Panels A and C depict the cumulative incidence curves for patients according to chronic kidney disease stages. Panels B and D stratify chronic kidney disease stages by warfarin exposure. Y-axis scale segments in blue indicate range from 0.0 to 0.14.
Panel A depicts the cumulative incidence curves for patients according to chronic kidney disease stages. Panel B stratifies chronic kidney disease stages by warfarin exposure.
eTable 1. Definition of comorbidities and undertaken procedures at inclusion
eTable 2. Definition of outcomes
eTable 3. Demographics, clinical and in-hospital course characteristics of 24317 myocardial infarction patients with atrial fibrillation stratified by CKD strata and warfarin treatment at discharge
eTable 4. Propensity score balance between warfarin-exposed and nonexposed individuals before/after matching in the whole patient population
eTable 5. Propensity score balance between warfarin-exposed and nonexposed individuals before/after matching in individuals with eGFR greater than 60 mL/min/1.73m2
eTable 6. Propensity score balance between warfarin-exposed and nonexposed individuals before/after matching in individuals with eGFR greater than 30 to 60 mL/min/1.73m2
eTable 7. Propensity score balance between warfarin-exposed and nonexposed individuals before/after matching in individuals with eGFR greater than 15 to 30 mL/min/1.73m2
eTable 8. Propensity score balance between warfarin-exposed and nonexposed individuals before/after matching in individuals with eGFR less than or equal to 15 mL/min/1.73m2
eTable 9. Performance of propensity score models in the whole dataset and for each CKD stratum
eTable 10. Risk of the composite outcome of death, myocardial infarction and ischemic stroke associated to warfarin treatment in subgroup populations
eTable 11. Subgroup population analysis of the composite outcome of death, myocardial infarction, ischemic stroke and bleeding
eFigure 1. Flowchart describing the inclusion of patients in the study
Customize your JAMA Network experience by selecting one or more topics from the list below.
Carrero JJ, Evans M, Szummer K, Spaak J, Lindhagen L, Edfors R, Stenvinkel P, Jacobson SH, Jernberg T. Warfarin, Kidney Dysfunction, and Outcomes Following Acute Myocardial Infarction in Patients With Atrial Fibrillation. JAMA. 2014;311(9):919–928. doi:10.1001/jama.2014.1334
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Conflicting evidence exists regarding the association between warfarin treatment, death, and ischemic stroke incidence in patients with advanced chronic kidney disease (CKD) and atrial fibrillation.
To study outcomes associated with warfarin treatment in relation to kidney function among patients with established cardiovascular disease and atrial fibrillation.
Design, Setting, and Participants
Observational, prospective, multicenter cohort study from the Swedish Web-System for Enhancement and Development of Evidence-Based Care in Heart Disease Evaluated According to Recommended Therapies (SWEDEHEART) registry (2003-2010), which includes all Swedish hospitals that provide care for acute cardiac diseases. Participants included consecutive survivors of an acute myocardial infarction (MI) with atrial fibrillation and known serum creatinine (N = 24 317), including 21.8% who were prescribed warfarin at discharge. Chronic kidney disease stages were classified according to estimated glomerular filtration rate (eGFR).
Main Outcomes and Measures
(1) Composite end point analysis of death, readmission due to MI, or ischemic stroke; (2) bleeding (composite of readmission due to hemorrhagic stroke, gastrointestinal bleeding, bleeding causing anemia, and others); or (3) the aggregate of these 2 outcomes within 1 year from discharge date.
A total of 5292 patients (21.8%) were treated with warfarin at discharge, and 51.7% had manifest CKD (eGFR <60 mL/min/1.73 m2 [eGFR<60]). Compared with no warfarin use, warfarin was associated with a lower risk of the first composite outcome (n = 9002 events) in each CKD stratum for event rates per 100 person-years: eGFR>60 event rate, 28.0 for warfarin vs 36.1 for no warfarin; adjusted hazard ratio (HR), 0.73 (95% CI, 0.65 to 0.81); eGFR>30-60: event rate, 48.5 for warfarin vs 63.8 for no warfarin; HR, 0.73 (95% CI, 0.66 to 0.80); eGFR>15-30: event rate, 84.3 for warfarin vs 110.1 for no warfarin; HR, 0.84 (95% CI, 0.70-1.02); eGFR≤15: event rate, 83.2 for warfarin vs 128.3 for no warfarin; HR, 0.57 (95% CI, 0.37-0.86). The risk of bleeding (n = 1202 events) was not significantly higher in patients treated with warfarin in any CKD stratum for event rates per 100 person-years: eGFR>60 event rate, 5.0 for warfarin vs 4.8 for no warfarin; HR, 1.10 (95% CI, 0.86-1.41); eGFR>30-60 event rate, 6.8 for warfarin vs 6.3 for no warfarin; HR, 1.04 (95% CI, 0.81-1.33); eGFR>15-30 event rate, 9.3 for warfarin vs 10.4 for no warfarin; HR, 0.82 (95% CI, 0.48-1.39); eGFR≤15 event rate, 9.1 for warfarin vs 13.5 for no warfarin; HR, 0.52 (95% CI, 0.16-1.65). Warfarin use in each CKD stratum was associated with lower hazards of the aggregate outcome (n = 9592 events) for event rates per 100 person-years: eGFR>60 event rate, 32.1 for warfarin vs 40.0 for no warfarin; HR, 0.76 (95% CI, 0.69-0.84); eGFR>30-60 event rate, 53.6 for warfarin vs 69.0 for no warfarin; HR, 0.75 (95% CI, 0.68-0.82); eGFR>15-30 event rate, 90.2 for warfarin vs 117.7 for no warfarin; HR, 0.82 (95% CI, 0.68-0.99); eGFR≤15 event rate, 86.2 for warfarin vs 138.2 for no warfarin; HR, 0.55 (95% CI, 0.37-0.83).
Conclusions and Relevance
Warfarin treatment was associated with a lower 1-year risk for the composite outcome of death, MI, and ischemic stroke without a higher risk of bleeding in consecutive acute MI patients with atrial fibrillation. This association was not related to the severity of concurrent CKD.
Patients with chronic kidney disease (CKD) experience a marked increased risk of atherothrombotic and thromboembolic complications, such as myocardial infarction and stroke, and simultaneously have an increased risk of bleeding.1,2 Both of these risks increase exponentially as renal function deteriorates. Embolism from the heart is a major cause of ischemic stroke, particularly in atrial fibrillation. Patients with concomitant atrial fibrillation and reduced kidney function are increasingly common, not only because both conditions share common risk factors, but also because they promote one another.3,4 Not unexpectedly, patients with CKD and atrial fibrillation have higher risk of ischemic stroke and thrombembolism than patients without CKD.5,6
Anticoagulation therapies, such as warfarin, are indicated in most patients with atrial fibrillation and supported by prospective, randomized trials demonstrating a marked efficacy. Although such trials have excluded individuals with renal dysfunction, the use of anticoagulation has also been extended to patients with CKD. However, in recent years, observational studies suggest that the use of warfarin may actually increase the risk of death and stroke among such patients.7- 9 On the other hand, other reports show no increased risk in patients with hemodialysis who are treated with warfarin and patients with CKD and atrial fibrillation.10- 12 This conflicting evidence allows no clear recommendation regarding the use of warfarin in patients with CKD, despite their high risk of stroke. Further, it is unknown whether there is a level of renal dysfunction at which the risk may outweigh the benefit. The objective of this study was to investigate outcomes associated with warfarin treatment in relation to kidney function among patients with established cardiovascular disease and atrial fibrillation (ie, a group of patients having a strong rationale for anticoagulation therapy according to the CHA2DS2-VASc score for atrial fibrillation stroke risk criteria).13,14
All patients were informed about their participation in the registry and the follow-up and had the right to refuse. The registry and the merging of registries were approved by The National Board of Health and Welfare and the ethics committee at Karolinska Institutet, Stockholm. We used data from 2003 to 2010 from the nationwide Swedish Web-System for Enhancement and Development of Evidence-Based Care in Heart Disease Evaluated According to Recommended Therapies (SWEDEHEART) registry. This registry includes all Swedish hospitals (N = 72) that provide care for acute cardiac diseases, and enrolls all consecutive patients admitted to a coronary care unit with symptoms suggestive of an acute coronary syndrome.15
Previous history of diabetes mellitus, hypertension, myocardial infarction, heart failure, peripheral vascular disease, atrial fibrillation, ischemic stroke, hemorrhagic stroke, other bleeding, chronic obstructive pulmonary disease, dementia, or cancer were obtained from the SWEDEHEART registry form and enriched with data linkages to the National Inpatient Registry, which includes diagnoses for all patients hospitalized in Sweden since 1987 to December 31, 2011 (eTable 1 in the Supplement). Information on patient presentation at admission, hospital course variables, and medication at admission and discharge were also collected. Atrial fibrillation was defined as a history of atrial fibrillation, electrocardiography showing signs of atrial fibrillation, or both, either on admission or during the in-hospital course.
Estimated glomerular filtration rate (eGFR) was estimated from serum creatinine values with the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation: mL/min/1.73 m2.16 The majority of creatinine assessments were performed by either enzymatic or corrected Jaffe method (alkaline picrate reaction), both of which are traceable to isotope dilution mass spectroscopy standards. For creatinine measurements performed with non–isotope dilution mass spectroscopy-traceable methods, values were reduced by 5% prior to being entered into the CKD-EPI formula.17,18 On the basis of current International Society of Nephrology’s Kidney Disease Improving Global Outcomes (KDIGO) recommendations,19 the following categorization for renal dysfunction was used: normal renal function and CKD stages 1 and 2, defined as eGFR higher than 60 mL/min/1.73 m2 (eGFR>60); moderate dysfunction (CKD stage 3), defined as eGFR higher than 30 to 60 mL/min/1.73 m2 (eGFR>30-60); severe dysfunction (CKD stage 4), defined as eGFR higher than 15 to 30 mL/min/1.73 m2 (eGFR>15-30); and end-stage renal disease (CKD stage 5), defined as eGFR less than or equal to 15 mL /min/1.73 m2 (eGFR≤15). In the absence of data on albuminuria, the category eGFR>60 includes individuals with both normal renal function and CKD stages 1 and 2.
The exposure was prescription of warfarin at discharge as registered in SWEDEHEART. No information on warfarin dosages was available. Mortality data were obtained by data linkages to the Swedish population registry, which included information about the vital status of all Swedish citizens through December 31, 2011. Hospitalizations during follow-up due to myocardial infarction, ischemic stroke, or bleeding were obtained from the National Inpatient Registry. Bleeding was defined as readmission due to hemorrhagic stroke, gastrointestinal bleeding, bleeding causing anemia, and others. Outcome definitions are further expanded in eTable 2 in the Supplement.
The following outcomes were defined a priori: (1) the composite outcome of death, readmission due to myocardial infarction and ischemic stroke within 1 year from discharge date, (2) readmission because of bleeding within 1 year from discharge date, and (3) the aggregate of these 2 outcomes. For these composite end points, the earliest (if any) event was selected. When analyzing separate end points, patients were censored for death. No other censoring was applied.
Data analysis was performed with R (http://www.r-project.org), version 2.15.0. Statistical significance was a P value less than .05 with 2-sided testing. Missing data were handled using multiple imputation by the method of chained equations.20 The R package’s Multivariate Imputation by Chained Equations (MICE) was used to form 20 imputed data sets.
Warfarin use vs no warfarin use was considered as a time-fixed binary variable throughout the follow-up period. Event rates for the study outcomes were calculated for each renal dysfunction category, reporting crude risk ratios. The association between warfarin and outcomes was studied by Cox proportional hazard models. Two sets of multivariable models were investigated to study the effect of potential confounders: (1) adjustment for age (5-knot restricted cubic spline), sex, eGFR, preexisting comorbidities (diabetes mellitus, hypertension, myocardial infarction, chronic heart failure, peripheral vascular disease, ischemic stroke, bleeding, chronic obstructive pulmonary disease, and cancer within 3 years), patient presentation characteristics at admission (ST segment elevation myocardial infarction [STEMI] and decompensated heart failure [Killip class ≥2]), hospital course (percutaneous coronary intervention and coronary artery bypass graft), discharge medication (antiplatelet therapy [none, mono, or dual]), β-blockers, angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, and statins), and center effect (as a random effect via a γ frailty distribution); and, in a separate model, (2) we further adjusted for left ventricular ejection fraction due to the large amount of imputed data for this variable. Inverse-variance–weighted linear regression was used to assess linear trends across eGFR categories (using equidistant CKD stages) on studied outcomes. Time to event (cumulative incidence) Kaplan-Meier curves were plotted across CKD stages with and without warfarin treatment.
Two sensitivity analyses were performed: to assess the effect of imputation, we (1) ran a complete-case analysis for the adjusted model (there were 23 056 such patients, corresponding to 94.8% of the total study population); and to assess the effect of confounding by indication, (2) propensity scores were calculated using random-effects logistic regression models with warfarin at discharge as the outcome and all variables of the primary analysis model as explanatory variables.21 Patients were then matched on estimated propensity scores using full matching22 via the R package’s Optmatch. A caliper of 0.01 was used, except for the patient category of eGFR≤15, where this led to many unmatched patients. For this stratum, a caliper of 0.02 was used instead. The analysis was then done as a Cox regression with warfarin at discharge as a predictor, and the matching indicator as a random effect using a γ frailty model.
In a final step, subgroup analyses were performed to assess the robustness of our results. Not all of these models could be run successfully due to (1) no warfarin contrast (no contrast in the exposure), (2) too few events (strata with <10 events), and (3) the model failed to converge. When this occurred it is indicated in the corresponding tables.
Between 2003 and 2010, a total of 158 059 individuals were admitted with an acute myocardial infarction. Atrial fibrillation was found in 34 087 individuals. Patients were excluded if they lacked information on serum creatinine (n = 2351), warfarin treatment (n = 357), or age (n = 2). They were also excluded if they died before hospital discharge (n = 2580), had been registered in SWEDEHEART previously (n = 4367) or because of registration errors (n = 113). A patient flowchart is illustrated in eFigure 1 in the Supplement.
The remaining 24 317 patients were included in the analysis and 21.8% of these were prescribed warfarin at discharge. Table 1 and Table 2 summarize baseline characteristics of included patients stratified by warfarin treatment. Patients treated with warfarin did not differ from patients receiving no warfarin with regard to basic demographics. However, several comorbidities, such as heart failure, diabetes, and ischemic stroke, were more common in individuals treated with warfarin. In both groups, the CHADS2 score was 2 or higher in the majority of patients. Clinical history of hemorrhagic stroke and bleeding seemed more frequent among patients who did not receive warfarin at discharge. There was a similar rate of patient inclusion during the different admission years.
As many as 51.7% of patients were considered to have CKD stage 3 or higher, with an eGFR higher than 60 mL/min/1.73 m2. Of these, 41.7% of patients had CKD stage 3, 8.1% had CKD stage 4, and 2.0% had CKD stage 5. eTable 3 in the Supplement shows patient characteristics according to CKD stages and warfarin treatment. As CKD categories increased in severity, patients tended to be older and less often smokers. The prevalence of comorbid diabetes, hypertension, cardiovascular disease (myocardial infarction, chronic heart failure, peripheral vascular disease, and ischemic stroke), as well as the number of patients with a CHADS2 score of 2 or higher increased with the worsening of CKD stages. Patients with CKD also presented more severe symptoms, such as heart failure (Killip class ≥2), at admission, but the number of patients presenting with STEMI and the proportion of percutaneous coronary interventions performed decreased. Dual antiplatelet therapy and statin treatment at discharge seemed less common across worsening CKD stages, but the proportion of patients receiving calcium antagonists or diuretics tended to increase. Patients with CKD stage 5 were less often prescribed warfarin (13.8%) compared with other categories (eGFR>60, 22.0%; eGFR>30-60, 22.4%; eGFR>15-30, 18.9%). In all CKD strata, patients receiving warfarin more often had a history of heart failure, ischemic stroke, and diabetes, as well as a higher CHADS2 score. History of hemorrhagic stroke or prior bleeding tended to be less common among patients treated with warfarin across the different CKD strata.
The number of patients who developed the composite outcome, bleeding events, and the aggregate of these 2 outcomes increased with the worsening of CKD categories (Table 3), as did the rate at which these events occurred (Table 3, Figure 1 [panels A and C], and Figure 2 [panel A]). Within each CKD category and compared with patients receiving no warfarin, those receiving warfarin had a lower number of events of the composite of death, myocardial infarction, or stroke (eGFR>60, 23.5% with warfarin vs 28.8% without warfarin; eGFR>30-60, 36.3% with warfarin vs 44.0% without warfarin; eGFR>15-30, 53.2% with warfarin vs 61.0% without warfarin; eGFR≤15, 53.0% with warfarin vs 66.0% without warfarin), as well as the number of aggregated events (death, myocardial infarction, stroke, and bleeding; eGFR>60, 26.4% with warfarin vs 31.1% without warfarin; eGFR>30-60, 39.1% with warfarin vs 46.4% without warfarin; eGFR>15-30, 55.1% with warfarin vs 63.0% without warfarin; eGFR≤15, 54.5% with warfarin vs 68.7% without warfarin), as well as a lower incidence rate (Table 3, Figure 1 [panel B], and Figure 2 [panel B]). The crude relative risk of bleeding was not significantly higher in patients treated with warfarin in any CKD stratum (eGFR>60: relative risk [RR], 1.08 [95% CI, 0.88 to 1.32]; eGFR>30-60: RR, 1.14 [95% CI, 0.94 to 1.38]; eGFR>15-30: RR, 1.00 [95% CI, 0.66 to 1.52]; eGFR≤15: RR, 0.73 [95% CI, 0.27 to 2.00], Figure 1, panel D). Regardless of CKD stage, the number of deaths was lower in patients treated with warfarin compared with patients receiving no warfarin. The rate of readmissions due to myocardial infarction and ischemic strokes were lower in the patients treated with warfarin across all renal function strata.
Table 4 shows the results from the Cox regression analyses. For each stratum of CKD, warfarin treatment was associated with a lower risk of the composite outcome of death, myocardial infarction, and stroke (eGFR>60: adjusted hazard ratio [HR], 0.73 [95% CI, 0.65 to 0.81]; eGFR>30-60: HR, 0.73 [95% CI, 0.66 to 0.80]; eGFR>15-30: HR, 0.84 [95% CI, 0.70 to 1.02]; eGFR≤15: HR, 0.57 [95% CI, 0.37 to 0.86]), as well as a lower risk of the aggregated composite outcome further considering bleeding (eGFR>60: HR, 0.76 [95% CI, 0.69 to 0.84]; eGFR>30-60: HR, 0.75 [95% CI, 0.68 to 0.82]; eGFR>15-30: HR, 0.82 [95% CI, 0.68 to 0.99]; eGFR≤15: HR, 0.55 [95% CI, 0.37 to 0.83]). For the patients treated with warfarin, the bleeding risk was not higher and there was no significant trend across differing CKD categories (eGFR>60: HR, 1.10 [95% CI, 0.86 to 1.41]; eGFR>30-60: HR, 1.04 [95% CI, 0.81 to 1.33]; eGFR>15-30: HR, 0.82 [95% CI, 0.48 to 1.39]; eGFR≤15: HR, 0.52 [95% CI, 0.16 to 1.65]). The magnitude of the association between warfarin treatment and outcomes was little modified by multivariable adjustment, and no linear trends were observed across strata. Similar results for the primary analysis were observed in complete-case analyses or propensity score–matching. Information on the performance of the propensity score–matching and the balance between variables in warfarin-exposed and nonexposed individuals is shown in eTable 4 through eTable 9 in the Supplement.
Several subgroup analyses were undertaken to study the robustness of our findings (eTable 10 and eTable 11 in the Supplement). Similar results for the primary analysis were found for the composite outcome of death, myocardial infarction, and ischemic stroke, and the same was true when further considering bleeding risk as an aggregate. The low number of bleeding events in the subgroup populations limits drawing solid conclusions and these analyses were not performed.
Recent observational studies have prompted concern about the value of warfarin anticoagulation in CKD stage 5 or dialysis patients with atrial fibrillation. In these studies, warfarin treatment was associated with an increased risk of stroke, death, or both,7- 9 resulting in modification of current guidelines, which call for caution regarding warfarin use in this patient group.23 Our study addressed this issue in a large cohort of consecutive patients with myocardial infarction, atrial fibrillation, and varying severity of CKD. We report that warfarin treatment was associated with a lower 1-year risk of the composite end point of death, myocardial infarction, and ischemic stroke without a higher risk of bleeding. This association was observed in patient strata with moderate, severe, or end-stage CKD. Our finding partially accords with the recent study by Olesen and coauthors12 from Denmark, where warfarin treatment was associated with a lower risk of stroke or thromboembolism in patients with atrial fibrillation irrespective of their renal function. Compared with that study, our analysis may offer some advantages, as we estimated renal function by eGFR rather than International Classification of Diseases codes for renal replacement therapy initiation. The use of International Classification of Diseases codes probably underestimates CKD prevalence24 and also offers no possibility to assess risks associated with impaired renal function prior to renal replacement therapy. In addition, we expand to a nationally representative sample of high-risk individuals that virtually all have an indication for warfarin therapy and have detailed information on concurrent anticoagulant and antiplatelet therapy. Our observations also accord with post hoc analyses from the Stroke Prevention in Atrial Fibrillation Study III (SPAF III) trial,11 in which 516 participants with atrial fibrillation and CKD stage 3 were studied and the risk for ischemic stroke or systemic embolism was reduced by 76% for adjusted-dose warfarin compared with aspirin plus low, subtherapeutic doses of warfarin. Finally, the higher stroke risk associated with warfarin use in dialysis patients reported by Wizemann and coauthors7 was only observed for individuals aged 75 or older. Because adverse effects of treatment are not open to bias, findings in our study may suggest not denying warfarin to patients with atrial fibrillation after a myocardial infarction because of compromised renal function. Similar to preceding evidence, however, our results are also observational in nature and do not provide conclusive guidance regarding anticoagulant therapy in patients with atrial fibrillation and CKD. Nevertheless, clear ethical concerns may not allow such trials.
Our study has several strengths that merit consideration, including its richness of information on patient clinical characteristics and confirmation of atrial fibrillation by both patient history and electrocardiographic findings during hospital course. Although no independent adjudication or verification of clinical events has been performed in our study, the validity of the Swedish Inpatient Register is considered high, in particular for diagnoses such as myocardial infarction, heart failure, atrial fibrillation, or stroke.25 The use of a unique personal identification number for all Swedish citizens and continuously updated national registries on death date, cause of death, and emigration allow a virtually complete follow-up with no losses. Furthermore, Swedish health care is tax-funded and access is homogenous throughout the country. Therefore the likelihood of confounding by different access to health care or treatment due to socioeconomic differences may not be relevant in this analysis.
Previous studies have reported an increased risk of bleeding in patients treated with warfarin with moderate to advanced CKD12 or requiring dialysis.7- 10,26,27 In our data, we observed no higher bleeding risk or nonstatistically significant hazards in relation to warfarin treatment and CKD strata. The discrepancy between our results and those observed in dialysis populations7- 9 may in part be attributed to differences in indications for warfarin treatment and to general differences in standard care between countries and centers.28 In this sense, patients with CKD stage 5 requiring or not requiring dialysis are reported to have a greater need of warfarin dose adjustment and poor time in therapeutic international normalized ratio (INR) range, with a tendency to supratherapeutic INR values.29,30 In Scandinavian countries, the care and time in therapeutic range for INR is generally very good and above 75%.28,31 Warfarin-associated bleeding in patients with CKD may be more common when INR is not tightly controlled. This was observed by Chan and coauthors,9 where the association with mortality in patients with hemodialysis was only observed for high warfarin-prescribed dosages. Also, the study by Knoll and coauthors32 from a single-center study with weekly and tight INR control showed that no patient with hemodialysis under sufficient warfarin anticoagulation had a stroke or a fatal bleeding event.
Some limitations need to be taken into consideration when interpreting our findings. The first limitation pertains to the inclusion criteria (patients with myocardial infarction and atrial fibrillation), which may make generalizability of the study results to other patient populations difficult. Although we are able to account for the most important confounders and confirm our results by propensity score–matching, residual confounding is still likely to exist and should be acknowledged as a limitation of this and any other observational study. For instance, we could not control for the duration of atrial fibrillation, deep venous thrombosis, pulmonary embolism, or valvular replacement. Besides the lack of information on the patients’ INRs, we acknowledge that the number of patients with CKD stage 5 included is low and merits caution in its interpretation, even more in subgroup analyses. Nevertheless, no linear trends were observed across the different CKD strata. Finally, we base our results on a single creatinine measurement, which may result in nondifferential misclassification. Certainly, there is concern that warfarin use in dialysis patients may accentuate vascular calcification33 or induce calcific uremic arteriopathy, a vasculopathy with a high mortality rate.34 Although our study is not designed to address this issue, death rates were not increased in patients with CKD stage 5 receiving warfarin in our analysis. It is plausible, nevertheless, that a 1-year follow-up is too short to capture the longer-term effects of warfarin on the vascular calcification process.
Warfarin treatment was associated with a lower 1-year risk of the composite outcome of death, myocardial infarction, and ischemic stroke, without a higher risk of bleeding, in consecutive patients with acute myocardial infarction and atrial fibrillation. This association was not related to the severity of concurrent CKD.
Corresponding Author: Juan Jesús Carrero, Karolinska Institutet, Divisions of Renal Medicine and Baxter Novum, Karolinska University Hospital Huddinge, K56, S-141 86 Stockholm, Sweden (email@example.com).
Author Contributions: Dr Jernberg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Carrero, Evans, Szummer, Spaak, Edfors, Stenvinkel, Jernberg.
Acquisition of data: Jacobson, Jernberg.
Analysis and interpretation of data: Carrero, Evans, Szummer, Spaak, Lindhagen, Jernberg.
Drafting of the manuscript: Carrero.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Lindhagen.
Obtained funding: Jacobson, Jernberg.
Administrative, technical, and material support: Jacobson.
Study supervision: Jernberg.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Carrero reports receiving grant funding from the Centre for Gender Medicine at Karolinska Institutet, the Swedish Medical Research Council, and the Westman Foundation. Dr Evans reports consulting for Amgen. Dr Szummer reports receiving grant funding from ALF Medicin (the regional agreement on medical training and clinical research between Stockholm County Council and Karolinska Institutet). Dr Spaak reports receiving grant funding from AbbVie; payment for lectures from Abbot and Merck (MSD); payment for the development of educational presentations from MSD; travel accommodations from Medtronic; and receiving support from the Karolinska Cardiorenal Theme Center. Dr Stenvinkel reports serving on the board for Abbott, Gambro, Keryx Biopharmaceuticals, Takeda, and Vifor Pharma; giving expert testimony for the Norwegian Research Council; receiving grant funding from Bayer, the Swedish Medical Research Council, and the Westman Foundation; and receiving payment for lectures from Amgen, Baxter, Bayer, Diaverum, Keryx, and Shire. Dr Jacobson reports receiving support from the Karolinska Cardiorenal Theme Center and payment for lectures from Amgen and Fresenius Medical Care. Dr Jernberg reports receiving grant funding from the Swedish Heart and Lung Foundation and ALF Medicin. No other disclosures were reported.
Funding/Support: This work was supported by a grant from the Swedish Foundation for Strategic Research.
Role of the Sponsor: The funder was not involved in the design and conduct of the study; collection, management, analysis, or interpretation of the data; and preparation, review, or approval of the manuscript, or decision to submit the manuscript for publication.