Table 3.

Observational studies on anticoagulants for stroke prevention in CKD stages 3–5 not on dialysis with atrial fibrillation

Study Characteristics and PublicationIntervention, Control, and Analysis TypeEffectiveness and Harm OutcomesaConclusion
Retrospective population analysis Denmark, 1997–2008. Follow-up not given, n=3587, CKD stages 3–5, not on dialysis, Olesen, NEJM 2012Warfarin±aspirin (n=609) versus nonwarfarin±aspirin (n=2978). As treated.Incidence rate of all-cause stroke or systemic embolism in 100 person yr 6.4. Risk with warfarin compared with no warfarin, HR, 0.84 (95% CI, 0.69 to 1.01). Incidence rate of major bleeding in 100 person yr 8.8b. Increased risk with warfarin compared with no warfarin, HR, 1.36 (95% CI, 1.17 to 1.59).Warfarin approaches statistical significance to reduce all-cause stroke or thromboembolism but also does associate with higher bleeding risk compared with nonwarfarin.
Retrospective population analysis Denmark, 1997–2011. Follow-up 0.85 yr, n=4519, CKD stages 3–5, not on dialysis, Bonde, JACC 2014Warfarin±aspirin (n=1130) versus no warfarin±aspirin (n=3389). As treated.“Net clinical benefit” defined as composite of fatal all-cause stroke/fatal bleeding. 753 fatal all-cause stroke/fatal bleed events. Warfarin associates with a lower risk of fatal stroke/fatal bleeding, HR, 0.71 (95% CI, 0.57 to 0.88)c.There is a net clinical benefit to warfarin over nonwarfarin in preventing deaths from stroke and bleeding.
Retrospective population analysis Alberta, Canada, 2003–2012. Follow-up 1 yr, n=2272, CKD stages 3–5, not on dialysis, age≥66 yr, Jun, AJKD 2017Warfarin±aspirin (n=1136) matched to no anti-coagulation±aspirin (n=1136). Intention to treat.Ischemic stroke/TIA: HR (95% CI) warfarin versus nonwarfarin eGFR 45–59: 0.60 (0.44 to 0.84); eGFR 30–44: 0.59 (0.38 to 0.94); eGFR<30: 0.54 (0.26 to 1.13). Major bleedingd: HR (95% CI) warfarin versus nonwarfarin eGFR 45–59: 1.00 (0.79 to 1.27); eGFR 30–44: 0.82 (0.62 to 1.09); eGFR<30: 0.95 (0.60 to 1.50).In older patients, warfarin associates with reduced risk of ischemic stroke/TIA compared with nonwarfarin with no significant difference in major bleeding.
Retrospective population analysis Ontario, Canada. Follow-up 0.73 yr, n=1417, CKD stages 3–5, not on dialysis on anti-coagulation, age≥66 yr, Keskar, KI 2017Anti-coagulation±aspirin (n=1417) matched to no anti-coagulation±aspirin (n=1417). Anti-coagulation group includes: warfarin 91%, heparin 3%, DOACs 6%. As treated.Incidence rate per 1000 person yr of ischemic stroke 41.3 on anti-coagulation and 34.4 not on anti-coagulation. Risk on anti-coagulation compared with no anti-coagulation, HR, 1.12 (95% CI, 0.90 to 1.39). Incidence rate per 1000 person yr of major bleedinge 61.3 on and 34.3 not on anti-coagulation. Risk on anti-coagulation compared with no anti-coagulation, HR, 1.60 (95% CI, 1.31 to 1.97).Anti-coagulation associates with increased risk of bleeding and no reduction in stroke compared with no anti-coagulation in older adults.
Retrospective population analysis England and Wales, 2006–2016. Follow-up 1.39 yr, n=6977, CKD stages 3–5, not on dialysis, age>65 yr, Kumar, BJM 2018Anti-coagulation±anti-PLTs (n=2424) versus no anti-coagulation±aspirin (n=2424). Anti-coagulation group includes: warfarin 72%, rivaroxaban 13%, apixaban 11%, dabigatran 3%, edoxaban 0.2%, heparin 2%. As treated.Incidence rates per 100 person yr for ischemic stroke were 4.6 on anti-coagulation and 1.5 not on anti-coagulation. Risk in anti-coagulation versus no anti-coagulation, HR, 2.6 (95% CI, 2.0 to 3.4). Incidence rates per 100 person yr for hemorrhagic stroke and GIB were 1.2 on anti-coagultion and 0.4 not on anti-coagulation. Risk on anti-coagulation versus not on anti-coagulation, HR, 2.4 (95% CI, 1.4 to 4.1)Anti-coagulation associates with increased risk of ischemic strokes and bleeding compared with no anti-coagulation in older adults.
Retrospective analysis in United States, 2010–2017. Follow-up not givenf, n=1990, with CKD stages 3–5, Jung-Im Shin, CJASN 2018DOACs (n=1990) apixaban 32%, rivaroxaban 41%, dabigatran 27%, matched to warfarin (n=1990). As treated.Ischemic stroke event rates per 100 person yr 8.8 in DOACs versus 7.8 in warfarin users, HR, 1.02 (0.76 to 1.37). Major bleedinge event rates per 100 person yr 26.3 bleeding DOACs versus 20.1 in warfarin users, HR, 1.23 (95% CI, 1.02 to 1.48)DOACs associated with similar benefits from prevention of ischemic stroke but slightly higher risk of bleeding versus warfarin.
  • As-treated analysis refers to time on drug as often done with statistical techniques using time-varying exposures (e.g., time-dependent Cox proportional hazards models) or censoring if there is no anticoagulant prescription refill within a defined period of time. Only the primary analysis is reported. Intention-to-treat analyses use exposure (e.g., warfarin versus no anticoagulation) as a time-fixed binary variable. NEJM, New England Journal of Medicine; HR, hazard ratio; 95% CI, 95% confidence interval; JACC, Journal of the American College of Cardiology; AJKD, American Journal of Kidney Disease; TIA, transient ischemic attack; KI, Kidney International; DOACs, direct-acting oral anticoagulants; BJM, British Journal of Medicine; anti-PLTs, antiplatelets (includes aspirin and clopidogrel); GIB, gastrointestinal bleed; CJASN, Clinical Journal of the American Society of Nephrology.

  • a Adjusted analyses are reported.

  • b Hospitalization or death from all-cause stroke (ischemic and or hemorrhagic) or systemic thromboembolism (peripheral-artery embolism, ischemic stroke, and TIA). Hospitalization or death from major bleeding (gastrointestinal, intracranial, urinary tract, or airway). Outcome events were not separated by treatment group.

  • c This study was designed to evaluate net clinical benefit; outcomes of stroke/systemic thromboembolism/all bleeding and fatal stroke/fatal bleeding were not reported separately. The numbers of outcome events were not separated by treatment group.

  • d Major bleeding defined by first hospitalization or emergency department visit for intracranial, upper or lower gastrointestinal, or other bleeding.

  • e Bleeding defined as a code for major (i.e., bleeding at critical site [intracranial, retroperitoneal, intraspinal, intraocular, pericardial, or intraarticular] or bleeding requiring transfusion) and minor. Bleeding with a corresponding trauma code was excluded.

  • f Major bleeding defined as fatal, intracranial, ocular causing blindness, articular, or retroperitoneal requiring intervention or transfusion ≥2 U of packed red cells, or hemoglobin drop ≥2 g/dl. Results from intention-to-treat and time-varying analyses were similar. Four percent (n=253) had eGFR<30 ml/min and dialysis patients were included. A composite of all minor and major bleeding was used.