|Home | About | Journals | Submit | Contact Us | Français|
Antithrombotic medications (anticoagulants and antiplatelets) are often withheld in the periprocedural period and after bleeding complications to limit the risk of new or recurrent bleeding. These medications are also stopped by patients for various reasons such as cost, side effects, or unwillingness to take medication.
Patient records from the population-based Greater Cincinnati / Northern Kentucky Stroke Study were reviewed to identify cases of ischemic stroke in 2005 and determine the temporal association of strokes with withdrawal of antithrombotic medication. Ischemic strokes and reasons for medication withdrawal were identified by study nurses for subsequent physician review.
In 2005, 2,197 cases of ischemic stroke among residents of the region were identified via hospital discharge records. Of the 2,197 ischemic strokes, 114 (5.2%) occurred within 60 days of an antithrombotic medication withdrawal: 61 (53.5%) of these after stoppage of warfarin and the remainder after stoppage of an antiplatelet medication. Of the strokes following withdrawal, 71 (62.3%) were first-ever, and 43 (37.7%) were recurrent; 54 (47.4%) occurred after withdrawal of medication by a physician in the periprocedural period.
The withdrawal of antiplatelet and antithrombotic medications in the 60 days preceding an acute ischemic stroke was associated with 5.2% of ischemic strokes in our study population. This finding emphasizes the need for thoughtful decision making concerning antithrombotic medication use in the periprocedural period and efforts to improve patient compliance.
The frequency and burden of ischemic strokes following discontinuation of antithrombotic (AT) medication in the population overall is not defined.1The cessation of antiplatelet (AP) and anticoagulant (AC) medications has been associated with an increased incidence of ischemic stroke and other thrombotic events within several months of medication discontinuation.1–4 Experimental studies in humans following discontinuation of aspirin have also demonstrated a stimulating effect on the cyclooxygenase activity in human platelets under certain conditions that may last up to six weeks.3, 5
Reasons for discontinuance of AT medications by physicians and patients include planned procedures, bleeding complications, fall risk, financial burden, and poor compliance. The most common reason in several cohorts is discontinuation in the periprocedural period to limit the risk of bleeding complications.1, 3
This project was a retrospective review designed to determine the number of acute ischemic strokes associated with cessation of AP or AC medications in an ongoing population-based study. The data regarding stoppage of medication, timing of stoppage, and reasons for stoppage were a priori included in the chart abstraction packet for the epidemiology study for 2005 to address the issue of medication withdrawal.
The Greater Cincinnati/Northern Kentucky Stroke Study (GCNKSS) is a population-based study of stroke and transient ischemic events in a region that includes two southwestern Ohio counties and three contiguous Northern Kentucky counties separated by the Ohio River. Only residents of the five study counties are considered for case ascertainment. The most recent completed study period was calendar year 2005. In the GCNK region, 17 hospitals were active in 2005. Previous studies have documented that residents of the five counties who have a stroke exclusively seek care at these hospitals rather than at hospitals in the outlying region.6, 7 This study was approved by the institutional review boards at each participating hospital. Details of case ascertainment for 2005 and previous study periods have been previously published.6 More detailed methodology about case ascertainment, detailed abstraction of data from the medical record from all sources in the community, and physician phenotyping of cases can be found in the online supplement. Only acute ischemic strokes, both first-ever-in-a-lifetime or recurrent, were included in the present analysis.
Each verified case of acute ischemic stroke was reviewed in detail by a study nurse and physician investigators to determine whether there was a proximal withdrawal of AC or AP medications prior to stroke onset. The time interval from withdrawal of medication to stroke event was ascertained, and the record was reviewed to determine the reason for withdrawal of medication. Reasons were broadly categorized into those stopped in accordance with physician direction and those stopped independently by the patient. The reasons for stoppage by physician were specified as periprocedural, bleeding, supratherapeutic anticoagulation, history of falls, or “other,” and reasons why patients discontinued medications were specified as non-compliance, financial, “other,” and unknown. AP medications included clopidogrel, aspirin (ASA), dipyridamole, and ticlopidine; AC medications included warfarin sodium and unfractionated and low molecular weight heparins.
For the analysis, stroke events were classified into three groups, based on AT status at time of stroke onset: (1) patient was not on an AT medication within 60 days of stroke onset; (2) patient was on an AT medication at time of onset; and (3) patient stopped taking an AT medication within 60 days of onset. The prospective rationale for the cut-off of 60 days included the following: 1) Evidence of increased risk of stroke and thrombotic events for 1–2 months in prior observational studies and clinical trials that is consistent with experimental studies showing rebound effects after discontinuation of antiplatelet medication that can last up to six weeks.3–5, 8, 9 2) Some patients have no bridging therapy and it may be 1–2 months before stable anticoagulation with warfarin is reattained. 3) Some patients may have substantial delays in restarting of AT medication after a procedure because of complications or because of failure by the responsible physician to restart medication.
Data were managed and analyzed using SAS®, versions 8.02 and 9.2 respectively (SAS Institute, Cary, NC). Generalized estimating equations (GEE)10 were used to examine bivariate differences between strokes in the Not on AT group vs. the Stopped AT and On AT groups combined, and also between the “On AT” group and “Stopped AT” group. This statistical method was used to account for patients with more than one event in the time period. The working correlation structure that gave the best model fit was obtained. A binary or multinomial distribution was specified for categorical variables, as appropriate. Appropriate transformations were used when necessary. Data are reported as frequencies and percentages, or means with associated standard errors, as appropriate.
We identified 2,197 ischemic strokes in 2,090 adult patients in the GCNKSS population during 2005. The mean (SEM) age of the stroke patients was 70.6 (0.32); 55.1% were female, and 21.6% were black. Of the 2,197 ischemic strokes, 1,991 occurred in patients who had a single stroke during the study period, 91 occurred in patients who had two strokes, and 8 in patients who had three strokes;1,710 (77.3%) were first-ever-in-a-lifetime ischemic strokes, and 487 (22.2%) were recurrent ischemic strokes.
The AT status breakdown of the 2,197 strokes is (1) 999 not on AT within 60 days of stroke onset (Not on AT Group), (2) 1,084 on AT at time of onset (On AT Group), and (3) 114 where AT was stopped within 60 days of onset (Stopped AT Group). Of the 114 patients in the Stopped AT group, 108 had clinical symptoms lasting more than 24 hours and six had clinical symptoms of less than 24 hours with a diffusion positive stroke on MRI.
Table 1 compares the demographics and clinical characteristics of the three groups. As compared with the On AT group, the Stopped AT group was more likely to have atrial fibrillation (41.2 vs. 22.9%), to be male (59.7% vs. 44.9%), and tended toward a higher baseline NIHSSS (5.7 vs. 5.0). Of the 114 patients in the Stopped AT group, 42.1% had died within one year as compared with 29.2% of the On AT Group. The percentage of strokes in the Stopped AT Group with a mRS of 0–2 or a return to pre-stroke mRS at discharge (33.3%) was significantly lower than those in the On AT Group (43.9%, p=0.03). In multivariate logistic regression models of mortality at 3 months and one year which controlled for atrial fibrillation, diabetes, black race, age, prior infarct and baseline mRS, being on an AT medication at time of stroke onset remained signficantly associated with lower mortality, compared with stopping an AT medication prior to onset: 3-month mortality odds ratio 0.55 (95% CI - 0.34–0.90); 1-year mortality odds ratio 0.56 (95% CI - 0.36–0.87).
Table 2 shows the distribution of the 114 withdrawals of medication, by type of drug and time interval; 61 patients (53.5%) had strokes after stoppage of warfarin, which include 10 patients who had stoppage of both warfarin and an antiplatelet agent. The remainder had strokes after stoppage of an antiplatelet agent.
Table 3 shows the timing of ischemic strokes after discontinuation of AT medication and the reasons why the medications were stopped. Nearly half of the patients had medications stopped by physicians for procedures, and over half of these patients’ strokes occurred within 7 days of medication stoppage. Nineteen (35.2%) of the 54 patients who had medication stopped for a procedure had recurrent strokes. Patient compliance, bleeding complications, and financial pressures were the next largest reasons for stoppage of medication. Stroke events were clustered mostly in the first two weeks after stoppage of medications, which was particularly the case for medications stopped for procedures or for supratherapeutic anticoagulation. Strokes for other categories of medication withdrawal also occurred more frequently in the first several weeks after stoppage of medication, but strokes continued to occur up to 60 days after stoppage. The number of ischemic strokes overall after discontinuation of AT medication was 59 in week one, 21 in week two, 22 in weeks 3–4, and 12 in weeks 5–8.
Supplemental Table 1 shows the specific procedures for which AT medication was discontinued. The most common surgeries and procedures were cardiac surgeries and procedures (11), hip and knee surgeries (9), carotid artery surgeries and procedures (8), epidural steroid injections (5), and endoscopic procedures (5). Of these 38 most common procedures, only 5 (13.2%) patients had bridging heparin/LMWH, of which two were cardiac valve replacement surgeries.
Antithrombotic medications, which include both antiplatelet agents and anticoagulants, have had a large impact in the reduction of first-ever and recurrent ischemic stroke in patients with atrial fibrillation, prior ischemic stroke or transient ischemic attack, prior myocardial infarction, and other subpopulations at higher risk for ischemic events.11–13 Yet, the effectiveness and safety of AT medications depend upon patient compliance as well as their appropriately prescribed use in various clinical situations in which bleeding risks of these medications may be temporarily or chronically increased. Decisions regarding perioperative management of AT medication are among the most common and complex treatment decisions that physicians face.
Our population-based study indicates that about 5% of ischemic strokes are associated with withdrawal of AT medications within 60 days of onset. In our study, strokes associated with discontinuation of AT medication had significantly greater mortality and morbidity than strokes occurring while on AT medication, even after controlling for higher rates of atrial fibrillation and other variables associated with poorer outcome. This observation is consistent with a Kaiser Permanente study of patients with atrial fibrillation in which stroke patients who had an INR of less than 2.0 at admission had significantly more severe strokes and higher mortality as compared with patients with an INR of ≥ 2.0.14
Acute stroke patients are often aphasic or cognitively impaired, ascertainment of medication history may be incomplete, particularly in those patients on medications that are purchased over-the-counter and may not be listed in pharmacy records. Thus, our estimated percent of strokes (5.2%) after stopping AT may be an underestimate. In a prospective survey of discontinuation of antiplatelet medication in stroke patients, Sibon and colleagues reported that 5% of 289 ischemic stroke patients at a single University Hospital in Bordeaux, France had stopped an antiplatelet agent within 30 days of stroke onset.1 All of the strokes occurred within 6–10 days after stoppage of medication, and 54% of patients had medication held by a physician for surgical procedures (6) or another cause (1). Maulaz and colleagues reported a case-control study from Centre Hospitalier Universitaire Vaudois in Lausanne in which 4.2% of ischemic stroke patients had discontinued aspirin within the prior month; all strokes occurred within 25 days of discontinuation.3 Neither of these studies included anticoagulant medications.
Our data suggest that the risk of AT medication withdrawal is related to time since withdrawal as well as to reinstitution of an effective AT medication. Stroke events were clustered mostly in the first two weeks after stoppage of medications, which was particularly the case for medications stopped for procedures or for supratherapeutic anticoagulation. Strokes for other categories of medication withdrawal also occurred more frequently in the first several weeks after stoppage of medication, but strokes continued to occur up to 60 days after stoppage. This is illustrated by events associated with medication stoppage due to bleeding complications in which AT medications would be less likely to be restarted quickly, if at all. Our data are potentially susceptible to recall and documentation biases given our methodology. However, given that the time period prior to stroke was only 60 days, recall bias is probably small.
Our data also suggest that withdrawal of warfarin is associated with a greater risk of ischemic stroke than withdrawal of antiplatelet agents, which likely reflects in part the higher risk of ischemic events in subjects on warfarin. In a 2005 telephone survey of a random population sample from the same Greater Cincinnati study region, 40% of whites (mean age 62) and 36% of blacks (mean age 59) reported regular aspirin use.15 Questions regarding use of warfarin and other antiplatelet agents were not part of the 2005 survey. Other recent population studies provide useful comparison of relative use of AT agents. The REGARDS cohort study includes 30,166 community dwelling adults over the age of 45 who were enrolled throughout the U.S. during 2003–2006. Of this cohort (mean age 66.1, s.d. 9.0 years), 47.5% were on aspirin, 5.0% were on a thienopyridine or dipyridamole, and 3.6% were on warfarin (personal written communication, Suzanne Judd, George Howard, 8/9/10). Using the United Kingdom General Practice Research Database from 2000 through 2005, Delaney and colleagues reported a case-control study of gastointentinal bleeding that included matched 40,171 controls (mean age 69.1 years, s.d. 17.7); 3.2% of these controls were on warfarin and 1.8% were on a thienopyridine.16 In contrast to the much greater use of antiplatelet agents than warfarin in the general population, over half of ischemic strokes in our study population occurred in patients who had withdrawal of warfarin. The majority of these patients are on warfarin for atrial fibrillation which represents a group at higher risk of thromboembolic event, particularly those with prior strokes or high CHAD2 scores. 2, 17
To calculate an accurate risk of thrombotic events associated with stoppage of AT medication, one would need to know the prevalence of AT medication in the general population, the frequency and duration of medication stoppage, the number of ischemic and hemorrhagic events on and off AT medication, and the use of bridging therapy during the same time period. Such data are currently not available in our population nor in any larger population of which we are aware. In addition, determination of the risk of ischemic events following stoppage of AT medication can be confounded in part by the reason for stoppage (e.g. cardiac surgery which has a known risk of ischemic stroke). Several smaller cohort studies of subjects who had interruption of AT therapy indicate the absolute risk of thrombotic events associated with interruption of warfarin for procedures is about 1% but is higher in those subjects with atrial fibrillation and prior stroke. 8, 9, 17–19 The effect of bridging therapy on reduction of thrombo-embolic events in these cohort studies is unclear since subjects who receive bridging therapies are at higher baseline risk of thrombo-embolic events.19
The ongoing BRIDGE Study, a randomized trial funded by the National Heart, Lung and Blood Institute, and the PERIOP 2 Study, a randomized trial sponsored by the Lawson Health Research Institute, will hopefully provide more clarity about the role of bridging therapy in patients with atrial fibrillation and/or mechanical valves with a higher risk of thromboembolic events. Nonetheless, our data support current guidelines recommending minimization of the time off AT medication, particularly in those patients with prior stroke, mechanical heart valves, or atrial fibrillation with higher CHAD2 scores, and continuation of antiplatelet medication in those patients undergoing cardiac and carotid artery procedures.2 In addition, the specific properties of new oral AT medications like dabigatran (half-life, measurement of biologic effect, etc.) will need to be considered in these ongoing studies and in clinical practice.20
The ACCP Guidelines for perioperative management of AT therapy provide a stratification of risk as well as recommendations for continuation or bridging of AT therapy based upon the underlying indication for AT medication and the risk of bleeding for a given surgery.2 For example, for minor dental, ophthalmic, and dermatologic procedures, maintainence of anticoagulation or antiplatelet therapy throughout the perioperative period is recommended by the ACCP. However, the ACCP Guidelines do not address all minor procedures such as lumbar punctures. Lumbar punctures have a very low risk of epidural hematomas in the setting of antiplatelet agents, yet national guidelines recommend stoppage of thienopyridines for seven days prior to performance of a lumbar puncture, one of the most common procedures in neurologic practice, without stratification of risk.21 Local institutions or physicians may be even more rigid and require stoppage of any antiplatelet agents, including aspirin, for minor procedures even in those patients with higher risk of thrombo-embolic events.22
The financial burden of medication also impacts patient compliance. We found that 8.4% of infarcts that occurred within 60 days of medication withdrawal were associated with patient-directed cessation of medication as a result of financial burden. This may underestimate the true burden represented by financial constraints in that 16.7% of patients with infarcts within 60 days of medication withdrawal were categorized as noncompliant with medications, yet the reasons for their non-compliance were not always available.
The withdrawal of antitcoagulant and antiplatelet medications is associated with a substantial number of acute first-ever and recurrent ischemic strokes. This finding emphasizes the need for thoughtful decision making concerning antithrombotic medication use in the periprocedural period and for recruitment in ongoing trials addressing this issue.
The authors would also like to express their appreciation to the following hospitals for their participation in the Greater Cincinnati/Northern Kentucky Stroke Study: University Hospital; Good Samaritan Hospital; Bethesda North Hospital; Christ Hospital; St. Elizabeth Edgewood, Covington, Florence, and Ft. Thomas Hospitals; Mercy Anderson, Clermont, Fairfield, Mt. Airy and Western Hills Hospitals; the Jewish Hospital; Deaconess Hospital; Cincinnati Children’s Hospital Medical Center; and the Cincinnati Veteran Affairs Medical Center. The contents of this manuscript do not represent the view of the Department of Veterans Affairs or U.S. Government or the other hospital systems. The authors also thank George Howard, DPH and Suzanne Judd, PhD for the use of REGARDS data.
Funding support: This study was supported by NIH Grant R01 NS30678.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: Dr. Flaherty and Dr. Kleindorfer have acted as consultants for Boehringer Ingelheim and Dr. Khatri has acted as consultant for Otsuka pharmaceuticals. Both companies produce anti-thrombotic medications.