Preadmission Oral Anticoagulant Treatment and Clinical Outcome Among Patients Hospitalized With Acute Stroke and Atrial Fibrillation
A Nationwide Study
Background and Purpose—Preadmission oral anticoagulant treatment (OAT) has been linked with less severe stroke and a better outcome in patients with atrial fibrillation. However, the existing studies have methodological limitations and have, with one exception, not included hemorrhagic strokes. We performed a nationwide historic follow-up study using data from population-based healthcare registries to assess the effect of preadmission OAT on stroke outcomes further.
Methods—We identified 11 356 patients with atrial fibrillation admitted to hospital with acute stroke (including ischemic stroke and intracerebral hemorrhage) between 2003 and 2009. Propensity score–matched analyses were used to compare stroke severity (Scandinavian Stroke Scale score) and mortality among 2175 patients with preadmission OAT and 2175 patients without preadmission OAT.
Results—A total of 2492 (21.9%) patients received OAT at the time of their stroke. Preadmission OAT was associated with a lower risk of severe stroke (Scandinavian Stroke Scale score at time of admission, <30 point; propensity score–matched odds ratio, 0.74; 95% confidence interval, 0.63–0.86) and lower 30-day mortality rate (propensity score–matched adjusted odds ratio, 0.83; 95% confidence interval, 0.71–0.98).
Conclusions—Only a minority of hospitalized patients with acute stroke with atrial fibrillation received OAT at the time of stroke. Preadmission OAT was associated with less severe stroke and lower 30-day mortality rate in a propensity score–matched analysis.
Atrial fibrillation (AF) is the most common cardiac arrhythmia and is a major risk factor for ischemic stroke,1,2 particularly among elderly patients.3 Stroke is the most feared complication in patients with AF because AF is associated with a higher risk of adverse outcome after stroke.4,5
Several clinical trials have documented the efficacy of oral anticoagulation treatment (OAT) for prevention of cardioembolic stroke in patients with AF.6 Less is known about the possible effect of preadmission OAT on stroke outcomes. Only a few observational studies have examined the association, and overall preadmission OAT has been found to be associated with less severe stroke, lower case fatality, and improved functional level.7–13 If this association is truly causal, it constitutes a strong argument for increasing the efforts to ensure that eligible patients with AF receive OAT according to clinical guideline recommendations. This is important because several studies have found that high proportions of eligible patients with AF, including high-risk patients, are withheld OAT.14
However, several uncertainties related to limitations of the existing studies remain to be clarified before a more firm conclusion can be made. Most of the existing studies have not been population based, and none are nationwide, which raises concerns about the generalizability of the findings. Furthermore, the ability to control adequately for possible confounding factors and the statistical precision could be questioned. Finally, all except for one small study11 have been restricted to patients with ischemic stroke although the possible beneficial effect of OAT on ischemic events could potentially be outweighed by a potential harm in patients with intracerebral hemorrhage (ICH). There is consequently a need for additional large-scale population-based studies. Therefore, we examined the preadmission use of OAT among hospitalized Danish patients with stroke with AF and the association with stroke severity and case fatality in a nationwide, population-based follow-up study.
The study was based on national Danish registries covering the entire population (≈5.5 million). The Danish National Health Service provides tax-financed healthcare to all residents, and unambiguous individual-level linkage between registries is enabled by a unique 10-digit civil registration number that is assigned to every citizen and used in all registries.
We identified all Danes (≥18 years) admitted with acute stroke (including ischemic stroke and ICH) from January 2003 to December 2009 (n=11 356; Figure). Patients with multiple strokes during the study period were only included with their first stroke. All the patients had a known history of AF or were diagnosed with AF on admission with the stroke.
The patients were identified in the Danish Stroke Registry. Participation in the project is mandatory for all hospital departments in Denmark treating patients with acute stroke.15 Almost all patients have brain computed tomography or MRI scans in the acute stroke phase (97% of the study population). The completeness of the registration of patients is estimated to be >90% when compared with the Danish National Registry of Patients.
Data on all OAT (warfarin and phenprocoumon) prescriptions filled before the date of admission with stroke were obtained on each patient by linkage with the Danish Medicines Agency’s Medical Registry. The register contains data from 1995 on all prescription drugs dispensed at Danish pharmacies, including type of drug according to the Anatomic Therapeutic Chemical classification system and date of dispensing the drug. In Denmark, OATs are available by prescription only. Patients who had filled a prescription within 90 days before admission, which is the conventional prescription length, were considered current users at the time of admission.
30-Day Mortality Rate
Mortality rate was assessed using information from the Danish Civil Registration System. The Danish Civil Registration System keeps daily updated electronic records on change of address, date of emigration, and changes in vital status.18
Length of Hospital Stay
Length of stay was defined as the time span from admission to discharge. The admission date was defined as the date the patient was admitted to the hospital with stroke or the date of stroke occurrence if the patient was already hospitalized with another diagnosis.
We obtained information on a range of patient characteristics to identify patient-related predictors of the preadmission OAT use and to minimize the risk of confounding when comparing clinical outcome among patients with and without preadmission OAT use (Tables 1 and 2). The data were obtained from the Danish Stroke Registry, the National Registry of Patients,19 the Danish Medicines Agency’s Medical Register, and the Integrated Database for Labour Market Research20 and include age, sex, comorbidity (specific comorbidities and Charlson comorbidity score),21 type of stroke, previous admissions with AF, education, employment, income, preadmission use of other drugs (blood pressure–lowering drugs, lipid-lowering drugs, platelet inhibitors, and antidiabetic drugs), treatment with intravenous thrombolysis, body mass index, alcohol intake, smoking habits, quality of early stroke care, and calendar year.
Furthermore, we computed the preadmission CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, previous stroke/transient ischemic attack, vascular disease, age 65–74 years, sex category; age ≥75 years, and previous stroke carry doubled risk weight)22,23 and HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history, labile international normalized ratio [INR], elderly >65 years, drug consumption/alcohol abuse)24,25 scores for each patient. In the HAS-BLED score, the labile INR component of the score was not included because this information was not available for the entire study population. Both algorithms have been shown to predict thromboembolism and bleeding accurately.23,25
Information on the quality of in-hospital care reflected whether the patient had received a range of recommended processes of care, including early admission to a specialized stroke unit, early administration of antiplatelet or anticoagulant therapy, early examination with computed tomography or MRI, early assessment by a physiotherapist and an occupational therapist, and early nutritional risk assessment.26 We computed the percentage of relevant processes of care received for each patient as a measure of the quality of in-hospital stroke care.
INR measurements were available from all hospitals and general practices for patients from Central Region Denmark and the North Jutland Region (≈32% of the total Danish population). We retrieved data on INR measurements on the day of stroke admission.
We first examined the association between the patient-related characteristics and the preadmission use of OAT within the entire study population to identify patient-related predictors of OAT use. The association was examined using multivariable logistic regression with mutual adjustment of the covariates.
We used multiple imputation to impute the missing values among the covariates assuming that data were missing at random (Stata command: ice).27 We created 5 data sets on the basis of aforementioned covariates. The outcome measures were averaged across the 5 imputations correcting for between- and within- imputation variation.
Second, we examined the association between the preadmission OAT use and the clinical outcomes. In these analyses, we used propensity score matching (5-1 digit matching, Greedy method) to reduce the risk of bias attributable to confounding because preadmission OAT use was not randomly assigned in the study population.28,29 We aimed to match each OAT user with 1 nonuser with similar propensity score using replacement. The propensity score was based on the covariates listed in Table 1 (except for quality of early stroke care). We did not stratify by other covariates when matching. The variables included in the propensity score model were selected among available baseline variables on the basis of known associations with use of OAT and the studied outcomes. As recommended, the model was not specified according to statistical criteria.30 An absolute standardized difference <10% and a variance ratio between 0.8 and 1.25 were considered to support the assumption of balance between the groups30,31 (Figures I and II in the online-only Data Supplement, ○ for the entire study population [n=11 356] and ● for the propensity score–matched patients [n=4350]).
The matching was followed by conditional logistic (stroke severity and 30-day mortality rate) or Cox proportional hazards regression stratified on the matched pairs (length of stay) analysis. In the analyses on 30-day mortality rate and length of stay, we adjusted for differences in the quality of early stroke care. We repeated the analyses while stratifying for type of stroke (ischemic stroke versus ICH). Furthermore, the analyses on 30-day mortality rate were repeated while also adjusting for stroke severity (Scandinavian Stroke Scale score) to evaluate stroke severity as a mediator. Finally, we examined the role of the intensity of preadmission OAT as reflected by the INR value (measured on the date of admission with stroke). These analyses were restricted to patients residing in Central Region Denmark or the North Jutland Region, where data on INR values were available (564 OAT users and 564 propensity score–matched with non–OAT users).
All data analyses were performed using SAS version 9.2 (SAS Institute) and Stata version 11.0 (StataCorp). Figures were generated using R (x64 version 2.12.1).
Table 1 displays patient characteristics of the entire study population (n=11 356) and the propensity score–matched patients (n=4350). A total of 2492 of 11 356 patients (21.9%) used OAT at the time of admission with stroke. When restricting the study population to patients with a previous admission with AF, 2109 of 6679 patients (31.6%) used OAT at the time of admission. The median time from admission with AF to admission with stroke was 573 days (25%/75% percentiles, 147/1471) among these patients.
Predictors of Preadmission OAT Use
Table 2 presents the crude and mutually adjusted association between a range of patient-related characteristics and preadmission use of OAT in the entire study population (n=11 356). Higher CHA2DS2-VASc score at the time of admission was associated with a higher chance of preadmission OAT use, whereas no overall association was found between HAS-BLED score and preadmission OAT use. Furthermore, we observed an increase in OAT use with calendar time.
No substantial differences were seen restricting the analyses to patients with a previous admission with AF (data not shown).
Preadmission OAT Use and Clinical Outcomes
The comparisons of the clinical outcomes were made among the propensity score–matched patients (n=4350). We were able to identify a suitable non–OAT users for 2175 of 2492 OAT users (87.3%). The imbalances in the covariates were largely removed by the propensity score matching (Figures I and II in the online-only Data Supplement).
Preadmission OAT users had a lower overall risk of severe stroke at the time of admission when compared with non–OAT users (Table 3). The lower risk was restricted to patients with ischemic stroke, whereas no difference in severity was observed among patients with ICH. When restricting to the OAT users with available data on admission INR and the corresponding propensity score–matched nonusers, we found the lowest odds ratios of severe stroke among OAT users with an INR value between ≥2.00 and 3.00 or >3.00 when compared with non–OAT users.
Median length of stay was 7 days (25%–75% percentiles; 3–17) among the preadmission OAT users versus 8 days (25%–75% percentiles, 4–18) among the non–OAT users, corresponding to a marginally increased propensity score–matched adjusted hazard ratio of discharge of 1.06 (95% confidence interval, 1.00–1.14).
Overall 30-day mortality rate was also lower among OAT users (Table 4). The lower overall mortality was driven by a low mortality among OAT users admitted with ischemic stroke, whereas OAT users with ICH had an increased mortality when compared with non–OAT users with ICH. We found no differences in 30-day mortality rate between preadmission OAT users and non–OAT users when also adjusted for stroke severity (propensity score–matched adjusted odds ratio, 1.06; 95% confidence interval, 0.81–1.38). When restricting to patients with available INR data, we found indications of a lower mortality among OAT users with an INR value on the day of admission <3.00 although the risk estimates did not reach statistical significance.
Only 1 of 5 patients with AF used OAT at the time of admission with stroke in this nationwide study although the patients in general were characterized by a high predicted thromboembolic risk. Even among patients who had previously been admitted to hospital with AF, <1 of 3 patients with AF used OAT. In propensity score–matched analyses, preadmission use of OAT was associated with lower stroke severity, lower 30-day mortality rate, and marginally shorter length of stay.
The low use of OAT in our study population is comparable with previous reports on patients with AF admitted with stroke7–13 and patients with AF in general.14 Improvements over time have been observed, as is also the case in our study, across geographical settings and healthcare systems, which may reflect the effect of national and international promotion of guidelines recommendations; however, still a high proportion of eligible patients with AF remains insufficiently treated, in particular, when the intensity of the OAT is also taken into account.
The underuse of OAT is particularly troublesome when considering the findings on the association between OAT and clinical outcome after stroke among patients with AF found in our study and in previous studies. Our findings among patients with ischemic stroke are in line with several smaller existing studies, which have also reported preadmission OAT to be associated with lower ischemic stroke severity7,10–13 and lower mortality risk, in particular, among patients with an INR>2.00.7,11,12 We also found that the lower mortality among OAT users seemed to be mediated by the lower stroke severity, which is in accordance with previous findings.11
A key challenge in the process of implementing OAT among patients with AF, which have lasted since the 1990s, is the concerns about bleeding risk.14 These concerns are related to inadequate assessment of stroke risk, including both thromboembolic and bleeding related, in patients with AF. The risk of inducing a fatal or severe ICH on the basis of OAT is a particular concern in this context. The outcome of patients with ICH receiving OAT is, therefore, of major importance when examining the overall effect of OAT on stroke outcomes on a population level. However, to date, only the study by Audebert et al11 seems to have tried to account for the potentially adverse prognostic effect of preadmission OAT on patients with ICH (patients with ICH were also included in the study by Haeusler et al13 but seem not to have been included in the analyses on clinical outcome). Audebert et al11 included 86 patients with ICH in a study based on data from 2 academic and 12 community hospitals in Germany participating in a nonrandomized, community intervention study. Therefore, our study is the first population-based, nationwide study examining the effect of preadmission OAT use on hospitalized stroke outcomes, including both ischemic stroke and ICH. The relevance of including patients with ICH when studying the effect of OAT is underlined by the fact that ≈1 of 5 strokes among patients with AF in our study was an ICH. However, although we found a higher mortality among hospitalized patients with ICH using OAT at the time of admission, the increased risk did not offset the substantially lower mortality among the much higher number of patients with AF with ischemic stroke using preadmission OAT. In combination with the data from the existing studies, these findings clearly suggest that preadmission OAT have an overall beneficial effect on clinical outcome after stroke. It is noteworthy that no studies to date seem to have found preadmission OAT to be associated with a worse clinical outcome. It remains to be clarified whether the possible beneficial effects of OAT are also found with use of the new generation of oral anticoagulants, including dabigatran, rivaroxaban, and apixaban.
The strengths of our study included the prospective, population-based, and propensity score–matched design with complete follow-up, limiting the risk of bias and confounding. Furthermore, the total number of patients included in our study (n=11 356) was >3-fold higher than the total number of patients in all previous studies (n=3686).7–13 The large sample size was particularly evident with regards to patients with ICH (1055 patients included in our study versus 114 patients in all previous studies).
The validity of our estimates depends on the accuracy of the registries. Lack of information on nonhospitalized patients with stroke, including the patients who died before reaching the hospital, could at least theoretically introduce a selection bias if OAT users were more (or less) likely not to be hospitalized when compared with non–OAT users. However, only few patients with stroke die early after the start of symptoms, and there is a long tradition in Denmark for admitting almost all patients with acute stroke to hospital.32 The patients with AF included in our study population either had a known history of AF or were diagnosed with AF on admission with the stroke. We could not differentiate completely between these 2 groups. However, this limitation seemed to be of minor importance because we also found a low rate of OAT use when restricting the population to patients with a previous AF admission.
Although we used a well-balanced propensity score–matched design, we cannot exclude the possibility that our results remain influenced by confounding factors because of the observational nature of the study design. Furthermore, we used filled prescriptions as a proxy measure for actual OAT use although it is likely that a proportion of the patients was not compliant with the treatment, or their INRs might not have been in good control because of other factors (eg, change dietary habits or use of new medications). Such misclassification of OAT use would lead to an underestimation of the true effect of preadmission OAT use on stroke outcomes. Finally, we used stroke severity, length of hospital stay, and 30-day mortality rate as clinical outcomes. Other outcomes, in particular functional level after discharge (eg, modified Rankin score), are also of major interest; however, such data were not available in our study population.
In conclusion, only few patients with acute stroke with AF received OAT at the time of hospitalization with stroke. In a propensity score–matched analysis, preadmission OAT was associated with less severe stroke and lower 30-day mortality rate. Further efforts seem warranted to ensure OAT to all eligible patients with AF because OAT seems to reduce the risk of cardioembolic stroke and at the same time the risk of an adverse clinical outcome should a stroke occur.
Sources of Funding
The study was supported by a grant from Bristol-Myers Squibb (BMS) and Pfizer. BMS and Pfizer had no influence on the study design, data collection, the presentation of data, or the conclusions made.
Dr Johnsen is member of an advisory board for BMS/Pfizer and has received speakers honoraria from Bayer, Boehringer-Ingelheim, BMS, Pfizer, and AstraZeneca. Dr Hansen is member of advisory boards for BMS/Pfizer and Boehringer-Ingelheim and has received speakers honoraria from Boehringer-Ingelheim, BMS/Pfizer, and AstraZeneca. Dr Husted is member of advisory boards for AstraZeneca, Bayer, and BMS/Pfizer and has received research grants from GSK, Boehringer-Ingelheim, BMS, Pfizer, and GlaxoSmithKline. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.001792/-/DC1.
- Received April 12, 2013.
- Accepted October 22, 2013.
- © 2013 American Heart Association, Inc.
- Wolf PA,
- Abbott RD,
- Kannel WB
- Jørgensen HS,
- Nakayama H,
- Reith J,
- Raaschou HO,
- Olsen TS
- Aguilar MI,
- Hart R
- O’Donnell M,
- Oczkowski W,
- Fang J,
- Kearon C,
- Silva J,
- Bradley C,
- et al
- Mainz J,
- Krog BR,
- Bjørnshave B,
- Bartels P
- 17.↵Scandinavian Stroke Study Group. Multicenter trial of hemodilution in ischemic stroke--background and study protocol. Stroke. 1985;16:885–890.
- Lynge E,
- Sandegaard JL,
- Rebolj M
- Langagergaard V,
- Palnum KH,
- Mehnert F,
- Ingeman A,
- Krogh BR,
- Bartels P,
- et al
- Olesen JB,
- Lip GY,
- Hansen ML,
- Hansen PR,
- Tolstrup JS,
- Lindhardsen J,
- et al
- Sterne JA,
- White IR,
- Carlin JB,
- Spratt M,
- Royston P,
- Kenward MG,
- et al
- Thorvaldsen P,
- Davidsen M,
- Brønnum-Hansen H,
- Schroll M