Comparative effectiveness of warfarin, dabigatran, rivaroxaban and apixaban in non-valvular atrial fibrillation: A nationwide pharmacoepidemiological study
Comparative effectiveness of warfarin, dabigatran, rivaroxaban and apixaban in non- valvular atrial fibrillation: A nationwide pharmacoepidemiological study
Lars J. KjerpesethID 0 1
Randi Selmer 1
Inger Ariansen 1
?ystein Karlstad 1
Hanne Ellekjaer 1
Eva Skovlund 0 1
0 Department of Public Health and Nursing, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology (NTNU) , Trondheim, Norway, 2 Chronic Diseases and Ageing , Division of Mental and Physical Health, Norwegian Institute of Public Health , Oslo , Norway , 3 Department of Neuromedicine and Movement Science, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology (NTNU) , Trondheim , Norway , 4 Stroke Unit, Department of Internal Medicine, St. Olav's Hospital , Trondheim , Norway
1 Editor: Pablo Garcia de Frutos, Institut d'Investigacions Biomediques de Barcelona , SPAIN
Data Availability Statement: The study is based
on data from the Norwegian Prescription Database,
the Norwegian Cause of Death Registry (both held
by the Norwegian Institute of Public Health) and
the Norwegian Patient Registry (held by the
Norwegian Directorate of Health). The data cannot
be shared publicly because of the Norwegian data
protection legislation. Qualifying researchers can
apply for access to relevant data with the
Norwegian Institute of Public Health and the
Norwegian Directorate of Health upon the approval
To compare effectiveness and safety of warfarin and the direct oral anticoagulants (DOAC)
dabigatran, rivaroxaban and apixaban in non-valvular atrial fibrillation in routine care.
From nationwide registries, we identified treatment-na?ve patients initiating warfarin,
dabigatran, rivaroxaban or apixaban for non-valvular atrial fibrillation from July 2013 to December
2015 in Norway. We assessed prescription duration using reverse waiting time distribution.
Adjusting for confounding in a Cox proportional hazards model, we estimated one-year risks
for ischemic stroke, transient ischemic attack (TIA) or systemic embolism, major or clinically
relevant non-major bleeding; intracranial; gastrointestinal; and other bleeding. We censored
at switch of treatment or 365 days of follow-up.
We included 30,820 treatment-na?ve patients. Compared to warfarin, the adjusted hazard
ratios (HR) for ischemic stroke, TIA or systemic embolism were 0.96 (95% CI 0.71?1.28) for
dabigatran, 1.12 (95% CI 0.87?1.45) for rivaroxaban and 0.97 (95% CI 0.75?1.26) for
apixaban. Corresponding hazard ratios for major or clinically relevant non-major bleeding were
0.73 (95% CI 0.62?0.86) for dabigatran, 0.97 (95% CI 0.84?1.12) for rivaroxaban and 0.71
(95% CI 0.62?0.82) for apixaban. Statistically significant differences of other safety
outcomes compared to warfarin were fewer intracranial bleedings with dabigatran (HR 0.28,
95% CI 0.14?0.56), rivaroxaban (HR 0.40, 95% CI 0.23?0.69) and apixaban (HR 0.56, 95%
CI 0.34?0.92); fewer gastrointestinal bleedings with apixaban (HR 0.70, 95% CI 0.52?0.93);
from the Regional Committees for Medical and
Health Research Ethics. For further details, please
contact Vidar Hjellvik, Norwegian Institute of Public
Health, Oslo, Norway ().
Funding: The study is funded by an internal grant
from the Norwegian University of Science and
Technology. The funder had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
and fewer other bleedings with dabigatran (HR 0.67, 95% CI 0.55?0.81) and apixaban (HR
0.70, 95% CI 0.59?0.83).
After 1 year follow-up in treatment-na?ve patients initiating oral anticoagulation for
non-valvular atrial fibrillation, all DOACs were similarly effective as warfarin in prevention of
ischemic stroke, TIA or systemic embolism. Safety from bleedings was similar or better, including
fewer intracranial bleedings with all direct oral anticoagulants, fewer gastrointestinal
bleedings with apixaban and fewer other bleedings with dabigatran and apixaban.
European guidelines recommend prophylactic oral anticoagulation in patients with
non-valvular atrial fibrillation who have a moderate to high risk of stroke . Warfarin has been the
mainstay for oral anticoagulation, but requires frequent monitoring and dose adjustments due
to a narrow therapeutic window and many interactions with food and drugs . In the last
decade, easier-to-use direct-acting oral anticoagulants (DOACs) such as dabigatran,
rivaroxaban and apixaban have proven as effective and safe as warfarin for stroke prevention in large
randomized controlled trials [3?5]. The DOACs have been quickly incorporated in European
guidelines on oral anticoagulation in atrial fibrillation [1, 6]. Among users of oral
anticoagulation for atrial fibrillation in Norway, we have seen a shift in market shares from complete
warfarin coverage in 2010 to a market share of more than 80% DOACs in new users and 50% in
prevalent users in 2015 [7, 8]. Other countries have also seen a rapid uptake in use of DOACs
for atrial fibrillation [9?16].
These changes in routine clinical care may have huge implications on the public health
burden. Atrial fibrillation is common, especially among the elderly, and its prevalence and
associated complications are expected to surge in the next decades owing to an ageing population
and increase in predisposing risk factors such as obesity, diabetes and unhealthy lifestyle .
The introduction of DOACs has increased the drug repertoire, but also complicated
decisionmaking for prescribers and patients considering oral anticoagulation . While each DOAC
has been tested against warfarin, head-to-head trials of DOACs are lacking and indirect
comparisons of the trials are difficult because of differences in study populations. Also, the
standardized approach to treatment and highly selected patients in clinical trials may limit their
generalizability. Patients with severe renal impairment were excluded from the DOAC trials
and especially dabigatran requires a high renal clearance. Other concerns with regard to
applicability of the trial results in routine care are elderly patients with multiple comorbidities,
polypharmacy, and compliance issues . Taken together, these issues warrant observational
comparative effectiveness studies in everyday clinical practice [19, 20].
Although a number of such studies have been performed, few have compared both
dabigatran, rivaroxaban and apixaban separately to warfarin in the same study population [21, 22].
In addition, many of these studies are based on selected populations [21, 22]. In contrast, the
cradle-to-grave health registers of Norway allow inclusion of a nationwide patient population
from primary and secondary care, and with little loss to follow-up . Furthermore,
Norway?s tax-supported universal healthcare system funds nearly all patient expenses and insures
equal access to warfarin and the more expensive DOACs . And while Halvorsen et al.
investigated the safety of oral anticoagulants in atrial fibrillation using Norwegian registers,
2 / 18
they did not address effectiveness outcomes . Thus, we have conducted a study to compare
both effectiveness and safety of warfarin, dabigatran, rivaroxaban and apixaban in an
unselected and nationwide population in Norway.
Materials and methods
We have performed a register-based cohort study from 15 July 2013 to 31 December 2015.
Dabigatran, rivaroxaban and apixaban received Norwegian marketing authorization for
nonvalvular atrial fibrillation in August 2011, December 2011 and November 2012, respectively.
15 July 2013 was chosen as start date of this study because this was when apixaban, as the last
study drug, received general reimbursement for treatment of atrial fibrillation, and
reimbursement codes were used to identify patients treated for atrial fibrillation. This also allowed for a
run-in period for the DOACs in general and attenuated potential bias from including early
adopters of DOACs . We adopted an active comparator, new-user design to reduce
prevalent user bias and confounding by indication [27, 28].
Using the personal identification number unique to all Norwegian residents, we linked data
from four nationwide databases: the Norwegian Prescription Database, the Norwegian Patient
Registry, the Norwegian Cause of Death Registry and the National Registry. The Norwegian
Prescription Database contains the dispensing date, Anatomical Therapeutic Chemical (ATC)
classification code, package size, tablet strength, and reimbursement code of every prescription
claimed since 2004. The indication for use is indicated by the reimbursement codes that since
2008 have been coded according to the International Classification of Diseases, 10th revision
(ICD-10) and the International Classification of Primary Care, 2nd Edition (ICPC-2). The
Norwegian Patient Registry contains discharge dates and ICD diagnoses of in- and outpatient
visits to hospitals and private specialist practices on an individual level since 2008. Up to two
primary diagnoses and (potentially) an unlimited number of secondary diagnoses are recorded
at each in- and outpatient visit. The Norwegian Cause of Death Registry has records of all
deaths since 1951, and uses an international algorithm to ensure uniform assessment of the
underlying cause of death and code it according to the ICD coding system. The National
Registry is a civil registration system of all residents that includes information on sex, date of birth,
and date of emigration, death and other changes in resident status.
We identified all individuals who had been oral anticoagulant-na?ve since 2004 and then
received at least one dispensing of a study drug between 15 July 2013 and 31 December 2015.
We further restricted the population to adults 18 years on the date of the first dispensing
who received a reimbursement code for atrial fibrillation/-flutter (ICD-10 code I48 or ICPC-2
code K78) on the first dispensing and had no previous history of mitral stenosis or prosthetic
heart valves (?valvular? atrial fibrillation) . We excluded individuals who on the date of first
dispensing received more than one type of oral anticoagulant or reimbursement code, or a
tablet strength not approved for treatment of atrial fibrillation. We also excluded a few patients
who failed linkage, had an uncertain resident status or could not be followed, e.g. citizens
living abroad (Fig 1 and S1 Table).
3 / 18
Fig 1. Study population selection. Flowchart showing selection of the final study population of new users of warfarin,
dabigatran, rivaroxaban or apixaban for non-valvular atrial fibrillation from 15 July 2013 to 31 December 2015 in
Norway by linkage of nationwide registries.
Exposure to oral anticoagulants
We considered a patient?s use of oral anticoagulation to start on the date he or she first
received a dispensing for a study drug (index date), and continue as long as the patient
received a new dispensing within the duration of the drug supply of the preceding dispensing.
We estimated the duration of each dispensing using reverse parametric waiting time
distribution [29, 30]. According to a validation study using warfarin as a case, which has a large
interindividual variation in daily dose, this approach gives less bias and higher precision than
traditional methods assuming a fixed daily dose . The method considers the time from the
last dispensing of each patient before an index time point, or reverse waiting time. From the
distribution of these reverse waiting times, one can establish maximum likelihood estimates of
specified interarrival percentiles, or the time it takes for a percentile of continued users to refill
a prescription. If a patient has not redeemed a new prescription within the specified percentile,
the treatment is considered to have stopped. Furthermore, the method can take into account
covariates predictive of the time from one dispensing to the next, such as age, sex and amount
filled. This can reduce misclassification on the individual level . In practice, we pooled all
dispensings of a study drug to new users within the last year before 1 November 2015, and
used their last filling in this time interval in a reverse parametric waiting time distribution
model with 90th percentile interarrival distribution and adjustment for age, sex, and number
of tablets dispensed. We repeated the analysis for each study drug separately and used the
results to predict dispensing durations for subjects under study. We used 1 November 2015 as
the index time point to avoid bias from seasonal stockpiling of study drugs towards the end of
the calendar year, which is common in Norway due to the reimbursement rules. We used a
4 / 18
log-normal parametric model as this provides a better fit than other models . The analyses
can be automated using the publically available wtdttt package for Stata.
We included both primary and secondary diagnoses of the Norwegian Patient Registry in the
assessment of outcomes. From the Cause of Death Registry, we included deaths with the
outcomes as the underlying cause. Our primary effectiveness outcome was the composite of
hospitalizations with a diagnosis of ischemic stroke, transient ischemic attack (TIA) or systemic
embolism (arterial thromboembolism outside the brain or lungs), or death from ischemic
stroke or systemic embolism. Secondary effectiveness outcomes were hospitalization or death
from ischemic stroke or systemic embolism; hospitalization or death from ischemic stroke;
and hospitalization with TIA. The primary safety outcome, major or clinically relevant
nonmajor bleeding, was a composite of the secondary safety outcomes, which were intracranial
bleeding, gastrointestinal bleeding and other bleeding. We defined intracranial bleeding as
hospitalization or death from this diagnosis. Gastrointestinal and other bleeding were similarly
defined, with the addition of outpatient diagnoses however, since treatment of such bleedings
does not necessarily require hospitalization. Since it can be difficult to discern the cause of
intracranial bleedings in register data, we included diagnosis codes of both traumatic and
nontraumatic bleedings . Similarly, the ischemic stroke outcome included diagnosis codes of
both ischemic and unspecified stroke since the latter will be ischemic in most cases . See S2
Table for further details on diagnosis codes.
We used the patients? sex, age, county of residence (19 in total), comorbidities, concomitant
drug use, the number of hospitalizations (
0, 1, 2
) and the number of outpatient visits (
) in the 6 months before index date to adjust for potential confounding. We defined
comorbidities from in- and outpatient primary and secondary diagnoses within 5 years before or on
index date, and from reimbursement codes for drugs within the same time frame. We
considered drugs dispensed within the last 6 months as concomitant drug use. A detailed description
of all covariates is found in S3 Table. We calculated CHA2DS2-VASc score (congestive heart
failure, hypertension, age 75, diabetes, stroke/TIA/systemic embolism, vascular disease, age
65?74, and sex [female]) and modified HAS-BLED score (hypertension, abnormal liver/kidney
function, stroke, bleeding history, age >65, alcohol abuse or concomitant use of non-steroidal
anti-inflammatory drugs or platelet inhibitors) for each patient at index date (see S4 Table for
calculation of the scores).
Patients were followed from the index date until the first event of the outcome under study, or
censoring. We censored patients at switch to another oral anticoagulant; switch to a tablet
strength not approved for atrial fibrillation; end of drug supply; death (if not caused by the
outcome under study), emigration; 365 days of follow-up; or study end. To avoid bias from
differing follow-up times due to differences in the uptake of dabigatran, rivaroxaban and apixaban,
we censored follow-up for all study drugs at 365 days. In addition, baseline adjustment of
time-varying covariates such as age and co-medications get less valid with longer follow-up.
We conducted all analysis in Cox proportional hazards models to estimate hazard ratios (HR)
with 95% confidence intervals, after checking the proportional hazards assumption. We report
effect estimates for three models in the main analyses and the sensitivity analyses: crude,
adjusted for risk factors in the CHA2DS2-VASc and modified HAS-BLED scores (partially
5 / 18
adjusted model), and adjusted for all predefined covariates (fully adjusted model). The results
are reported together with median follow-up time, number of events and one-year crude
incidence rate for each study drug. In a supplementary analysis, we made plots of the cumulative
incidence of each outcome during one year, adjusted for competing risk from deaths not
caused by the event (S1 and S3 Figs).
We repeated the primary effectiveness and safety outcome analyses for several subgroups:
patients with and without specialist-diagnosed atrial fibrillation; history or no history of
vascular disease (i.e. acute myocardioal infarction, atherosclerosis or peripheral vascular disease),
stroke/TIA/systemic embolism, serious bleeding, or chronic kidney disease; concomitant or
no concomitant use of acetylsalicylic acid or platelet inhibitor; age 75 and <75 years; male
and female sex; CHA2DS2-VASc 2 and <2; and modified HAS-BLED 3 and <3;
highintensity DOAC users; and low-intensity DOAC users. In the high-intensity and low-intensity
DOAC subgroup analyses, we respectively compared rivaroxaban 20 mg and apixaban 5 mg to
dabigatran 150 mg and rivaroxaban 15 mg and apixaban 2.5 mg to dabigatran 110 mg, and we
censored patients who switched to another tablet strength. We only report estimates from the
fully adjusted model for the subgroup analyses (S2 and S4 Figs).
We performed several sensitivity analyses on the primary outcomes to test the validity of
our results (S5 and S6 Tables). First, we repeated the analyses including only events registered
as a primary diagnosis at hospitalization (thereby excluding events registered as secondary
diagnoses or underlying cause of death). Second, we repeated the analyses using 50th, 60th,
70th, 80th and 99th interarrival density percentiles in the reverse parametric waiting time
distribution model to estimate the duration of dispensings (90th percentile in the main analyses).
Third, we repeated the analyses using a fixed dose model where we estimated the duration of a
dispensing from the number of tablets dispensed. We assumed a dosing regimen of one tablet
twice daily for dabigatran and apixaban and one tablet once daily for rivaroxaban. For
warfarin, we calculated a mean daily tablet dose for the study population using the accumulated
time from a dispensing of warfarin to the next dispensing as the denominator and the total
amount of tablets of warfarin in these dispensings as the nominator. We allowed a 30-day gap
in the oral anticoagulant supply (grace period) in this model.
To test how well we were able to control for confounding, we performed a sensitivity
analysis using a presumed ?neutral? outcome, hospitalization or death from pneumonia, as a
negative control  (S7 Table). Pneumonia have many overlapping risk factors with stroke and
bleeding, such as old age, underlying medical conditions and lifestyle factors , which we
expect to contribute to confounding since they are known to influence the choice of oral
anticoagulant in atrial fibrillation [8, 36, 37]. Channeling of patients towards specific study drugs
based on these risk factors would lead to differences in the risk of pneumonia associated with
each drug. A similar risk of pneumonia between drug groups after adjustment for covariates
would thus provide indirect evidence of no residual confounding.
We used Stata/SE version 15.0 to analyze the data and considered two-sided p-values <0.05
as statistically significant. The South East Regional Committee for Medical and Health
Research Ethics approved the study and granted a waiver of the requirement for obtaining
patient consent (2010/131). The Norwegian Data Protection Authority gave a license to link
registry data (13/00577-4/CGN). Reporting conform to the Strengthening the Reporting of
Observational Studies in Epidemiology (STROBE) statement (S1 Checklist).
We identified 30,820 new users of warfarin, dabigatran, rivaroxaban or apixaban for
non-valvular atrial fibrillation from 15 July 2013 to 31 December 2015 (Table 1). Sixty-nine percent
6 / 18
7 / 18
TIA: transient ischemic attack. PPI: proton pump inhibitor. CHA2DS2-VASc: congestive heart failure, hypertension, age 75 [doubled], previous stroke/TIA/systemic
embolism [doubled], vascular disease, age 65?74, female sex. Modified HAS-BLED: hypertension, abnormal liver/kidney function, previous stroke/TIA/systemic
embolism, bleeding history, age >65, alcohol abuse or concomitant use of non-steroidal anti-inflammatory drugs or platelet inhibitor.
had previously been diagnosed with atrial fibrillation during hospitalization or an outpatient
visit, and 12% had valvular disease such as mitral/aortic/tricuspid insufficiency or
aortic/tricuspid stenosis (patients with mitral stenosis or heart valve prosthesis had been excluded from
the study cohort). Their mean age was 73.3 years and 47% were elderly ( 75 years). Women
constituted 44% of the cohort. Eighty-four percent had a CHA2DS2-VASc score of 2, and
47% a modified HAS-BLED score of 3. Fifteen percent had 15 or more prescription drugs
dispensed in the last 6 months before starting oral anticoagulation. Initiators of warfarin or
apixaban were generally older and had more comorbidities and higher scores on
VASc and modified HAS-BLED than users of dabigatran or rivaroxaban.
8 / 18
Between study entry and study end on 31 December 2015, 1965 patients (6.4%) died; 649
(2.1%) had an ischemic stroke; 271 (0.9%) had a TIA; 42 (0.1%) had a systemic embolism; 246
(0.8%) had an intracranial bleeding; 812 (2.6%) had a gastrointestinal bleeding; and 2091
(6.8%) had a bleeding in another location. For all outcomes in the main analyses, median
follow-up was longest for warfarin users and shortest for rivaroxaban users and apixaban users
Cumulative incidence curves for the effectiveness outcomes are presented in S1 Fig. After
adjustment for all covariates, the risk of ischemic stroke, TIA or systemic embolism differed
little from warfarin for both dabigatran, rivaroxaban and apixaban (Table 2 and Fig 2).
Although the estimates were imprecise, we observed similar results in the subgroup analyses,
including the high- and low-intensity DOAC subgroups were the risk of the primary
effectiveness outcome did not differ for rivaroxaban or apixaban compared to dabigatran (S2 Fig).
Likewise, the risks were similar compared to warfarin for the secondary effectiveness outcomes
ischemic stroke or systemic embolism; ischemic stroke; and TIA (Table 2 and Fig 2).
Cumulative incidence curves for the safety outcomes are presented in S3 Fig. Compared to
warfarin, dabigatran and apixaban were superior with regard to the risk of major or clinically
relevant non-major bleeding after adjustment for all covariates, while patients receiving
rivaroxaban had a similar risk (Table 2 and Fig 2). We observed the same trend for the direction of
the hazard ratios in all subgroups comparing DOACs to warfarin, although it was less clear
among patients with a history of stroke/TIA/systemic embolism, major bleeding, or chronic
kidney disease (S4 Fig). Compared to dabigatran, rivaroxaban use was associated with a higher
risk of major or clinically relevant non-major bleeding in the high-intensity DOAC subgroup,
while the risk was similar with rivaroxaban in the low-intensity DOAC subgroup and with
apixaban in both DOAC subgroups (S4 Fig).
The crude incident rate for intracranial bleeding was higher with use of warfarin than any
DOAC, and the risk for this outcome remained lower for all three DOACs after adjustment for
all covariates. The risk of gastrointestinal bleeding was comparable to warfarin for both
dabigatran and rivaroxaban. In contrast, the risk was lower in apixaban users than warfarin users.
With regard to bleedings in locations other than the intracranial and gastrointestinal cavities,
the risk compared to those using warfarin was lower for patients using dabigatran or apixaban,
and similar for rivaroxaban users (Table 2 and Fig 2).
The sensitivity analysis on the primary outcomes using the fixed dose model and the reverse
parametric waiting time distribution model with different percentiles of interarrival density
yielded similar results as the main analyses (S5 and S6 Tables). The risk of ischemic stroke,
TIA or systemic embolism with rivaroxaban compared to warfarin was higher with the fixed
dose model however, and reached statistical significance. Using only primary diagnosis events
in the effectiveness outcome resulted in a lower risk for the primary effectiveness outcome
with all three DOACs compared to warfarin, the estimates were uncertain due to fewer total
events however. Including only primary diagnosis events in the primary safety outcome did
not markedly change the results from the main analysis. In the analysis on the negative control
outcome, we observed a lower crude risk of hospitalization or death from pneumonia with all
DOACs compared to warfarin. The difference diminished with increasing adjustment for
covariates, although not completely (S7 Table).
Comparing the DOACs dabigatran, rivaroxaban or apixaban to warfarin, we did not find any
decisive differences in the risk of ischemic stroke, TIA or systemic embolism; ischemic stroke
or systemic embolism; ischemic stroke; and TIA. Small differences could not be excluded
9 / 18
PLOS ONE | https://doi.org/10.1371/journal.pone.0221500
CHA2DS2-VASc: congestive heart failure, hypertension, age 75 [doubled], previous stroke/TIA/systemic embolism [doubled], vascular disease, age 65?74, female sex.
Modified HAS-BLED: hypertension, abnormal liver/kidney function, previous stroke/TIA/systemic embolism, bleeding history, age >65, alcohol abuse or concomitant
use of non-steroidal anti-inflammatory drugs or platelet inhibitor.
?: Hazard ratios for the risk of an outcome for dabigatran, rivaroxaban or apixaban compared to warfarin in Cox proportional hazard models with no adjustment
(crude); adjusted for year and risk factors of CHA2DS2-VASc and modified HAS-BLED (partially adjusted); and adjusted for all predefined covariates (fully adjusted).
however. Furthermore, in the current study dabigatran and apixaban were superior to warfarin
for the risk of major or clinically relevant non-major bleeding, while the risk was similar with
rivaroxaban. We found that the risk of intracranial bleeding was lower with all three DOACs
than warfarin however. With regard to gastrointestinal bleeding, we observed a lower risk with
apixaban than warfarin and comparable risk with dabigatran or rivaroxaban, while other
bleedings were less common with dabigatran or apixaban than warfarin and similar with
Similar findings have been observed in both randomized controlled trials and observational
studies. In the trials, the risk of ischemic or unspecified stroke did not differ from warfarin
with rivaroxaban or apixaban. For dabigatran, the risk was lower with the high-intensity dose
and numerically higher with low-intensity dose in the RE-LY trial, which is consistent with no
net better efficacy when we compared both regimens to warfarin in the present study. A
metaanalysis of 28 observational studies reported similar rates of ischemic stroke and ischemic
stroke or systemic embolism with all three DOACs compared to vitamin K antagonists . In
accordance with our results, a Swedish study found comparable risks between all three
DOACs and warfarin for the combined outcome of TIA and ischemic and unspecified stroke
. With regard to TIA as an individual outcome, we observed no differences in risk between
individual DOACs and warfarin. A German claims-based study found an increased risk of
TIA in atrial fibrillation for DOACs combined compared to vitamin K antagonists .
However, in the latter group >99% received phenprocoumon, which differs pharmacokinetically
from warfarin . The DOAC trials did not report this outcome [3?5].
The aforementioned meta-analysis reported similar rates of major bleeding with
rivaroxaban and fewer with apixaban compared to vitamin K antagonists; more gastrointestinal
bleedings with dabigatran and rivaroxaban and fewer with apixaban; and a large reduction in
intracranial bleeding with all three DOACs. It also suggested a lower risk of major bleeding
with dabigatran than warfarin, although the result was not statistically significant. These results
are in line with the current study: Compared to warfarin, we found the risk of major or
clinically relevant non-major bleeding to be 27% lower with dabigatran, similar with rivaroxaban,
and 30% lower with apixaban; while the risk of gastrointestinal bleeding was numerically
higher with dabigatran and rivaroxaban, and significantly lower with apixaban. In randomized
controlled trials, the risk of major or minor bleeding was 9% lower with high-intensity
11 / 18
Fig 2. Effectiveness and safety of DOACs vs warfarin in non-valvular atrial fibrillation. Adjusted one-year hazard ratio (HR) and 95% confidence interval (CI)
for effectiveness and safety outcomes in new users of direct oral anticoagulants compared to warfarin for non-valvular atrial fibrillation in Norway July 2013 to
dabigatran and 22% lower with low-intensity dabigatran compared to warfarin; the risk of
major or clinically relevant non-major bleeding did not differ with rivaroxaban and was 32%
lower with apixaban; and there were more gastrointestinal bleedings with rivaroxaban and
both regimens of dabigatran and fewer with apixaban [3?5].
A lower risk of intracranial bleeding with DOACs compared to warfarin has been reported
consistently in all randomized controlled trials [3?5, 40], a finding which is also replicated in
the present study. The reason for the relatively higher risk of intracranial bleeding with
warfarin is unclear. A possible explanation could be disturbance of the hemostasis by inhibition of
coagulation factor VIIa, which forms complexes with tissue factor, highly expressed in brain
vessels, and together is a key initiator of the coagulation cascade . In another nationwide
Norwegian registry study, Gulati et al. found increased risks of intracranial bleeding with
warfarin, rivaroxaban and apixaban compared to no antithrombotic use, while the risk did not
increase with dabigatran . This finding is consistent with a particularly low risk of
intracranial bleeding with dabigatran in the current study.
With regard to bleeding in other locations than the intracranial and gastrointestinal
cavities, the risk was similar with rivaroxaban compared to warfarin, and significantly lower with
12 / 18
dabigatran and apixaban. This is consistent with findings in several other observational studies
[25, 43?45] and the ARISTOTLE trial . The RE-LY and ROCKET AF trials did not report
this outcome. A meta-analysis of observational studies comparing dabigatran, rivaroxaban and
apixaban found that the latter had the best safety profile. Our results also point in this direction
since, unlike the other DOACs, the risk for all three subtypes of bleeding was lower with
apixaban than warfarin. The current study also gives credence to studies that find DOACs,
especially dabigatran and apixaban, to be more cost-effective than warfarin . While DOACs do
not seem to be more effective than warfarin with regard to stroke prevention, their uptake was
followed by fewer ischemic strokes but no more bleedings in a regional Swedish atrial
fibrillation population . This might be mediated through a lowered threshold of oral
anticoagulation coupled with a lower bleeding risk with DOACs. The trend was also present before their
introduction however . Interestingly, our results on arterial thromboembolism and
bleeding are consistent with the major randomized controlled trials comparing DOACs and
warfarin despite the reportedly high quality of anticoagulation treatment with warfarin in Norway
[49?52]. However, this is in line with post-hoc analyses indicating that the results of the major
clinical trials are consistent across a wide range of warfarin treatment qualities, although any
advantages of the DOAC lessens with increasing warfarin treatment quality [49?51].
Using the reverse parametric waiting time distribution model to estimate prescription
length, we could define continuous use based on how long it takes for a given percentile of
users to refill a prescription. As shown in S5 and S6 Tables, changing the percentiles to 50, 60,
70, 80 or 99 did not markedly change the results for the primary effectiveness and safety
outcomes. Using a fixed daily dose and allowing up to 30 days gap in treatment yielded similar
results except for a lower effectiveness of rivaroxaban compared to warfarin.
A strength of the present study is the use of reimbursement codes to identify atrial
fibrillation as an indication for oral anticoagulation regardless of the level of care the patient receives.
While these reimbursement codes have not been validated, 69% of the study population had
an atrial fibrillation diagnosis in the Norwegian Patient Registry at the initiation of oral
anticoagulation and 80% within 3 months after initiation, and this diagnosis has a positive
predictive value of 89% . Restricting the analyses to patients with a confirmed atrial fibrillation
diagnosis at baseline did not change the conclusion with regard to the primary outcomes (S2
and S3 Figs). Apart from atrial fibrillation, few validation studies of diagnoses in the
Norwegian Patient Registry exist. However, the diagnosis of acute stroke (ischemic and hemorrhagic)
and intracranial bleeding (traumatic and non-traumatic) have been found to be of adequate
quality for epidemiological studies [32, 54]. To improve the validity of these outcomes even
further in the current study, we included only events that required hospitalization. Linking
with the Norwegian Cause of Death Registry we were also able to include fatal events that were
otherwise not detected. Reassuringly, the sensitivity analyses using only primary diagnosis
events in the outcomes did not lead to different conclusions than the main analyses.
We believe the inclusion of routine care patients from nationwide and mandatory registers
in Norway, where healthcare is universal and affordable, ensure generalizability of our results
to other atrial fibrillation populations with a similar constitution. While we did not have
information on ethnicity or race, and Norway has become more multiethnic in recent decades, the
elderly population constituting most atrial fibrillation patients is likely predominantly white
(95% of the cohort were born in Norway). Likewise, we did not have information on other
potentially prognostic factors for physicians choosing between different drugs for specific
patients, such as body mass index, consumption of alcohol and tobacco, creatinine clearance
and socioeconomic status.
The higher crude rate of hospitalization or death from pneumonia suggests that initiators
of warfarin were frailer than DOAC initiators, especially those initiating dabigatran and
13 / 18
rivaroxaban. Upon adjusting for additional covariates, the adjusted hazard for DOAC users
moved incrementally closer to that of warfarin users, but it still remained lower than for
warfarin with all three DOACs in the fully adjusted model (S7 Table). Thus, channeling of elderly
and less healthy patients towards warfarin and to some degree apixaban may have generated
some residual confounding. Still, we do not believe this largely impacted our main results as
they are consistent in subgroups, sensitivity analyses and with trials and other observational
After 1 year of follow-up in new users, all DOACs were similarly effective as warfarin in
prevention of ischemic stroke, TIA or systemic embolism, while safety from bleedings was similar
or better, including fewer intracranial bleedings with all DOACs, fewer gastrointestinal
bleedings with apixaban and fewer other bleedings with dabigatran and apixaban.
S1 Checklist. STROBE checklist of items that should be included in reports of cohort
S1 Fig. Cumulative incidence of effectiveness outcomes.
S2 Fig. Risk of primary effectiveness outcome in subgroups.
S3 Fig. Cumulative incidence of safety outcomes.
S4 Fig. Risk of primary safety outcome in subgroups.
S1 Table. Definition of inclusion and exclusion criteria.
S2 Table. Definition of study outcomes.
S3 Table. Definition of covariates.
S4 Table. Calculation of CHA2DS2-VASc and modified HAS-BLED risk scores from
variables in S3 Table.
S5 Table. Results of sensitivity analyses on primary effectiveness outcome.
S6 Table. Results of sensitivity analyses on primary safety outcome.
S7 Table. Results of sensitivity analysis on negative control outcome.
14 / 18
We would like to thank the Norwegian Prescription Database, the Norwegian Patient Registry
and the Norwegian Cause of Death Registry for supplying data for this study. The
interpretation and reporting of these data are the sole responsibility of the authors, and no endorsement
by the registries is intended nor should be inferred.
Conceptualization: Lars J. Kjerpeseth, Randi Selmer, Inger Ariansen, Hanne Ellekjaer, Eva
Data curation: Lars J. Kjerpeseth.
Formal analysis: Lars J. Kjerpeseth.
Investigation: Lars J. Kjerpeseth.
Funding acquisition: Randi Selmer, Inger Ariansen, Hanne Ellekjaer, Eva Skovlund.
Methodology: Lars J. Kjerpeseth, Randi Selmer, ?ystein Karlstad, Eva Skovlund.
Project administration: Lars J. Kjerpeseth, Randi Selmer, Eva Skovlund.
Resources: Randi Selmer, Eva Skovlund.
Software: Lars J. Kjerpeseth.
Supervision: Randi Selmer, Hanne Ellekjaer, Eva Skovlund.
Validation: Lars J. Kjerpeseth.
Visualization: Lars J. Kjerpeseth.
Writing ? original draft: Lars J. Kjerpeseth.
Writing ? review & editing: Randi Selmer, Inger Ariansen, ?ystein Karlstad, Hanne Ellekjaer,
15 / 18
16 / 18
17 / 18
1. Camm AJ , Lip GY , De Caterina R , Savelieva I , Atar D , Hohnloser SH , et al. 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association . Europace. 2012 ; 14 ( 10 ): 1385 - 413 . Epub 2012/08/28. https://doi.org/10.1093/europace/ eus305 PMID: 22923145 .
2. Holbrook AM , Pereira JA , Labiris R , McDonald H , Douketis JD , Crowther M , et al. Systematic overview of warfarin and its drug and food interactions . Arch Intern Med . 2005 ; 165 ( 10 ): 1095 - 106 . Epub 2005/ 05/25. https://doi.org/10.1001/archinte.165.10.1095 PMID: 15911722 .
3. Connolly SJ , Ezekowitz MD , Yusuf S , Eikelboom J , Oldgren J , Parekh A , et al. Dabigatran versus warfarin in patients with atrial fibrillation . N Engl J Med . 2009 ; 361 ( 12 ): 1139 - 51 . Epub 2009/09/01. https://doi. org/10.1056/NEJMoa0905561 PMID: 19717844 .
4. Patel MR , Mahaffey KW , Garg J , Pan G , Singer DE , Hacke W , et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation . N Engl J Med . 2011 ; 365 ( 10 ): 883 - 91 . Epub 2011/08/13. https://doi.org/10. 1056/NEJMoa1009638 PMID: 21830957 .
5. Granger CB , Alexander JH , McMurray JJ , Lopes RD , Hylek EM , Hanna M , et al. Apixaban versus warfarin in patients with atrial fibrillation . N Engl J Med . 2011 ; 365 ( 11 ): 981 - 92 . Epub 2011/08/30. https:// doi.org/10.1056/NEJMoa1107039 PMID: 21870978 .
6. European Heart Rhythm A , European Association for Cardio-Thoracic S , Camm AJ , Kirchhof P , Lip GY , Schotten U , et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC) . Eur Heart J . 2010 ; 31 ( 19 ): 2369 - 429 . Epub 2010/08/31. https://doi.org/10.1093/eurheartj/ehq278 PMID: 20802247 .
7. Kjerpeseth LJ , Ellekjaer H , Selmer R , Ariansen I , Furu K , Skovlund E. Trends in use of warfarin and direct oral anticoagulants in atrial fibrillation in Norway, 2010 to 2015 . Eur J Clin Pharmacol . 2017 ; 73 ( 11 ): 1417 - 25 . Epub 2017/07/25. https://doi.org/10.1007/s00228-017-2296-1 PMID: 28735494 .
8. Kjerpeseth LJ , Ellekjaer H , Selmer R , Ariansen I , Furu K , Skovlund E . Risk factors for stroke and choice of oral anticoagulant in atrial fibrillation . Eur J Clin Pharmacol . 2018 ; 74 ( 12 ): 1653 - 62 . Epub 2018/08/18. https://doi.org/10.1007/s00228-018-2540-3 PMID: 30116828 .
9. Barnes GD , Lucas E , Alexander GC , Goldberger ZD . National Trends in Ambulatory Oral Anticoagulant Use . Am J Med . 2015 ; 128 ( 12 ): 1300 - 5 e2. Epub 2015/07/07. https://doi.org/10.1016/j.amjmed. 2015 . 05 .044 PMID: 26144101; PubMed Central PMCID : PMC4658248 .
10. Gadsboll K , Staerk L , Fosbol EL , Sindet-Pedersen C , Gundlund A , Lip GYH , et al. Increased use of oral anticoagulants in patients with atrial fibrillation: temporal trends from 2005 to 2015 in Denmark . Eur Heart J . 2017 ; 38 ( 12 ): 899 - 906 . Epub 2017/01/23. https://doi.org/10.1093/eurheartj/ehw658 PMID: 28110293 .
11. Huiart L , Ferdynus C , Renoux C , Beaugrand A , Lafarge S , Bruneau L , et al. Trends in initiation of direct oral anticoagulant therapies for atrial fibrillation in a national population-based cross-sectional study in the French health insurance databases . BMJ Open . 2018 ; 8 ( 3 ): e018180 . Epub 2018 /04/01. https://doi. org/10.1136/bmjopen-2017-018180 PMID: 29602837; PubMed Central PMCID : PMC5884337 .
12. Loo SY , Dell'Aniello S , Huiart L , Renoux C. Trends in the prescription of novel oral anticoagulants in UK primary care . Br J Clin Pharmacol . 2017 ; 83 ( 9 ): 2096 - 106 . Epub 2017/04/09. https://doi.org/10.1111/ bcp.13299 PMID: 28390065; PubMed Central PMCID : PMC5555878 .
13. Desai NR , Krumme AA , Schneeweiss S , Shrank WH , Brill G , Pezalla EJ , et al. Patterns of initiation of oral anticoagulants in patients with atrial fibrillation- quality and cost implications . Am J Med . 2014 ; 127 ( 11 ): 1075 - 82 e1. Epub 2014/05/27. https://doi.org/10.1016/j.amjmed. 2014 . 05 .013 PMID: 24859719 .
14. Marzec LN , Wang J , Shah ND , Chan PS , Ting HH , Gosch KL , et al. Influence of Direct Oral Anticoagulants on Rates of Oral Anticoagulation for Atrial Fibrillation . J Am Coll Cardiol . 2017 ; 69 ( 20 ): 2475 - 84 . Epub 2017/05/20. https://doi.org/10.1016/j.jacc. 2017 . 03 .540 PMID: 28521884 .
15. Komen J , Forslund T , Hjemdahl P , Andersen M , Wettermark B. Effects of policy interventions on the introduction of novel oral anticoagulants in Stockholm: an interrupted time series analysis . Br J Clin Pharmacol . 2017 ; 83 ( 3 ): 642 - 52 . Epub 2016/10/13. https://doi.org/10.1111/bcp.13150 PMID: 27730670; PubMed Central PMCID : PMC5306490 .
16. Mochalina N , Joud A , Carlsson M , Sandberg MEC , Sjalander A , Juhlin T , et al. Antithrombotic therapy in patients with non-valvular atrial fibrillation in Southern Sweden: A population-based cohort study . Thromb Res . 2016 ; 140 : 94 - 9 . Epub 2016/03/05. https://doi.org/10.1016/j.thromres. 2016 . 02 .023 PMID: 26938155 .
17. Rahman F , Kwan GF , Benjamin EJ . Global epidemiology of atrial fibrillation . Nat Rev Cardiol . 2014 ; 11 ( 11 ): 639 - 54 . Epub 2014/08/13. https://doi.org/10.1038/nrcardio. 2014 .118 PMID: 25113750 .
18. Eek AK , Oie E , Granas AG . Prescribing of NOACs has outnumbered warfarin: exploring how physicians choose anticoagulant treatments . Eur J Clin Pharmacol . 2018 ; 74 ( 3 ): 323 - 30 . Epub 2017/11/18. https:// doi.org/10.1007/s00228-017-2374-4 PMID: 29149366 .
19. Sorensen HT , Lash TL , Rothman KJ . Beyond randomized controlled trials: a critical comparison of trials with nonrandomized studies . Hepatology . 2006 ; 44 ( 5 ): 1075 - 82 . Epub 2006/10/24. https://doi.org/10. 1002/hep.21404 PMID: 17058242 .
20. Schneeweiss S. Developments in post-marketing comparative effectiveness research . Clin Pharmacol Ther . 2007 ; 82 ( 2 ): 143 - 56 . Epub 2007/06/08. https://doi.org/10.1038/sj.clpt.6100249 PMID: 17554243; PubMed Central PMCID : PMC2905665 .
21. Ntaios G , Papavasileiou V , Makaritsis K , Vemmos K , Michel P , Lip GYH . Real-World Setting Comparison of Nonvitamin-K Antagonist Oral Anticoagulants Versus Vitamin-K Antagonists for Stroke Prevention in Atrial Fibrillation: A Systematic Review and Meta-Analysis . Stroke . 2017 ; 48 ( 9 ): 2494 - 503 . Epub 2017/07/19. https://doi.org/10.1161/STROKEAHA.117.017549 PMID: 28716982 .
22. Mitchell A , Watson MC , Welsh T , McGrogan A . Effectiveness and Safety of Direct Oral Anticoagulants versus Vitamin K Antagonists for People Aged 75 Years and over with Atrial Fibrillation: A Systematic Review and Meta-Analyses of Observational Studies . Journal of clinical medicine . 2019 ; 8 ( 4 ). Epub 2019 /04/27. https://doi.org/10.3390/jcm8040554 PMID: 31022899; PubMed Central PMCID : PMC6518135 .
23. Furu K. Establishment of the nationwide Norwegian Prescription Database (NorPD)-new opportunities for research in pharmacoepidemiology in Norway . Nor J Epidemiol . 2008 ; 18 ( 2 ): 129 - 36 .
24. Furu K , Wettermark B , Andersen M , Martikainen JE , Almarsdottir AB , Sorensen HT . The Nordic countries as a cohort for pharmacoepidemiological research . Basic Clin Pharmacol Toxicol . 2010 ; 106 ( 2 ): 86 - 94 . https://doi.org/10.1111/j.1742- 7843 . 2009 . 00494 . x PMID : 19961477 .
25. Halvorsen S , Ghanima W , Fride Tvete I , Hoxmark C , Falck P , Solli O , et al. A nationwide registry study to compare bleeding rates in patients with atrial fibrillation being prescribed oral anticoagulants . Eur Heart J Cardiovasc Pharmacother . 2017 ; 3 ( 1 ): 28 - 36 . Epub 2016/09/30. https://doi.org/10.1093/ehjcvp/ pvw031 PMID: 27680880; PubMed Central PMCID : PMC5216196 .
26. Schneeweiss S , Gagne JJ , Glynn RJ , Ruhl M , Rassen JA . Assessing the comparative effectiveness of newly marketed medications: methodological challenges and implications for drug development . Clin Pharmacol Ther . 2011 ; 90 ( 6 ): 777 - 90 . Epub 2011/11/04. https://doi.org/10.1038/clpt. 2011 .235 PMID: 22048230 .
27. Ray WA . Evaluating medication effects outside of clinical trials: new-user designs . Am J Epidemiol . 2003 ; 158 ( 9 ): 915 - 20 . Epub 2003/10/31. https://doi.org/10.1093/aje/kwg231 PMID: 14585769 .
28. Lund JL , Richardson DB , Sturmer T. The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application . Curr Epidemiol Rep . 2015 ; 2 ( 4 ): 221 - 8 . Epub 2016/03/10. https://doi.org/10.1007/s40471-015-0053-5 PMID: 26954351; PubMed Central PMCID : PMC4778958 .
29. Stovring H , Pottegard A , Hallas J . Refining estimates of prescription durations by using observed covariates in pharmacoepidemiological databases: an application of the reverse waiting time distribution . Pharmacoepidemiol Drug Saf . 2017 ; 26 ( 8 ): 900 - 8 . Epub 2017/05/04. https://doi.org/10.1002/pds.4216 PMID: 28466973 .
30. Stovring H , Pottegard A , Hallas J . Estimating medication stopping fraction and real-time prevalence of drug use in pharmaco-epidemiologic databases. An application of the reverse waiting time distribution . Pharmacoepidemiol Drug Saf . 2017 ; 26 ( 8 ): 909 - 16 . Epub 2017/05/06. https://doi.org/10.1002/pds.4217 PMID: 28474439 .
31. Thrane JM , Stovring H , Hellfritzsch M , Hallas J , Pottegard A . Empirical validation of the reverse parametric waiting time distribution and standard methods to estimate prescription durations for warfarin . Pharmacoepidemiol Drug Saf . 2018 ; 27 ( 9 ): 1011 - 8 . Epub 2018/06/29. https://doi.org/10.1002/pds. 4581 PMID: 29952049 .
32. Oie LR , Madsbu MA , Giannadakis C , Vorhaug A , Jensberg H , Salvesen O , et al. Validation of intracranial hemorrhage in the Norwegian Patient Registry . Brain Behav . 2018 ; 8 ( 2 ): e00900 . Epub 2018 /02/28. https://doi.org/10.1002/brb3.900 PMID: 29484261; PubMed Central PMCID : PMC5822577 .
33. McCormick N , Bhole V , Lacaille D , Avina-Zubieta JA . Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review . PLoS One . 2015 ; 10 ( 8 ): e0135834 . Epub 2015 /08/21. https://doi.org/10.1371/journal.pone.0135834 PMID: 26292280; PubMed Central PMCID : PMC4546158 .
34. Lipsitch M , Tchetgen Tchetgen E , Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies . Epidemiology . 2010 ; 21 ( 3 ): 383 - 8 . Epub 2010/03/26. https://doi.org/10.1097/ EDE.0b013e3181d61eeb PMID: 20335814; PubMed Central PMCID : PMC3053408 .
35. Torres A , Peetermans WE , Viegi G , Blasi F . Risk factors for community-acquired pneumonia in adults in Europe: a literature review . Thorax . 2013 ; 68 ( 11 ): 1057 - 65 . Epub 2013/10/17. https://doi.org/10. 1136/thoraxjnl-2013-204282 PMID: 24130229; PubMed Central PMCID : PMC3812874 .
36. Komen J , Forslund T , Hjemdahl P , Wettermark B . Factors associated with antithrombotic treatment decisions for stroke prevention in atrial fibrillation in the Stockholm region after the introduction of NOACs . Eur J Clin Pharmacol . 2017 ; 73 ( 10 ): 1315 - 22 . Epub 2017/07/01. https://doi.org/10.1007/ s00228-017-2289-0 PMID: 28664360; PubMed Central PMCID : PMC5612279 .
37. Gundlund A , Staerk L , Fosbol EL , Gadsboll K , Sindet-Pedersen C , Bonde AN , et al. Initiation of anticoagulation in atrial fibrillation: which factors are associated with choice of anticoagulant? J Intern Med . 2017 ; 282 ( 2 ): 164 - 74 . Epub 2017/05/10. https://doi.org/10.1111/joim.12628 PMID: 28480507 .
38. Forslund T , Wettermark B , Andersen M , Hjemdahl P . Stroke and bleeding with non-vitamin K antagonist oral anticoagulant or warfarin treatment in patients with non-valvular atrial fibrillation: a populationbased cohort study . Europace . 2017 . Epub 2017/02/09. https://doi.org/10.1093/europace/euw416 PMID: 28177459 .
39. Mueller S , Groth A , Spitzer SG , Schramm A , Pfaff A , Maywald U . Real-world effectiveness and safety of oral anticoagulation strategies in atrial fibrillation: a cohort study based on a German claims dataset . Pragmatic and observational research . 2018 ; 9 : 1 - 10 . https://doi.org/10.2147/POR.S156521 PMID: 29750067 .
40. Giugliano RP , Ruff CT , Braunwald E , Murphy SA , Wiviott SD , Halperin JL , et al. Edoxaban versus warfarin in patients with atrial fibrillation . N Engl J Med . 2013 ; 369 ( 22 ): 2093 - 104 . Epub 2013/11/21. https:// doi.org/10.1056/NEJMoa1310907 PMID: 24251359 .
41. Mackman N. The role of tissue factor and factor VIIa in hemostasis . Anesth Analg . 2009 ; 108 ( 5 ): 1447 - 52 . Epub 2009/04/18. https://doi.org/10.1213/ane.0b013e31819bceb1 PMID: 19372318; PubMed Central PMCID : PMC2838713 .
42. Gulati S , Solheim O , Carlsen SM , Oie LR , Jensberg H , Gulati AM , et al. Risk of intracranial hemorrhage (RICH) in users of oral antithrombotic drugs: Nationwide pharmacoepidemiological study . PLoS One . 2018 ; 13 ( 8 ): e0202575 . Epub 2018 /08/24. https://doi.org/10.1371/journal.pone.0202575 PMID: 30138389; PubMed Central PMCID : PMC6107180 .
43. Villines TC , Schnee J , Fraeman K , Siu K , Reynolds MW , Collins J , et al. A comparison of the safety and effectiveness of dabigatran and warfarin in non-valvular atrial fibrillation patients in a large healthcare system . Thromb Haemost . 2015 ; 114 ( 6 ): 1290 - 8 . Epub 2015/10/09. https://doi.org/10.1160/TH15-06- 0453 PMID: 26446456 .
44. Nishtala PS , Gnjidic D , Jamieson HA , Hanger HC , Kaluarachchi C , Hilmer SN . ' Real-world' haemorrhagic rates for warfarin and dabigatran using population-level data in New Zealand . Int J Cardiol . 2016 ; 203 : 746 - 52 . Epub 2015/11/23. https://doi.org/10.1016/j.ijcard. 2015 . 11 .067 PMID: 26590888 .
45. Lauffenburger JC , Farley JF , Gehi AK , Rhoney DH , Brookhart MA , Fang G . Effectiveness and safety of dabigatran and warfarin in real-world US patients with non-valvular atrial fibrillation: a retrospective cohort study . J Am Heart Assoc . 2015 ; 4 ( 4 ): e001798 . Epub 2015 /04/12. https://doi.org/10.1161/JAHA. 115.001798 PMID: 25862791; PubMed Central PMCID : PMC4579955 .
Wisloff T , Hagen G , Klemp M. Economic evaluation of warfarin, dabigatran, rivaroxaban, and apixaban for stroke prevention in atrial fibrillation . Pharmacoeconomics . 2014 ; 32 ( 6 ): 601 - 12 . Epub 2014/04/10.
https://doi.org/10.1007/s40273-014-0152-z PMID: 24715603; PubMed Central PMCID : PMC4031399 .
47. Forslund T , Komen JJ , Andersen M , Wettermark B , von Euler M , Mantel-Teeuwisse AK , et al. Improved Stroke Prevention in Atrial Fibrillation After the Introduction of Non-Vitamin K Antagonist Oral Anticoagulants . Stroke . 2018 ; 49 ( 9 ): 2122 - 8 . Epub 2018/10/26. https://doi.org/10.1161/STROKEAHA.118. 021990 PMID: 30354995 .
48. Shroff GR , Solid CA , Herzog CA. Temporal trends in ischemic stroke and anticoagulation therapy among Medicare patients with atrial fibrillation: a 15-year perspective (1992-2007) . JAMA Intern Med . 2013 ; 173 ( 2 ): 159 - 60 . Epub 2012/12/26. https://doi.org/10.1001/jamainternmed. 2013 .1579 PMID: 23266657 .
49. Piccini JP , Hellkamp AS , Lokhnygina Y , Patel MR , Harrell FE , Singer DE , et al. Relationship between time in therapeutic range and comparative treatment effect of rivaroxaban and warfarin: results from the ROCKET AF trial . J Am Heart Assoc . 2014 ; 3 ( 2 ): e000521 . Epub 2014 /04/24. https://doi.org/10.1161/ JAHA.113.000521 PMID: 24755148; PubMed Central PMCID : PMC4187517 .
Wallentin L , Lopes RD , Hanna M , Thomas L , Hellkamp A , Nepal S , et al. Efficacy and safety of apixaban compared with warfarin at different levels of predicted international normalized ratio control for stroke prevention in atrial fibrillation . Circulation . 2013 ; 127 ( 22 ): 2166 - 76 . Epub 2013/05/04. https://doi.org/10.
1161/CIRCULATIONAHA.112.142158 PMID: 23640971 .
Wallentin L , Yusuf S , Ezekowitz MD , Alings M , Flather M , Franzosi MG , et al. Efficacy and safety of dabigatran compared with warfarin at different levels of international normalised ratio control for stroke prevention in atrial fibrillation: an analysis of the RE-LY trial . Lancet . 2010 ; 376 ( 9745 ): 975 - 83 . https:// doi.org/10.1016/S0140- 6736 ( 10 ) 61194 - 4 WOS:000282247600029. PMID: 20801496
52. Tveit A , Abdelnoor M , Enger S , Smith P . Atrial fibrillation and antithrombotic therapy in a 75-year-old population . Cardiology . 2008 ; 109 ( 4 ): 258 - 62 . Epub 2007/09/18. https://doi.org/10.1159/000107789 PMID: 17873490 .
53. Malmo V , Langhammer A , Bonaa KH , Loennechen JP , Ellekjaer H . Validation of self-reported and hospital-diagnosed atrial fibrillation: the HUNT study . Clin Epidemiol . 2016 ; 8 : 185 - 93 . Epub 2016/06/30. https://doi.org/10.2147/CLEP.S103346 PMID: 27354826; PubMed Central PMCID : PMC4910677 .
54. Varmdal T , Bakken IJ , Janszky I , Wethal T , Ellekjaer H , Rohweder G , et al. Comparison of the validity of stroke diagnoses in a medical quality register and an administrative health register . Scand J Public Health . 2016 ; 44 ( 2 ): 143 - 9 . Epub 2015/12/15. https://doi.org/10.1177/1403494815621641 PMID: 26660300 .