Intracranial Bleeds After Head Injury in Warfarin Patients
Intracranial Bleeds After Head Injury in Warfarin Patients
The prevalence of intracranial bleeds after minor head injuries in patients who are not anticoagulated ranges between 5.1% and 8.0%, compared with 21.9% in this study of anticoagulated patients. The prevalence with minimal injuries, however, is not clear. The definition of minimal head injury is not widely adopted, and the injury is usually considered trivial in individuals who are not anticoagulated. We have not found studies estimating the prevalence of intracranial bleeds after minimal head injury in patients who are not anticoagulated. One in every 20 patients with minimal head injuries had intracranial bleeding in this study. Loss of consciousness and younger age were the only clinical variables associated with intracranial bleeding. In this cohort, patients with higher INR were not more likely to bleed when compared with patients with a lower INR.
Previous studies approaching this topic used different methods or patient populations, thereby making it difficult to make direct comparisons with our results. Brewer et al. performed a trauma registry review of patients on antiplatelets or anticoagulants who sustained a head injury with a GCS of 15, and also found that loss of consciousness was associated with intracranial bleeds. Although headache failed to reach statistical significance in our study, Nishijima et al. performed a multi-center prospective study on patients with recent use of either clopidogrel or warfarin who sustained a minor head injury. They found that both headache and vomiting were significantly associated with the presence of intracranial bleeds. They found a 4.3% risk of intracranial bleeding in patients on warfarin. This figure is close to our finding with minimal head injuries (4.8%), but lower than what we found for minor head injuries (i.e., associated with loss of consciousness, amnesia or confusion, and GCS 13–15). INR levels were not measured, which might have also reduced the apparent prevalence of bleeding. Rendell and Batchelor looked at CT head requests that included the terms (warfarin*) OR (anticoagula*) and found that only a GCS < 15 was significantly associated with a greater risk of bleeds. It wasn't clear whether or not patients with GCS < 13 were included in that study. Garra et al. looked at 65 patients with head injuries while on warfarin and found no intracranial bleeds. However, prothrombin times were subtherapeutic or normal in a large portion of the patients and those with loss of consciousness and amnesia were excluded, which indicates that most patients had minimal head injuries and possibly subtherapeutic or normal INRs. Gittleman et al. included 89 anticoagulated patients with head injuries of any severity who had a CT scan performed, including all GCS levels, and found that for patients with bleeds, the average GCS was 12.0. Major and Reed examined 399 admitted patients with head injury who were either on warfarin or antiplatelet treatment of all severities. Of the 89 patients on warfarin, 4 had intracranial bleeding. The study by Li et al. included 144 patients with a head injury while on warfarin who had a CT scan, and found no associations between clinical findings and intracranial bleeding. Patients with loss of consciousness were, however, excluded. Cohen and colleagues examined patients with variable injury severity as well. However, among the patients with GCS 13–15, the mortality rate was 80.6%. This figure is very high and not in keeping with other studies. Although none of the patients who had a negative initial CT scan had a repeat positive CT scan in our study, Nishijima et al. followed up 687 patients on warfarin who had an initial negative CT scan after a minor head injury and identified 4 (0.6%) with delayed intracranial hemorrhages.
We found the mean age in patients with intracranial bleed to be statistically significantly younger than patients with no intracranial bleed. It is possible that the younger patients in this sample are more ambulatory or active and as such, prone to more significant mechanisms of injury. However, data on mechanisms of injury were not sufficient to test this hypothesis.
Our outcome assessment included all adult hospitals with CT scanners in our geographical region. This assessment strategy allowed us to minimize the chance of missing positive outcomes. Data extraction was performed using a standardized data extraction form.
The nature of minimal head injuries is such that in the presence of other injuries it could be easily overlooked (e.g., patients with a hip fracture and significant pain who also had minimal head injury). For such patients, the injury might not be coded as an admission or discharge diagnosis. This may have led to missing cases with inconsequential minimal head injuries, which could have led to overestimating the prevalence of intracranial bleeds in minimal head injury. Another factor that may overestimate the risk of bleeding is that patients who sustain minimal head injuries may not seek medical care, making for a smaller denominator. Given the retrospective nature of the study, it is possible that some eligible cases have been missed due to coding errors or poor documentation. We assumed that the absence of a symptom or a sign from the nursing and physicians' notes indicated the symptom or sign was not present. Although this is generally the case, it is possible that some of the patients did have one of the symptoms or signs without having it documented in their records. The use of other anticoagulants (e.g., heparin), use of antiplatelet agents, and the presence of thrombocytopenia were not investigated. In addition, data were obtained using a single unblinded abstractor and the process was not audited for accuracy.
Discussion
The prevalence of intracranial bleeds after minor head injuries in patients who are not anticoagulated ranges between 5.1% and 8.0%, compared with 21.9% in this study of anticoagulated patients. The prevalence with minimal injuries, however, is not clear. The definition of minimal head injury is not widely adopted, and the injury is usually considered trivial in individuals who are not anticoagulated. We have not found studies estimating the prevalence of intracranial bleeds after minimal head injury in patients who are not anticoagulated. One in every 20 patients with minimal head injuries had intracranial bleeding in this study. Loss of consciousness and younger age were the only clinical variables associated with intracranial bleeding. In this cohort, patients with higher INR were not more likely to bleed when compared with patients with a lower INR.
Previous studies approaching this topic used different methods or patient populations, thereby making it difficult to make direct comparisons with our results. Brewer et al. performed a trauma registry review of patients on antiplatelets or anticoagulants who sustained a head injury with a GCS of 15, and also found that loss of consciousness was associated with intracranial bleeds. Although headache failed to reach statistical significance in our study, Nishijima et al. performed a multi-center prospective study on patients with recent use of either clopidogrel or warfarin who sustained a minor head injury. They found that both headache and vomiting were significantly associated with the presence of intracranial bleeds. They found a 4.3% risk of intracranial bleeding in patients on warfarin. This figure is close to our finding with minimal head injuries (4.8%), but lower than what we found for minor head injuries (i.e., associated with loss of consciousness, amnesia or confusion, and GCS 13–15). INR levels were not measured, which might have also reduced the apparent prevalence of bleeding. Rendell and Batchelor looked at CT head requests that included the terms (warfarin*) OR (anticoagula*) and found that only a GCS < 15 was significantly associated with a greater risk of bleeds. It wasn't clear whether or not patients with GCS < 13 were included in that study. Garra et al. looked at 65 patients with head injuries while on warfarin and found no intracranial bleeds. However, prothrombin times were subtherapeutic or normal in a large portion of the patients and those with loss of consciousness and amnesia were excluded, which indicates that most patients had minimal head injuries and possibly subtherapeutic or normal INRs. Gittleman et al. included 89 anticoagulated patients with head injuries of any severity who had a CT scan performed, including all GCS levels, and found that for patients with bleeds, the average GCS was 12.0. Major and Reed examined 399 admitted patients with head injury who were either on warfarin or antiplatelet treatment of all severities. Of the 89 patients on warfarin, 4 had intracranial bleeding. The study by Li et al. included 144 patients with a head injury while on warfarin who had a CT scan, and found no associations between clinical findings and intracranial bleeding. Patients with loss of consciousness were, however, excluded. Cohen and colleagues examined patients with variable injury severity as well. However, among the patients with GCS 13–15, the mortality rate was 80.6%. This figure is very high and not in keeping with other studies. Although none of the patients who had a negative initial CT scan had a repeat positive CT scan in our study, Nishijima et al. followed up 687 patients on warfarin who had an initial negative CT scan after a minor head injury and identified 4 (0.6%) with delayed intracranial hemorrhages.
We found the mean age in patients with intracranial bleed to be statistically significantly younger than patients with no intracranial bleed. It is possible that the younger patients in this sample are more ambulatory or active and as such, prone to more significant mechanisms of injury. However, data on mechanisms of injury were not sufficient to test this hypothesis.
Strengths
Our outcome assessment included all adult hospitals with CT scanners in our geographical region. This assessment strategy allowed us to minimize the chance of missing positive outcomes. Data extraction was performed using a standardized data extraction form.
Limitations
The nature of minimal head injuries is such that in the presence of other injuries it could be easily overlooked (e.g., patients with a hip fracture and significant pain who also had minimal head injury). For such patients, the injury might not be coded as an admission or discharge diagnosis. This may have led to missing cases with inconsequential minimal head injuries, which could have led to overestimating the prevalence of intracranial bleeds in minimal head injury. Another factor that may overestimate the risk of bleeding is that patients who sustain minimal head injuries may not seek medical care, making for a smaller denominator. Given the retrospective nature of the study, it is possible that some eligible cases have been missed due to coding errors or poor documentation. We assumed that the absence of a symptom or a sign from the nursing and physicians' notes indicated the symptom or sign was not present. Although this is generally the case, it is possible that some of the patients did have one of the symptoms or signs without having it documented in their records. The use of other anticoagulants (e.g., heparin), use of antiplatelet agents, and the presence of thrombocytopenia were not investigated. In addition, data were obtained using a single unblinded abstractor and the process was not audited for accuracy.
