Inappropriate preinjury warfarin use in trauma patients: A call for a safety initiativeHH Hon1, A Elmously2, CD Stehly1, JC Stoltzfus3, MA Granson4, SP Stawicki5, BA Hoey1
1 Department of Surgery, Regional Level I Trauma Center, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA
2 Department of Surgery, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA
3 The Research Institute, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA
4 Department of Research and Innovation, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA
5 Department of Surgery; Regional Level I Trauma Center; Department of Research and Innovation, St. Luke's University Health Network, Bethlehem, Pennsylvania, USA
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0022-3859.175004
Source of Support: None, Conflict of Interest: None
Introduction: Warfarin continues to be widely prescribed for a variety of conditions. It has been shown that preinjury warfarin may worsen outcomes in trauma patients. We hypothesized that a substantial proportion of injured patients seen at our institution were receiving preinjury warfarin for inappropriate indications and that a significant number of such patients had subtherapeutic or supratherapeutic international normalized ratios as well as increased mortality. Materials and Methods: A retrospective review of registry data from a Level I trauma center was conducted for the period from January 2004 to July 2013. Included were patients aged ≥22 years (based on the youngest recorded patient on warfarin in this study). Abstracted variables included patient age, Injury Severity Score (ISS), Maximum Abbreviated Injury Score for Head (MAISH), mortality, hospital length of stay (HLOS), indication(s) for anticoagulant therapy, admission Glasgow Coma Scale (GCS), and admission international normalized ratio (INR). Suitability of warfarin indication(s) was determined using the most recent American College of Chest Physicians (ACCP) Guidelines. Inappropriate warfarin administration was defined as use inconsistent with these guidelines. For outcome comparisons, a case-control design with 1:1 ratio was used, matching patients taking preinjury warfarin to a random sample of trauma patients who were not taking warfarin. Severe traumatic brain injury (sTBI) was defined as MAISH ≥4. Results: A total of 700 out of 14,583 patients aged ≥22 years were receiving preinjury warfarin (4.8% incidence, WG). This group was age- and ISS-matched with 700 patients (4.8% total sample) who were not taking warfarin (NWG) in a total case-control sample of 1,400. The two groups were similar in age, gender, ISS, and initial GCS. According to the ACCP guidelines, 115/700 (16.4%) patients in the warfarin group were receiving anticoagulation for inappropriate indications. Nearly 65% of the patients were outside of their intended INR therapeutic window (43.4% subtherapeutic, 21.6% supratherapeutic). Overall, median HLOS was greater among patients taking preinjury warfarin (4 days vs 2 days, P < 0.010). Mortality was higher in the WG (7.4% or 52/700) than in the NWG (1.9% or 13/700, P < 0.010). Patients with sTBI in the WG had significantly greater mortality (12.8% or 34/266) than those with sTBI in the NWG (5.3% or 9/169, P < 0.013). Conclusion: A significant proportion of trauma patients admitted to our institution were noted to take warfarin for inappropriate indications. Moreover, many patients taking warfarin had either subtherapeutic or supratherapeutic INR. Although warfarin use did not independently predict mortality, preinjury warfarin use was associated with greater mortality and HLOS in the subset of patients with sTBI. Safety initiatives directed at better initiation and management of warfarin are needed.
Keywords: Anticoagulation, appropriate use, indications, patient safety, side effects, therapeutic window, warfarin
The Institute of Medicine's report, "To Err is Human," has been instrumental in placing patient safety initiatives at the forefront of current medical practice. The report found that nearly 100,000 patients die each year due to medical errors, costing society up to $29 billion annually. , As a result, the Joint Commission and other health policy organizations instituted safety initiatives that include such processes as the administration of perioperative antibiotics, deep vein thrombosis prophylaxis, as well as universal procedure "time-out"and "checklist" protocols. , Despite these recent initiatives, medical errors continue to occur, often involving medications with high risk profiles.  Warfarin, used for prevention and treatment of thromboembolic disease, is one such example. Acceptable indications for warfarin therapy per the American College of Chest Physicians (ACCP) Guidelines include atrial fibrillation, deep venous thrombosis (DVT), and prosthetic heart valves.  Despite challenges involving patient compliance and therapeutic monitoring, many practitioners choose to use warfarin for a variety of clinical uses, often involving questionable applications. ,, For warfarin users, the risk of mortality is 1.4% per year, that of major hemorrhage is 3% per year, and the overall hemorrhage risk is 9%. 
Trauma patients on preinjury warfarin are more likely to experience morbidity and mortality, often out of proportion to the injury mechanism or severity. ,, Howard et al.  demonstrated that preinjury warfarin use adversely affected outcomes in elderly patients following falls from standing, especially when severe brain injury was present. Lavoie et al.  found that among patients >55 years of age with a closed head injury, preinjury warfarin use was associated with greater brain injury severity and mortality. We hypothesized that a significant number of trauma patients receive warfarin for inappropriate indications. We also suspected that a significant number of these patients have either subtherapeutic or supratherapeutic international normalized ratios (INR). Finally, we set out to confirm that warfarin, particularly in patients with traumatic brain injury (TBI), has the potential to increase mortality.
A retrospective case-control study of blunt trauma patients taking pre-injury warfarin was conducted at our Level I trauma center between January 1, 2004 and July 31, 2013. Due to the youngest age noted in the warfarin group (WG) being 22 years, patients younger than 22 years were excluded. Also excluded were penetrating trauma victims, pregnant patients (due to warfarin being contraindicated during pregnancy), and prisoners [standard Institutional Review Board (IRB) exclusion]. For each patient in the warfarin group, an age- and ISS-matched control patient was selected from the cohort of 13,883 patients who did not have documented warfarin use. Key study comparisons involved 1:1 case-control matched group of 1,400 patients. This matching ratio was selected after considering the observed prevalence of warfarin use and the relatively large sample size of the current study, with the investigators accepting the slightly lower statistical power associated with such choice. 
Abstracted chart data included patient age; gender; injury severity score (ISS); Glasgow Coma Scale (GCS); INR on hospital admission; hospital length of stay (HLOS); stated indication for warfarin use, abbreviated injury scale (AIS) for head, chest, and abdomen; and in-hospital mortality. Regarding the INR, the therapeutic adequacy of warfarin was based on the admission level (as measured during the trauma resuscitation) in conjunction with the documented indication for warfarin use.
Our primary endpoints included the determination of appropriateness of the stated indication for warfarin use, appropriate level of anticoagulation in the context of the stated indication, hospital length of stay, and mortality. Additional subset analysis was performed on patients with severe TBI (sTBI), defined as Maximum Abbreviated Injury Score for Head (MAISH) of ≥4, focusing on in-hospital mortality and HLOS. The ACCP guidelines were used as the gold standard for determining the clinical appropriateness of warfarin therapy.  Of note, for cases where indication for warfarin was not known (or otherwise stated), a phone call to the primary care physician was made as part of our institutional medication reconciliation protocol. A summary of warfarin uses that were deemed either "indicated" or "not indicated" is provided in [Table 1].
Statistical analyses was performed using the Mann-Whitney U test for nonnormally distributed continuous data, analysis of variance (ANOVA) for normally distributed continuous data, and chi-square test for nominal data. Using mortality as the primary endpoint, variables achieving significance level of P<0.20 on univariate analysis were included in subsequent regression analysis. Finally, the resultant multivariate model's predictive ability in respect to mortality was evaluated using receiver operating characteristic (ROC) curves and by computing the corresponding area under the curve (AUC) values. We utilized SPSS 18 for Windows (IBM Corp., Armonk, New York, USA) to perform the analyses. Statistical significance was defined as P < 0.05. Data were reported as mean ± standard deviation or median with interquartile range (IQR). The study was approved by the IRB.
A total of 14,583 blunt trauma admissions aged 22 years and older were evaluated during the study period. Within this group, 700 were taking preinjury warfarin (346 males, 354 females in WG) and 700 were controls (372 males, 328 females in NWG). The combined 1,400-patient group's mean age was 76.9 ± 13.5 years (WG 76.7 ± 13.3 vs NWG 77.1 ± 13.7, P = n/s), with mean ISS of 10.0 ± 8.09 (median 9, IQR 4-14; WG 9.8 ± 7.4 vs NWG 10.2 ± 8.7, P = n/s), and mean GCS of 14.3 ± 2.51 (WG 14.4 ± 2.6 vs NWG 14.3 ± 2.4, P = n/s). The MAISH was 2.91 ± 1.20 (WG 2.6 ± 1.1 vs NWG 3.1 ± 1.22, P < 0.01). A summary of patient characteristics for each group (WG and NWG) is provided in [Table 2].
On multivariate analysis, independent predictors of mortality were ISS [adjusted odds ratio (aOR) 1.17, 95% confidence interval (CI) 1.098-1.247)], GCS (aOR 1.25, 95% CI 1.100-1.428), and MAISH (aOR 2.22, 95% CI 1.239-3.963). Patient age, the use of warfarin, and INR did not independently predict mortality. Upon sensitivity analysis, we found that the current multivariate model (ISS, GCS, and MAISH) resulted in excellent predictive characteristics, superior to each component variable in isolation[Figure 1].
In order to better define the relationship between warfarin use and severity of TBI (with MAISH as surrogate measure), we performed a subset analysis focusing on fatalities among patients with sTBI [e.g., MAISH ≥4, [Table 3]. Here, patients in the WG were noted to have a mortality of 12.8% (34/266), which was approximately two times higher than the control group's mortality of 5.3% (9/169, P < 0.02). However, the overall distribution of mortality did not significantly differ by the admission INR level. Finally, the median HLOS was greater in patients taking preinjury warfarin [4 days versus 2 days, P < 0.01, [Table 3].
When comparing indications for anticoagulation observed in the WG and the ACCP guidelines,  129/700 (18.4%) of the treatment group were taking warfarin for potentially inappropriate indications. Of those, only 14/129 (10.9%) patients were shown to have appropriate indications upon further determination, with the remaining 115/129 (89.1%) receiving warfarin for either inappropriate or unknown (and presumably inappropriate) indications. The most common appropriate indications were atrial fibrillation (AF, 68.5% or 401/585), followed by venous thromboembolism (VTE, 19.5% or 114/585), and prosthetic heart valves (9.6% or 56/585) [Figure 2]. Most common among the "inappropriate" indications were "cardiac uses" in 54% (62/115) of cases and "no known use" in 27% (31/115) of instances.
According to ACCP guidelines, therapeutic INR levels for atrial fibrillation and VTE are 2.0-3.0; For prosthetic valves, the goal INR is 2.5-3.5.  Among patients with appropriate warfarin use, 65.1% (381/585) of the group were outside the intended therapeutic window with regard to their INR (252/381 subtherapeutic, 129/381supratherapeutic). The appropriateness of admission INR levels in the context of desired therapeutic windows is summarized in [Figure 3].
Patient safety and quality of care are at the heart of the modern health care delivery environment. Adverse events occur inapproximately 1 in 30 of hospitalizations, with over 13.5% of these events resulting in mortality. , Medication errors alone, both in hospital and outpatient settings, are estimated to contribute to thousands of deaths annually.  Not all of the medication-related errors result in actual harm; however, those that do can be devastating, especially when serious drug-drug interactions or irreversible side effects occur. , A recent study found that approximately 2% of patients admitted to hospital experienced a preventable adverse drug event, costing tens of billions of dollars in aggregate. 
Warfarin is a medication with an inherently high risk profile.The potential for spontaneous bleeding is significant and studies have shown that warfarin can substantially elevate the risk of intracranial hemorrhage, with associated increase in fatal events.  Still, this anticoagulant continues to be widely prescribed as its overall benefits outweigh the risks when used properly and for appropriate indications. Having said that, debate continues regarding absolute indications for anticoagulation, especially in the context of the overall risk-benefit equation.
Clinical studies suggest a reduction of cardiovascular events following myocardial infarction (MI) with combined aspirin-warfarin therapy versus aspirin alone. [16-19] However, a substantial portion of this evidence was gathered prior to the widespread use of coronary interventions and the establishment of aspirin and clopidogrel as the preferred post-MI combination therapy. , In the current era, the great majority of patients with both non-ST elevation and ST elevation acute coronary syndromes (ACS) are treated with an early percutaneous coronary intervention (PCI) and stenting. The most recently published American Heart Association/American College of Cardiology (AHA/ACC) Guidelines for Secondary Prevention for Patients with Coronary and Other Atherosclerotic Vascular Diseases recommend the use of aspirin and clopidogrel for up to 12 months following an ACS.  These guidelines implicitly state that there is an increased risk of bleeding when warfarin is added to aspirin and clopidogrel. Warfarin use is recommended only when clinically indicated for selected post-MI patients, including those with history of atrial fibrillation, atrial flutter, or left ventricular thrombus.  Therefore, coronary artery disease (CAD) alone was excluded as an acceptable indication for warfarin therapy in the 2012 ACCP guidelines. 
In the current study, more than 16% patients did not have an appropriate indication for warfarin use as per the ACCP guidelines.  According to our institutional protocols, at the time of admission, patients or their family members were asked why they were on warfarin. If any questions remained regarding the indication, then the patient's primary physician was contacted directly. Overall, the most common reason for warfarin therapy was atrial fibrillation, accounting for more than half of warfarin use in this study. Not surprisingly, VTE was the next most common indication.
In the "inappropriate warfarin use" group, over half of the patients were on warfarin therapy for CAD. Given the discussion above regarding the benefits of aspirin coupled with clopidogrel, isolated CAD is no longer an appropriate indication for warfarin anticoagulation, especially given the elevated risk of bleeding. Interestingly, a number of primary care physicians contacted by our team acknowledged "anecdotally" that they were aware of the ACCP guidelines, yet were reluctant to stop warfarin in these specific circumstances. Perhaps even more surprising was the number of patients with no discernable indication for warfarin, accounting for approximately 1 in 4 individuals in the "inappropriate warfarin use" group. In this subset, the patients' physicians frequently attributed warfarin use to "other practitioners" and very rarely questioned the actual indication(s). When we queried those "other practitioners," many were unable to provide the indication for beginning and/or continuing warfarin. If one extrapolates the overall percentage of patients in this study who were taking warfarin inappropriately to the 2.9 million patients currently taking warfarin in the USA, there may be over 700,000 patients being improperly treated. Given the focus on patient safety and quality of care, this is an unacceptable phenomenon. Consequently, patient safety initiatives aimed at reducing unnecessary medication use are acutely needed, with focus on real-time, critical evaluation of ever more complex medication regimens. 
In the current study, admission INR determinations were recorded and cross-checked against both the validity of clinical indication and the desired therapeutic range.  Patients were then grouped into three categories: Subtherapeutic, therapeutic, and supratherapeutic [Figure 3]. According to the ACCP guidelines, desired therapeutic INR levels for atrial fibrillation and VTE are 2.0-3.0, and for prosthetic heart valves 2.5-3.5. Overall, only about 35% of patients with known appropriate indication were noted to have their INR within the desired therapeutic window. Consequently, patients with "subtherapeutic" INR levels are left without the protection that the anticoagulation was intended to provide, and the "supratherapeutic" group may be at an elevated risk of spontaneous hemorrhage. Interestingly, we were unable to demonstrate greater mortality with higher INR levels, as previously shown by others.  Although this may be due to various biases and potential confounders, the fact that nearly 2 out of every 3 patients in this study were "out of therapeutic window" indicates that our current management practices are inadequate and must be revised.
The impact of preinjury warfarin on trauma patients is a topic of significant debate. It is inherently difficult to reliably associate effects of anticoagulation with clinical outcomes across a broad spectrum of injuries that cannot be easily standardized or categorized. Intuitively, a medication that inhibits clotting factors seems likely to lead to an increase in the manifest severity of the injuries sustained. Having matched our warfarin and control groups for age, gender, and ISS, the difference in the proportion of patients with MAISH ≥4 could potentially be associated with such an effect. However, this cannot be proved using a retrospective study design, and we can only speculate that the observed increase in mortality among patients with MAISH ≥4 who were taking warfarin is due to the use of anticoagulation. In a number of previous trauma studies, there was no significant increase in mortality among patients receiving preinjury warfarin. ,,, When focusing on trauma patients with TBI, some evidence exists to suggest that anticoagulation may adversely affect outcomes in this subgroup.  One study found that preinjury anticoagulation in trauma patients with intracranial injury was associated with 4-5-fold higher risk of death compared to nonanticoagulated patients.  In other studies, although preinjury warfarin did not increase the risk of fatal hemorrhagic complications in the absence of head trauma, there was increased mortality in patients who sustained intracranial injury, , especially when patients older than 55 years of age were considered.  Franko et al.,  showed that warfarin use may be associated with higher risk of traumatic intracranial hemorrhage and mortality, an effect more pronounced in patients who are ≥70 years old.
Given the significant increase in the number of older trauma patients,  the effects of anticoagulation need to be further examined in this specific context. Here, the overall mortality was noted to increase with preinjury warfarin administration - a finding further accentuated by the concomitant presence of elevated MAISH. , Sarani et al.  demonstrated that among patients who fell from standing and sustained TBI, warfarin use and GCS ≤12 were independently associated with intensive care unit admission and mortality. Considering the above evidence in conjunction with this study's findings, there are both the urgency and the need for coordinated patient safety action to improve our current warfarin (and other anticoagulant) prescribing patterns, and to reduce the incidence of potentially preventable mortality. ,,
Among key outcomes examined in this study was the relationship between HLOS and warfarin use. Because it is difficult to assess the overall morbidity in the context of warfarin use, we chose to utilize HLOS as a surrogate marker for all-cause morbidity. We found that preinjury warfarin use was associated with a median HLOS of 4 days compared to 2 days in the NWG. This figure only begins to show the potential impact of warfarin use on the US health care system. At our institution, the charge for a surgical floor bed is approximately $3,000 per day and the charge for an ICU bed exceeds $6,600 per day. These amounts do not take into consideration any other costs of hospital stay. While the HLOS figures given above likely reflect a combination of factors (e.g., relative injury severity, the need for reversal of anticoagulation, any related procedures), they also inherently "capture" a variety of complications associated with postinjury deleterious effects of therapeutic anticoagulation. ,,
There are important limitations to this study. First, its retrospective nature and the inability to determine causal relationships must be mentioned. This is especially important when considering key associations between warfarin use, the observed head injury severity, and mortality. Second, the collected information is likely subject to reviewer bias. All of the data were obtained from an institutional trauma database, which is heavily dependent on the quality of source data. Although data entry, in general, is performed with high fidelity, it is not without errors. Our institutional policy mandates that every patient and/or the patient's family are interviewed on admission to determine known indications for anticoagulation and that the data are documented and entered into the trauma database. Again, when necessary the patient's physicians were also queried. Despite our efforts, this process was imperfect: Many patients had changed physicians along the way, and their current physicians may not have been fully aware of "how and why" warfarin therapy was initiated. Third, there is a continued lack of consensus on the absolute indications for warfarin use. We used the ACCP guidelines as the "gold standard" for appropriate versus inappropriate use of warfarin.  Admittedly, other guidelines exist that do not exactly mirror the ACCP paradigm. Fourth, our institution has a highly protocolized approach to warfarin reversal. Thus, most patients receive vitamin K, plasma transfusion, and possibly various combinations of coagulation factor concentrates as soon as any significant trauma-related hemorrhage is detected (e.g., skeletal, soft tissue, cavitary, or intracranial). For patients without obvious or major injuries, clinical observation alone is instituted, without necessarily reversing the effects of warfarin. Because of the associated clinical heterogeneity (e.g., patient with no brain injury who received warfarin reversal for severe abdominal injury versus patient who received aggressive warfarin reversal for a massive intracranial hemorrhage), an objective, well-controlled evaluation of "excess mortality" attributable to specific hemorrhagic complications would not be feasible or practical using the current dataset. Finally, this study included only trauma patients who sustained blunt injuries, and our results cannot be generalized to patients with penetrating trauma.
A significant proportion of trauma patients admitted to our institution were noted to take warfarin for inappropriate indications. Moreover, many patients taking warfarin had either subtherapeutic or supratherapeutic INR levels. Although warfarin use did not independently predict mortality, preinjury warfarin use was associated with greater mortality and HLOS in the subset of patients with severe traumatic brain injuries (e.g., MAISH ≥4). Safety initiatives directed at more appropriate initiation and management of warfarin therapy are needed. In addition, the authors encourage both providers and hospitals to institute more stringent processes for reviewing medication use appropriateness, including provisions for proactive intervention when potentially dangerous interactions and/or patterns of utilization are detected.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3]