheral artery disease 1.12 (1.09–1.14)* found that half of OAC treatment changes were preceded  Coronary artery disease 1.07 (1.06–1.08)* by a hospitalization, most frequently for stroke/SE, MB, Medication use new acute renal failure, new chronic renal failure, new  ACEI/ARB 0.94 (0.93–0.94)* cancer diagnoses, or cardioversions or catheter ablations  Amiodarone 1.18 (1.17–1.19)* [18, 27]. To build on this evidence, we tested whether  β-blockers 0.95 (0.95–0.96)* the differences in the risk of non-persistence across treat-   H -receptor antagonist 1.01 (1.00–1.02) ment groups remained significant after accounting for  Proton pump inhibitor 1.03 (1.02–1.03)* the differential risk of these events associated with treat-  Statins 0.88 (0.88–0.89)* ment changes. In doing so, we demonstrated that apixaban  Antiplatelets 0.99 (0.98–1.00) remained associated with a lower risk of non-persistence  NSAIDs 1.07 (1.06–1.07)* after accounting for ischemic and bleeding events and ACEI angiotensin-converting enzyme inhibitor, ARB angiotensin II other relevant clinical factors. receptor blocker, CI confidence interval, HR hazard ratio, NSAIDs Our analyses, including both baseline patient charac- non-steroidal anti-inflammatory drugs, SE systemic embolism teristics and time-dependent variables for major clinical Models adjusted for age, sex, region, atrial fibrillation index year, events, demonstrated that the occurrence of MB events Deyo–Charlson Comorbidity Index, bleeding history, history of con- after treatment initiation was the strongest predictor of gestive heart failure, diabetes mellitus, hypertension, renal disease, liver disease, cancer, myocardial infarction, cardioversion and cathe- non-persistence. However, including the occurrence of ter ablations, dyspepsia or stomach discomfort, non-stroke/SE periph- major clinical events during follow-up did not significantly eral vascular disease, stroke/SE, transient ischemic attack, anemia and change the comparative risk of non-persistence across coagulation defects, alcoholism, peripheral artery disease, coronary groups in our model that included time-dependent vari- artery disease, and baseline medication use ables. This may be because of other factors related to the *HR significant at P < 0.001 Oral Anticoagulant Non-persistence in Patients with NVAF Table 4 Adjusted hazard ratios of non-persistence with time-varying covariates a a a HR (95% CI) HR (95% CI) HR (95% CI) Cohort  Warfarin Reference  Apixaban 0.76 (0.75–0.76) 0.76 (0.76–0.77) 0.62 (0.61–0.62)  Dabigatran 1.23 (1.22–1.24) 1.25 (1.23–1.26) Reference  Rivaroxaban 0.99 (0.98–0.99) Reference Time-varying covariates  Stroke/SE (primary discharge) 1.57 (1.53–1.61) 1.58 (1.53–1.64) 1.52 (1.45–1.60)  Major bleeding (primary discharge) 2.96 (2.92–3.00) 3.31 (3.25–3.37) 3.12 (3.04–3.20)  New acute renal failure 1.44 (1.42–1.46) 1.45 (1.43–1.47) 1.44 (1.41–1.47)  New chronic renal failure 1.14 (1.12–1.15) 1.17 (1.14–1.19) 1.13 (1.10–1.16)  New cancer 1.22 (1.20–1.25) 1.23 (1.21–1.26) 1.23 (1.19–1.27)  Cardioversions and catheter ablations 1.17 (1.15–1.19) 1.15 (1.13–1.18) 1.13 (1.10–1.16) Data are presented as hazard ratio (95% confidence interval). All hazard ratios were significant at P  <  0.001. Models adjusted for age, sex, region, atrial fibrillation index year, Deyo–Charlson Comorbidity Index, bleeding history, history of congestive heart failure, diabetes mellitus, hypertension, renal disease, liver disease, cancer, myocardial infarction, cardioversion and catheter ablations, dyspepsia or stomach discomfort, non-stroke/SE peripheral vascular disease, stroke/SE, transient ischemic attack, anemia and coagulation defects, alcoholism, peripheral artery disease, coronary artery disease, baseline medication use, and time-varying covariates during the follow-up HR hazard ratio, CI confidence interval, SE systemic embolism and access issues, which we were unable to capture but have differences in non-persistence risk among OACs, including been documented as possible indicators for treatment switch drug tolerability and meal requirements for certain DOACs or discontinuation [29]. [28], which we were unable to ascertain because of data Nevertheless, our study is a major contribution to the lit- limitations. Among baseline characteristics, increasing erature on real-world persistence with OAC, since it is by age, statin use, and history of hypertension or stroke were far the largest retrospective observational study, with over associated with a lower risk of non-persistence. 1 million patients, examining the comparative risk of non- persistence between OACs. By pooling five datasets and 4.1 Limitations including a comprehensive comparison of the OACs, this study adds supplemental information to the literature and As with many real-world studies, our study has several limi- may assist in decisions around treatment selection for stroke tations. This study was designed to examine non-persistence prevention among patients with NVAF. The robustness of among patients with NVAF initiated on OACs and predictors our findings is evidenced by the similar results observed of non-persistence, so we could not evaluate causal rela- after limiting follow-up to the initial 12 months after OAC tionships. As is the nature with retrospective observational initiation, shortening of the gap length from 60 to 30 days, studies, our study was subject to confounders. This study and the inclusion of INR results in lengthening warfarin was bound by the limitations of the claims data; variables treatment line. such as over-the-counter use of aspirin, serum creatinine/ creatinine clearance, and laboratory values were unavailable and thus were not controlled for in the model. Codes from the International Classification of Diseases, Ninth and Tenth 5 Conclusion Revisions, Clinical Modification were used to identify base- line characteristics and outcomes, which may lack clinical In this real-world study of over 1 million patients with NVAF, accuracy. Additionally, we were unable to determine time in our results suggest that noteworthy differences in anticoagu- therapeutic range for patients prescribed warfarin or jointly lation treatment persistence exist across OAC agents, even assess the relationship between persistence and adherence. after accounting for clinical events after OAC initiation. It is Furthermore, we were unable to identify predictors related important for clinicians and patients to take these differences to cost or access. Our results might in fact have been driven into consideration, especially as non-persistence to OAC by non-medical reasons, including out-of-pocket costs, for- therapy is associated with thromboembolic complications, mulary changes, insurance changes, physician preferences, including stroke, MB, and all-cause mortality. A. D. Dhamane et al. Supplementary Information The online version contains supplemen- for thromboprophylaxis. J Am Coll Cardiol. 2010;56(11):827– tary material available at https://doi. or g/10. 1007/ s40256- 021- 00501-w . 37. https:// doi. org/10. 1016/j. jacc. 2010. 05. 028. 3. Sankaranarayanan R, Kirkwood G, Visweswariah R, et al. How does chronic atrial fibrillation influence mortality in the modern Declarations treatment era? Curr Cardiol Rev. 2015;11(3):190–8. https://d oi. org/10. 2174/ 15734 03x10 66614 09021 43020. Funding This study was sponsored by Bristol Myers Squibb and Pfizer. 4. Pastori D, Menichelli D, Del Sole F, Pignatelli P, Violi F. Long-term risk of major adverse cardiac events in atrial fibril- lation patients on direct oral anticoagulants. Mayo Clin Proc. Conflict of interest Amol Dhamane and Mauricio Ferri are paid em- 2021;96(3):658–65. https://doi. or g/10. 1016/j. ma yocp.2020. 06. ployees of Bristol Myers Squib. Inmaculada Hernandez is an employee of the University of Pittsburgh. Manuela Di Fusco, Cristina Russ, and 5. Vranckx P, Valgimigli M, Heidbuchel H. The significance of Birol Emir are paid employees and shareholders of Pfizer. Allison Ke- drug-drug and drug-food interactions of oral anticoagulation. shishian, Cynthia Gutierrez, and Wan-Lun Tsai are paid employees Arrhythm Electrophysiol Rev. 2018;7(1):55–61. https://doi. or g/ of STATinMED Research, which is a paid consultant to Pfizer and 10. 15420/ aer.2017. 50.1. Bristol Myers Squibb. Huseyin Yuce has no conflicts of interest that 6. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban ver- are directly relevant to the content of this article. sus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365(11):981–92. https:// doi. org/ 10. 1056/ NEJMo a1107 Ethics approval Since this study did not involve the collection, use, or transmittal of individually identifiable data, it was exempt from Insti- 7. Connolly SJ, Ezekowitz MD, Yusuf S, et  al. Dabigatran ver- tutional Review Board review. Both the datasets and the security of sus warfarin in patients with atrial fibrillation. N Engl J Med. the offices where analysis was completed (and where the datasets are 2009;361(12):1139–51. https://doi. or g/10. 1056/ NEJMo a0905 561 . kept) meet the requirements of the Health Insurance Portability and 8. Patel MR, Mahaffey KW, Garg J, et  al. Rivaroxaban ver- Accountability Act of 1996. sus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365(10):883–91. https://doi. or g/10. 1056/ NEJMo a1009 638 . Consent to participate Not applicable. 9. Giugliano RP, Ruff CT, Braunwald E, et  al. Edoxaban ver- sus warfarin in patients with atrial fibrillation. N Engl J Med. Consent for publication Not applicable. 2013;369(22):2093–104. https:// doi. org/ 10. 1056/ NEJMo a1310 Data availability The data, analytical methods, and study materials will 10. Rivera-Caravaca JM, Esteve-Pastor MA, Roldán V, et al. Non-vita- not be made available to other researchers for purposes of reproducing min K antagonist oral anticoagulants: impact of non-adherence the results or replicating the procedure. and discontinuation. Expert Opin Drug Saf. 2017;16(9):1051–62. https:// doi. org/ 10. 1080/ 14740 338. 2017. 13515 42. Code availability Code not publicly available. 11. Lowres N, Giskes K, Hespe C, Freedman B. Reducing stroke risk in atrial fibrillation: adherence to guidelines has improved, but Author Contributions Conceptualization: all authors; Methodology: all patient persistence with anticoagulant therapy remains subop- authors; Formal analysis and investigation: Allison Keshishian, Cynthia timal. Korean Circ J. 2019;49(10):883–907. https:// doi. org/ 10. Gutierrez, and Wan-Lun Tsai; Writing - original draft preparation: Alli- 4070/ kcj. 2019. 0234. son Keshishian and Cynthia Gutierrez; Writing - review and editing: 12. Ozaki AF, Choi AS, Le QT, et al. Real-world adherence and per- all authors; Funding acquisition: Amol D Dhamane Manuela Di Fusco, sistence to direct oral anticoagulants in patients with atrial fibrilla- Mauricio Ferri, Cristina Russ, Birol Emir. tion: a systematic review and meta-analysis. Circ Cardiovasc Qual Outcomes. 2020;13(3):e005969. https://doi. org/ 10. 1161/ CIRCO Open Access This article is licensed under a Creative Commons Attri- UTCOMES. 119. 005969. bution-NonCommercial 4.0 International License, which permits any 13. Wiley V, Franchina-Elder J, Fu AC, et al. Treatment and persis- non-commercial use, sharing, adaptation, distribution and reproduction tence with oral anticoagulants among newly diagnosed patients in any medium or format, as long as you give appropriate credit to the with non-valvular atrial fibrillation: a retrospective observational original author(s) and the source, provide a link to the Creative Com- study in a US commercially insured and Medicare Advantage mons licence, and indicate if changes were made. The images or other population. BMJ Open. 2018;8:e020676. https:// doi.or g/10. 1136/ third party material in this article are included in the article's Creative bmjop en-2017- 020676. Commons licence, unless indicated otherwise in a credit line to the 14. Mitchell SA, Simon TA, Raza S, et al. The efficacy and safety of material. If material is not included in the article's Creative Commons oral anticoagulants in warfarin-suitable patients with nonvalvular licence and your intended use is not permitted by statutory regula- atrial fibrillation: systematic review and meta-analysis. Clin Appl tion or exceeds the permitted use, you will need to obtain permission Thromb Hemost. 2013;19(6):619–31. https:// doi. org/ 10. 1177/ directly from the copyright holder. To view a copy of this licence, visit 10760 29613 486539. http:// creat iveco mmons.or g/ licen ses/ by- nc/4. 0/. 15. Lip GY, Keshishian A, Li X, et al. Ee ff ctiveness and safety of oral anticoagulants among nonvalvular atrial fibrillation patients: the ARISTOPHANES study. Stroke. 2018;49(12):2933–44. https:// doi. org/ 10. 1161/ STROK EAHA. 118. 020232. 16. Li X, Deitelzweig S, Keshishian A, et  al. Effectiveness and References safety of apixaban versus warfarin in non-valvular atrial fibrilla- tion patients in “real-world” clinical practice. Thromb Haemost. 1. Shea JB, Sears SF. Cardiology patient pages. A patient’s guide to 2017;117(6):1072–82. https:// doi. org/ 10. 1160/ TH17- 01- 0068. living with atrial fibrillation. Circulation. 2008;117(20):340– 3. 17. Jensen PN, Johnson K, Floyd J, Heckbert SR, Carnahan R, Dublin https:// doi. org/ 10. 1161/ CIRCU LATION AHA. 108. 780577. S. A systematic review of validated methods for identifying atrial 2. Marinigh R, Lip GY, Fiotti N, Giansante C, Lane DA. Age as a fibrillation using administrative data. Pharmacoepidemiol Drug risk factor for stroke in atrial fibrillation patients: implications Saf. 2012;21(suppl 1):141–7. https:// doi. org/ 10. 1002/ pds. 2317. Oral Anticoagulant Non-persistence in Patients with NVAF 18. Hellfritzsch M, Grove EL, Husted SE, Rasmussen L, Poulsen 25. Bancroft T, Lim J, Wang C, Sander SD, Swindle JP. Health care BK, Johnsen SP, Hallas J, Pottegård A. Clinical events preceding resource utilization, costs, and persistence in patients newly diag- switching and discontinuation of oral anticoagulant treatment in nosed as having nonvalvular atrial fibrillation and newly treated patients with atrial fibrillation. Ep Europace. 2017;19(7):1091 –5. with dabigatran versus warfarin in the United States. Clin Ther. https:// doi. org/ 10. 1093/ europ ace/ euw241. 2016;38(3):545–56. https:// doi. org/ 10. 1016/j. clint hera. 2016. 01. 19. Cunningham A, Stein CM, Chung CP, Daugherty JR, Smalley 008. WE, Ray WA. An automated database case definition for serious 26. Menichelli D, Del Sole F, Di Rocco A, et al. Real-world safety and bleeding related to oral anticoagulant use. Pharmacoepidemiol efficacy of direct oral anticoagulants in atrial fibrillation: a sys- Drug Saf. 2011;20(6):560–6. https:// doi. org/ 10. 1002/ pds. 2109. tematic review and meta-analysis of 605 771 patients. Eur Heart 20. Thigpen JL, Dillon C, Forster KB, et  al. Validity of interna- J Cardiovasc Pharmacother. 2021;7(FI1):f11–9. https:// doi. org/ tional classification of disease codes to identify ischemic stroke 10. 1093/ ehjcvp/ pvab0 02. and intracranial hemorrhage among individuals with associated 27. Manzoor BS, Walton SM, Sharp LK, Galanter WL, Lee TA, diagnosis of atrial fibrillation. Circ Cardiovasc Qual Outcomes. Nutescu EA. High number of newly initiated direct oral antico- 2015;8(1):8–14. agulant users switch to alternate anticoagulant therapy. J Thromb 21. Lip GYH, Pan X, Kamble S, Kawabata H, Mardekian J, Mas- Thrombolysis. 2017;44(4):435–41. https:// doi. org/ 10. 1007/ seria C, Phatak H. Discontinuation risk comparison among ‘real- s11239- 017- 1565-2. world’ newly anticoagulated atrial fibrillation patients: Apixaban, 28. Xarelto. Package insert. Janssen Pharmaceuticals, Inc. Food and warfarin, dabigatran, or rivaroxaban. PLoS ONE. 2018;13(4): Drug Administration website. https:// www. acces sdata. fda. gov/ e0195950. https:// doi. org/ 10. 1371/ journ al. pone. 01959 50. drugs atfda_ docs/ label/ 2017/ 20243 9s021 lbl. pdf. Published May 22. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, et al. Anti- 2016. Accessed 10 Dec 2020. coagulation therapy for stroke prevention in atrial fibrillation: how 29. Jackson LR 2nd, Kim S, Shrader P, et al. Early therapeutic per- well do randomized trials translate into clinical practice? JAMA. sistence on dabigatran versus warfarin therapy in patients with 2003;290(20):2685–92. https:// doi. org/ 10. 1001/ jama. 290. 20. atrial fibrillation: results from the Outcomes Registry for Better 2685. Informed Treatment of Atrial Fibrillation (ORBIT-AF) registry. 23. Rosendaal FR, Cannegieter SC, Van der Meer FJ, Briet EA. J Thromb Thrombolysis. 2018;46(4):435–9. https:// doi. org/ 10. Method to determine the optimal intensity of oral anticoagulant 1007/ s11239- 018- 1715-1. therapy. Thromb Haemost. 1993;69(3):236–9. 24. Nelson WW, Song X, Coleman CI, Thomson E, Smith DM, Damaraju CV, Schein JR. Medication persistence and discontinu- ation of rivaroxaban versus warfarin among patients with non-val- vular atrial fibrillation. Curr Med Res Opin. 2014;30(12):2461– 9. https:// doi. org/ 10. 1185/ 03007 995. 2014. 933577.

Journal

American Journal of Cardiovascular DrugsSpringer Journals

Published: Oct 21, 2021

References