Atrial Fibrillation
Conditions
Brief summary
The primary purpose of this study is to evaluate comparative effectiveness and safety outcomes of therapies to prevent thromboembolic events in patients with nonvalvular atrial fibrillation by using Korean nationwide health claims database.
Interventions
DRUGApixaban
Treatment for NVAF patients
DRUGDabigatran
Treatment for NVAF patients
DRUGRivaroxaban
Treatment for NVAF patients
DRUGwarfarin
Treatment for NVAF patients
DRUGAntiplatelets
Treatment for NVAF patients
Sponsors
Pfizer
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
1. Patients aged 18 years or older on the index date 2. Patients had ≥1 medical claim for AF (refer to Table 1) before or on the index date with at least one hospitalization or at least two outpatient visits: 3. Patients prescribed aspirin, warfarin, or NOACs during intake period (from July 1, 2015 to November 30, 2016)
Exclusion criteria
Patients meeting any of the following criteria will not be included in the study. 1. Medical claims indicating diagnosis or procedure for hip/knee replacement surgery within 6 weeks prior to index date 2. Medical claims indicating a diagnosis code indicative of rheumatic mitral valvular heart disease, mitral valve stenosis during the 12-month baseline period (Valvular AF / Prosthetic heart valves) 3. Medical claims indicating a diagnosis code of VTE (Venous thromboembolism) during the 12-month baseline period 4. Medical claims indicating a diagnosis or procedure code of transient AF, or cardiac surgery during the 12-month baseline period (Thyrotoxicosis, Hypertrophic cardiomyopathy, Elective defibrillation, radiofrequency ablation, or left atrial appendage occlusion) 5. Medical claims indicating a diagnosis code of other conditions during the 12-month baseline period (End-stage chronic kidney disease / Kidney transplant / Dialysis / Pericarditis) 6. For the comparison of NOAC versus NOAC, and NOAC versus warfarin, patients with any OACs (apixaban, dabigatran, rivaroxaban, or warfarin) in the pre-index period (from 1 year prior to the day before index date) 7. For the comparison of NOAC versus aspirin, patients with following medications in the pre-index period (from 1 year prior to the day before index date) * NOAC user: OACs (apixaban, dabigatran, rivaroxaban, warfarin) * Aspirin user: none
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years. Hemorrhagic stroke, ischemic stroke and systemic embolism requiring hospitalization identified using hospital claims which had hemorrhagic, ischemic stroke or systemic embolism Korean standard classification of diseases (KCD) code, whichever came first (first occurred event used). KCD code: hemorrhagic stroke = I60-62, I690-692; ischemic stroke = G459, I63, I693; systemic embolism = I74. Hospitalization and brain CT/MRI codes were used for ischemic stroke, hemorrhagic stroke.Hospitalization and any CT/MRI codes were used for systemic embolism. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years. Hemorrhagic stroke, ischemic stroke and systemic embolism requiring hospitalization identified using hospital claims which had hemorrhagic, ischemic stroke or systemic embolism Korean standard classification of diseases (KCD) code, whichever came first (first occurred event used). KCD code: hemorrhagic stroke = I60-62, I690-692; ischemic stroke = G459, I63, I693; systemic embolism = I74. Hospitalization and brain CT/MRI codes were used for ischemic stroke, hemorrhagic stroke.Hospitalization and any CT/MRI codes were used for systemic embolism. Index date = the first prescription date of study drugs during intake duration. |
| Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate: number of events divided by 100 participant-years. Intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding and other bleeding requiring hospitalization identified using hospital claims which had ICH, GI and other bleeding KCD code whichever came first (first occurred event used). KCD code: ICH = I60-62, I690-92, S064-66, S068; GI bleeding = I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922; other bleeding = D62,H448,H3572,H356,H313,H210,H113,H052,H470,H431,I312,N020-N029,N421,N831,N857,N920,N923,N930,N938-939,M250,R233,R040-042,R048-049,T792,T810,N950,R310, R311, R318, R58, T455, Y442, D683). Brain CT/MRI codes were used for ICH only. Index date= first prescription date of study drugs during intake duration. Participants were identified as NOAC user/Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years. Intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding and other bleeding requiring hospitalization identified using hospital claims which had ICH, GI and other bleeding KCD code whichever came first (first occurred event used). KCD code: ICH = I60-62, I690-92, S064-66, S068; GI bleeding = I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922; other bleeding = D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938-939, M250, R233, R040-042, R048-049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Brain CT/MRI codes were used for ICH only. Index date = the first prescription date of study drugs during intake duration. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of systemic embolism. Systemic embolism requiring hospitalization was identified using hospital claims which had systemic embolism KCD code = I74. For systemic embolism, hospitalization and any CT or MRI codes was used. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration |
| Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of systemic embolism. Systemic embolism requiring hospitalization was identified using hospital claims which had systemic embolism KCD code = I74. For systemic embolism, hospitalization and any CT or MRI codes was used. Index date= the first prescription date of study drugs during intake duration. |
| Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of GI bleeding events after index date was reported. GI bleeding requiring hospitalization was identified using hospital claims which had a GI bleeding KCD code (I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922). Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of GI bleeding events after index date was reported. GI bleeding requiring hospitalization was identified using hospital claims which had a GI bleeding KCD code (I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922). Index date = the first prescription date of study drugs during intake duration. |
| Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was number of events divided by 100 participant-years for first occurrence of hemorrhagic stroke events after index date was reported. Hemorrhagic stroke requiring hospitalization was identified using hospital claims which had a hemorrhagic stroke KCD code (I60-62, I690-692). For hemorrhagic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of intracranial hemorrhage events after index date was reported. Intracranial hemorrhage requiring hospitalization was identified using hospital claims which had an intracranial hemorrhage KCD code (I60, I61, I62, I690, I691, I692, S064, S065, S066, and S068) and brain CT or MRI codes. Index date = the first prescription date of study drugs during intake duration. |
| Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of other bleeding events after index date was reported. Other bleeding requiring hospitalization was identified using hospital claims which had other bleeding KCD code (D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938, N939, M250, R233, R040, R041, R042, R048, R049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of other bleeding events after index date was reported. Other bleeding requiring hospitalization was identified using hospital claims which had other bleeding KCD code (D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938, N939, M250, R233, R040, R041, R042, R048, R049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Index date = the first prescription date of study drugs during intake duration. |
| Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of intracranial hemorrhage events after index date was reported. Intracranial hemorrhage requiring hospitalization was identified using hospital claims which had an intracranial hemorrhage KCD code (I60, I61, I62, I690, I691, I692, S064, S065, S066, and S068) and brain CT or MRI codes. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of hemorrhagic stroke events after index date was reported. Hemorrhagic stroke requiring hospitalization was identified using hospital claims which had a hemorrhagic stroke KCD code (I60-62, I690-692). For hemorrhagic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. |
| Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of ischemic stroke events after index date was reported. Ischemic stroke requiring hospitalization was identified using hospital claims which had ischemic stroke KCD code (G459, I63, and I693). For ischemic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration. |
| Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016) | Event rate was defined as number of events divided by 100 participant-years for first occurrence of ischemic stroke events after index date was reported. Ischemic stroke requiring hospitalization was identified using hospital claims which had ischemic stroke KCD code (G459, I63, and I693). For ischemic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. |
Countries
South Korea
Participant flow
Recruitment details
Data was retrieved from Korean Health Insurance Review & Assessment Service (HIRA) database, of participants who were diagnosed with non-valvular atrial fibrillation (NVAF) from 1-January-2007 up to and including index date, who newly initiated non-vitamin K antagonist oral anticoagulants (NOACs), warfarin or used aspirin during intake duration.
Pre-assignment details
It is a retrospective, observational study. Index date: first prescription date of study drugs during intake duration. Intake period: between 1-July-2015 to 30-November-2016. Inverse probability of treatment weighted (IPTW) method was used to analyze outcome measures to balance participant's characteristics among reporting arms.
Participants by arm
| Arm | Count |
|---|---|
| NOAC - Apixaban Oral anticoagulants (OACs) treatment-naive participants diagnosed with NVAF, who initiated apixaban on the index date, were included in this study cohort and their data available in Korean HIRA database was retrospectively observed. Index date for apixaban was defined as the first prescription date of apixaban during the intake period from 1-July-2015 to 30-November-2016. | 10,548 |
| NOAC - Dabigatran OACs treatment-naive participants diagnosed with NVAF, who initiated dabigatran on the index date, were included in this study cohort and their data available in Korean HIRA database was retrospectively observed. Index date for dabigatran was defined as the first prescription date of dabigatran during the intake period from 1-July-2015 to 30-November-2016. | 11,414 |
| NOAC - Rivaroxaban OACs treatment-naive participants diagnosed with NVAF, who initiated rivaroxaban on the index date, were included in this study cohort and their data available in Korean HIRA database was retrospectively observed. Index date for rivaroxaban was defined as the first prescription date of rivaroxaban during the intake period from 1-July-2015 to 30-November-2016. | 17,779 |
| Warfarin OACs treatment-naive participants diagnosed with NVAF, who initiated warfarin on the index date, were included in this study cohort and their data available in Korean HIRA database was retrospectively observed. Index date for warfarin was defined as the first prescription date of warfarin during the intake period from 1-July-2015 to 30-November-2016. | 8,648 |
| Aspirin OACs treatment-naive participants diagnosed with NVAF, who initiated aspirin on the index date, were included in this study cohort and their data available in Korean HIRA database was retrospectively observed. Index date for aspirin was defined as the first prescription date of aspirin during the intake period from 1-July-2015 to 30-November-2016. | 16,295 |
| Total | 64,684 |
Baseline characteristics
| Characteristic | NOAC - Apixaban | NOAC - Dabigatran | NOAC - Rivaroxaban | Warfarin | Aspirin | Total |
|---|---|---|---|---|---|---|
| Age, Continuous | 73.89 years STANDARD_DEVIATION 9.52 | 72.23 years STANDARD_DEVIATION 9.52 | 73.27 years STANDARD_DEVIATION 9.55 | 68.73 years STANDARD_DEVIATION 12.62 | 72.10 years STANDARD_DEVIATION 9.17 | 72.29 years STANDARD_DEVIATION 10.04 |
| Race and Ethnicity Not Collected | — | — | — | — | — | 0 Participants |
| Sex: Female, Male Female | 5148 Participants | 4926 Participants | 8169 Participants | 3238 Participants | 6995 Participants | 28476 Participants |
| Sex: Female, Male Male | 5400 Participants | 6488 Participants | 9610 Participants | 5410 Participants | 9300 Participants | 36208 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk | EG002 affected / at risk | EG003 affected / at risk | EG004 affected / at risk |
|---|---|---|---|---|---|
| deaths Total, all-cause mortality | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 |
| other Total, other adverse events | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 |
| serious Total, serious adverse events | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 |
Outcome results
Primary
Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years. Intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding and other bleeding requiring hospitalization identified using hospital claims which had ICH, GI and other bleeding KCD code whichever came first (first occurred event used). KCD code: ICH = I60-62, I690-92, S064-66, S068; GI bleeding = I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922; other bleeding = D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938-939, M250, R233, R040-042, R048-049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Brain CT/MRI codes were used for ICH only. Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.70 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 8.65 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.77 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 9.86 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 8.18 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 9.36 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.761, 0.985]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.684, 0.862]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.786, 0.975]
Primary
Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate: number of events divided by 100 participant-years. Intracranial hemorrhage (ICH), gastrointestinal (GI) bleeding and other bleeding requiring hospitalization identified using hospital claims which had ICH, GI and other bleeding KCD code whichever came first (first occurred event used). KCD code: ICH = I60-62, I690-92, S064-66, S068; GI bleeding = I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922; other bleeding = D62,H448,H3572,H356,H313,H210,H113,H052,H470,H431,I312,N020-N029,N421,N831,N857,N920,N923,N930,N938-939,M250,R233,R040-042,R048-049,T792,T810,N950,R310, R311, R318, R58, T455, Y442, D683). Brain CT/MRI codes were used for ICH only. Index date= first prescription date of study drugs during intake duration. Participants were identified as NOAC user/Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 7.97 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 14.73 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 8.57 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 13.77 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 9.55 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Major Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 14.23 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.512, 0.664]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.597, 0.953]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.685, 1.04]
Primary
Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years. Hemorrhagic stroke, ischemic stroke and systemic embolism requiring hospitalization identified using hospital claims which had hemorrhagic, ischemic stroke or systemic embolism Korean standard classification of diseases (KCD) code, whichever came first (first occurred event used). KCD code: hemorrhagic stroke = I60-62, I690-692; ischemic stroke = G459, I63, I693; systemic embolism = I74. Hospitalization and brain CT/MRI codes were used for ischemic stroke, hemorrhagic stroke.Hospitalization and any CT/MRI codes were used for systemic embolism. Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.75 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.47 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.35 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.48 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.01 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 7.31 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.87, 1.134]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.839, 1.073]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.854, 1.082]
Primary
Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years. Hemorrhagic stroke, ischemic stroke and systemic embolism requiring hospitalization identified using hospital claims which had hemorrhagic, ischemic stroke or systemic embolism Korean standard classification of diseases (KCD) code, whichever came first (first occurred event used). KCD code: hemorrhagic stroke = I60-62, I690-692; ischemic stroke = G459, I63, I693; systemic embolism = I74. Hospitalization and brain CT/MRI codes were used for ischemic stroke, hemorrhagic stroke.Hospitalization and any CT/MRI codes were used for systemic embolism. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 7.66 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 13.52 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 7.2 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 13.53 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 7.18 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Stroke/Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 12.89 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of hazard ratio (HR) at year 1 as an extended Cox model was used.95% CI: [0.541, 0.707]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.53, 0.687]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.563, 0.884]
Secondary
Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of GI bleeding events after index date was reported. GI bleeding requiring hospitalization was identified using hospital claims which had a GI bleeding KCD code (I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922). Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.23 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.18 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.28 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.59 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.97 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.25 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.62, 0.91]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.586, 0.83]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.801, 1.094]
Secondary
Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of GI bleeding events after index date was reported. GI bleeding requiring hospitalization was identified using hospital claims which had a GI bleeding KCD code (I850, I983, K2211, K226, K228, K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K3181, K5521, K625, K920, K921, K922). Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 3.44 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.2 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 4.17 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 5.56 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 4.32 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Gastrointestinal (GI) Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 5.78 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.492, 0.731]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.719, 1.522]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.92, 1.824]
Secondary
Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of hemorrhagic stroke events after index date was reported. Hemorrhagic stroke requiring hospitalization was identified using hospital claims which had a hemorrhagic stroke KCD code (I60-62, I690-692). For hemorrhagic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.11 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.74 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.05 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.15 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.71 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.14 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.998, 2.145]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.647, 1.225]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.447, 0.888]
Secondary
Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was number of events divided by 100 participant-years for first occurrence of hemorrhagic stroke events after index date was reported. Hemorrhagic stroke requiring hospitalization was identified using hospital claims which had a hemorrhagic stroke KCD code (I60-62, I690-692). For hemorrhagic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.1 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.73 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.78 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.71 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.18 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Hemorrhagic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.72 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.47, 0.964]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.337, 0.715]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.548, 1.018]
Secondary
Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of intracranial hemorrhage events after index date was reported. Intracranial hemorrhage requiring hospitalization was identified using hospital claims which had an intracranial hemorrhage KCD code (I60, I61, I62, I690, I691, I692, S064, S065, S066, and S068) and brain CT or MRI codes. Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.39 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.12 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.35 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.43 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.08 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus NOAC Analysis | 1.41 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.871, 1.666]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.691, 1.218]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.578, 1.027]
Secondary
Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of intracranial hemorrhage events after index date was reported. Intracranial hemorrhage requiring hospitalization was identified using hospital claims which had an intracranial hemorrhage KCD code (I60, I61, I62, I690, I691, I692, S064, S065, S066, and S068) and brain CT or MRI codes. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.37 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 2.38 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.17 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 2.31 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 1.47 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Intracranial Hemorrhage Requiring Hospitalization: NOAC Versus Warfarin Analysis | 2.37 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.212, 0.658]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.392, 0.737]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.507, 0.87]
Secondary
Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of ischemic stroke events after index date was reported. Ischemic stroke requiring hospitalization was identified using hospital claims which had ischemic stroke KCD code (G459, I63, and I693). For ischemic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.99 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.84 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.60 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.58 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.37 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus NOAC Analysis | 6.43 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.85, 1.122]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.848, 1.1]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.877, 1.125]
Secondary
Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of ischemic stroke events after index date was reported. Ischemic stroke requiring hospitalization was identified using hospital claims which had ischemic stroke KCD code (G459, I63, and I693). For ischemic stroke, hospitalization and brain CT or MRI codes were also required. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.9 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 11.8 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.53 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 11.92 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.29 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Ischemic Stroke Requiring Hospitalization: NOAC Versus Warfarin Analysis | 11.19 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.556, 0.738]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.544, 0.715]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.588, 0.954]
Secondary
Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of other bleeding events after index date was reported. Other bleeding requiring hospitalization was identified using hospital claims which had other bleeding KCD code (D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938, N939, M250, R233, R040, R041, R042, R048, R049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Index date = the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.67 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.06 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.73 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.54 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 3.83 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus NOAC Analysis | 4.35 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.732, 1.061]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.678, 0.947]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.754, 1.032]
Secondary
Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of other bleeding events after index date was reported. Other bleeding requiring hospitalization was identified using hospital claims which had other bleeding KCD code (D62, H448, H3572, H356, H313, H210, H113, H052, H470, H431, I312, N020-N029, N421, N831, N857, N920, N923, N930, N938, N939, M250, R233, R040, R041, R042, R048, R049, T792, T810, N950, R310, R311, R318, R58, T455, Y442, D683). Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 3.75 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.93 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 3.86 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.61 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 4.42 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Other Bleeding Requiring Hospitalization: NOAC Versus Warfarin Analysis | 6.85 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.481, 0.701]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.534, 0.766]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.622, 1.13]
Secondary
Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of systemic embolism. Systemic embolism requiring hospitalization was identified using hospital claims which had systemic embolism KCD code = I74. For systemic embolism, hospitalization and any CT or MRI codes was used. Index date= the first prescription date of study drugs during intake duration.
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.20 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.25 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.20 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.28 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.24 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus NOAC Analysis | 0.26 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.343, 1.615]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.342, 1.402]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.498, 1.747]
Secondary
Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis
Event rate was defined as number of events divided by 100 participant-years for first occurrence of systemic embolism. Systemic embolism requiring hospitalization was identified using hospital claims which had systemic embolism KCD code = I74. For systemic embolism, hospitalization and any CT or MRI codes was used. Index date = the first prescription date of study drugs during intake duration. Participants were identified as NOAC user or Warfarin user depending on the date when they first used NOAC or Warfarin during intake duration
Time frame: Maximum of 1 year 4 months (From 1-July-2015 to 30-November-2016)
Population: OACs treatment-naive participants with NVAF, starting any OACs in intake duration; matched on propensity scores by IPTW method to balance participant characteristics among specified arms. It created virtual groups used in analysis here. No data for Aspirin arm: could not be matched using IPTW with other arms due to heterogeneity characteristics.
| Arm | Measure | Value (NUMBER) |
|---|---|---|
| Apixaban vs Warfarin (Apixaban) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.18 events per 100 participants-years |
| Apixaban vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.78 events per 100 participants-years |
| Dabigatran vs Warfarin (Dabigatran) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.24 events per 100 participants-years |
| Dabigatran vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.75 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Rivaroxaban) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.27 events per 100 participants-years |
| Rivaroxaban vs Warfarin (Warfarin) | Event Rate of Systemic Embolism Requiring Hospitalization: NOAC Versus Warfarin Analysis | 0.74 events per 100 participants-years |
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.119, 0.523]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.203, 0.711]
Comparison: Cox proportional hazards model was used to compare event rates between the treatment groups. To check the proportional hazards assumption, the graphical methods and statistical tests using the time-dependent covariate were used. If there were evidence of violation in the proportional hazards assumption in graphical methods, the statistical tests were used and if the evidence of assumption violation still existed, the value of HR at year 1 as an extended Cox model was used.95% CI: [0.247, 0.722]