- Email: [email protected]
Utilizing clinical factors to reduce head CT scan ordering for minor head trauma patients
PI1 SO7364679(97)00071-l
UTILIZING
CLlNlCAL
FACTORS TO REDUCE HEAD CT SCAN MlNOR liEAD TRAUMA PATfEW’S
Erik C. Miller, MD, James F. Holmes, Reprint Address:
and Robert W. Derlet, MD
Division of Emergency Medicine, University of California Davis Medical Center, Davis, Californra Erik C. Miller, MO, Emergency Department, Yakima Valley Memorial Hospital, 2811 Tieton Drive, Yakima. WA 98902
INTRODUCTION
Cl Abstract-Our study objective was to determine whether simple clinical criteria can be used to safely reduce the number of patients who require cranial computed tomography (CT) scan after sustaining minor head trauma. Awake patients (Glascow Coma Scale = 15) who presented to the emergency department with acute head injury associated with a loss of consciousness were evaluated for clinical predictors of head injury prior to CT scan. The studied risk factors included severe headache, nausea, vomiting, and depressed skull fracture on physical examination. Patients with no risk factors present were compared with patients with one or more risk factors with respect to abnormal CT rate and rate of operative intervention for head injury. Of the 2143 patients entered into the study, 1302 (61%) had no risk factor for head injury, whereas 841 (39%) had one or more risk factors present. A total of 138 (6.4%) of those studied had an abnormal CT scan. This number included 3.7% of those patients with no risk factors vs. 11% in patients with one or more risk factors. The CT scan abnormalities in the no-risk-factor group were not clinicaNy significant. AU 5 patients who required operative intervention had at least one of the risk factors present. The use of four simple clinical criteria in minor head trauma patients would allow a 61% reduction in the number of head CT scans performed and still identify all patients who require neurosurgical intervention and the majority of patients with an abnormal CT scan. This method could lead to a large savings in patient charges nationwide. Further studies may be helpful in confirming these findings. 0 1997 Elsevier Science Inc. Z Keywords-trauma; head trauma; phy: head CT scan; cost savings
RECEIVED: ACCEPTED:
MD,
There continues to be debate in the medical literature about which patients with head trauma require computed tomography (CT) in the emergency department (ED). The debate is driven by the need to immediately identify all patients with potentially serious injury vs. increasing demands to cut escalating medical costs. Prior studies and clinical experience in patients with altered mental status or sustained confusion after blunt head trauma have shown the need for immediate head CT scan to rule out a neurosurgical lesion (I-3). These studies have demonstrated that 20-35s of patients with a Glasglow Coma Scale (GCS) of 13 or 14 have abnormal CT scan results, and many of these patients require intervention. The work-up of patients with a normal GCS and neurologic examination after head trauma is much less clear. Previous studies have suggested that all patients with a loss of consciousness (LOC) or amnesia after head trauma should have a head CT scan performed in the ED (l-4). Recent studies, with better design, have shown that very few patients in this group require intervention, and most of the abnormalities seen on CT scan are minor. These studies have led to the possibility of a less conservative approach (5-7). Another matter that is clear from previous studies is that of all patients with minor or moderate head injury, those with a GCS =- IS represent the vast majority seen in the ED (65-X5%), and it is from this group that medical costs could be substantially re-. duced (2,6.X).
computed tomogra-
24 July 1996; FINAL SUBMISSION RECEIVED: 28 January 1997: I 1 February 1997 453
_I_...
--_II_ -_
454
E. C. Miller et al.
Table 1. Comparison of Risk Factors in Normal and Abnormal CT Scan Groups
Characteristic Severe headache Nausea Vomiting Skull depression on exam None of aboverisk factors TOTAL
Number With Normal CT Scan N = 2005 (%) 617 190 101 0 1097 2005
(31) (9) (5) (0) (55) (100%)
In a previous study, we found that there were four clinical risk factors that identified the majority of patients with abnormal CT scansafter minor head trauma (7). In the present study, we examined these risk factors to see whether we could safely determine which patients with a GCS = 15 after head trauma would require a CT scan in the ED. In addition, we attemptedto estimate the potential savings that could be made in the medical costs of treating these patients utilizing our clinical criteria.
METHODS
The study was conducted at a large, urban, Level 1 trauma center from October 1993 to October 1995. The trauma protocol at our institution required patients who have a history of LOC or amnesia after head trauma to have a head CT scan performed, even when they have a normal mental status in the ED. For the study, LOC was defined as a witness viewing the patient in a state of unconsciousnessand reporting this fact to emergency medical service personnel or the senior emergency resident or attending physician in charge of the patient’s care. Amnesia was defined as a patient being unable to remember or describe the incident that led to the head trauma in the history directly taken by the examing physician. All patients with a normal mental status, a history of LOC or amnesia,and who presentedless than 24 h after blunt headtrauma were eligible for enrollment. Patients were not excluded for drug or alcohol use, multisystem trauma, or use of anticoagulants or known coagulopathy as long as mental status was normal and GCS was 15. The patients were enrolled prospectively and consecutively. To be included in the study, patients must have had their injury less than 2 h prior to presenting to the ED. All patients had their CT scan performed within 8 h of presentation to the ED. A data sheet was completed on all patients prior to the CT scan by the senior emergency resident or attending physician in charge of the patient’s care. Normal mental status was defined as a GCS = 15 in adults and age-appropriatebehavior in children after
Number With Abnormal CT Scan N = 138 (%) 75 28 21 3 11 138
p-Value <.OOOl <.OOOl <.OOOl <.OOOl
(55) (20) (15) (2)
03) (100%)
arrival in the ED. The four key risk factors for head injury were severeheadache,nausea,vomiting, and skull depressionon examination. Severeheadachewas defined by comparison to mild or moderate headachesduring questioning of the patient. All patients were askedabout the subjective feeling of nausea,and a yes/no answer was recorded. Vomiting was recorded if witnessed prior to ED presentation or in the ED. Skull depression was defined as a palpable defect in the outer table of the skull on examination by a resident or attending physician. All CT scanswere performed on a model TCT 900s scannerfrom Toshiba America Medical Systems(Tustin, CA). Bone, brain, and epidural windows were obtained from all patients. The attending neuroradiologist’s final reading was used as the official result. There were several radiologists who read the CT scans,and no attempt was made to compare interrater reliability. An abnormal CT scan was defined as one showing an acute traumatic intracranial lesion (contusion, parenchymal hematoma, epidural hematoma, subdural hematoma, subarachnoid hemorrhage)or a skull fracture. Patientswith none of the key risk factors studied for head injury were compared with patients with one or more risk factors present with respect to the incidence of abnormal CT scans and the need for neurosurgical intervention. Neurosurgical intervention was defined as craniotomy to repair an acute traumatic injury or placement of a monitoring bolt. All patients with abnormal CT scanswere followed to hospital discharge to determine whether any intervention or change in mental statusoccurred. Data were analyzed with chi-square tests. Patients were not discharged until they had been monitored in the ED a total of 3 h from initial presentation.
RESULTS
A total of 2143 patients were enrolled in the study over the 2-yr time period. There were 138 patients (6.4%) who had an abnormal CT scan and only 5 patients (0.2%) who required neurosurgical intervention. Table 1 compares patients with abnormal and
-155
CT Scan for Minor Head Trauma Table 2. Abnormal CT Scan and Neuroswgkal
Abnormal CT scans Neurosurgical intervention
Rates in Patients With and WwIout Ctidcal -
Patients With No Risk Factors, n = 1302
Patients With One Or More Risk Factor, n = 641
46 (3.7%) 0 (0)
90 (11%) 5 (0.6%)
normal CT scans with respect to the four key risk factors studied. Table 1 shows that the risk factors are present significantly more often in patients with abnormal CT scans. Applying the four studied criteria to the patients revealed that 1302 (61%) had none of the key risk factors present, whereas 841 (39%) had at least one of the risk factors present. Table 2 shows a comparison of these two groups with respect to rate of abnormal CT scans and rate of neurosurgical intervention. Patients with no clinical risk factor had a significantly lower rate of abnormal CT scans, and no patient in this group required surgical intervention (95% confidence interval = O-0.03%). In addition, those patients with risk factors present tended to have more potentially serious CT scan abnormalities than patients without a risk factor (Table 3). All 5 patients who required neurosurgical intervention had at least one risk factor present, including 3 with depressed skull fracture on examination, 1 with severe headache, and 1 with both nausea and severe headache. The 3 patients with skull depression underwent elevation of a depressed skull fracture. The patient with a severe headache had an epidural hematoma that was evacuated. The final patient had multiple contusions on initial CT scan and required partial lobectomy during prolonged hospitalization for massive cerebral edema. Utilizing the four criteria for head injury would have lead to a 65% sensitivity and 63% specificity in detecting patients with an abnormal CT scan. There would have been a 100% sensitivity, however, for detecting patients who required surgical intervention.
Table 3. Dlatdbuth
Injury --p-Value x: .Ooool c; .Ooool _ _--__-
DISCUSSION Minor head trauma is an extremely common entity seen in most EDs, yet there are widely varying recommendations on the care of these patients (l-9). Many trauma centers obtain head CT scans on all patients with LOC or amnesia after head trauma, regardless of the level of consciousness in the ED (l-3). At our institution, this practice was very time consuming and could lead to a 3-4-h delay for other patients needing urgent CT scans. This approach is also very costly because the CT scan is still relatively expensive and the clinical management of these patients is almost never changed by the result of the CT scan. This problem has led to a search for clinical variables that would allow physicians to screen patients for the necessity of an emergent CT scan. This search has met with limited success. Reinus et al. (8) proposed four clinical variables that could decrease head CT scans by 60%. The variables proposed (abnormal neurological examination, intoxication, history of amnesia or focal neurological deficit) had a sensitivity of 90% and a specificity of 66% for detecting patients with an abnormal CT scan and detected all patients requiring operative intervention. However, there were only 373 patients in the study, which limited its conclusions, and the investigators reported that they had insufficient data to perform all the calculations. Madden et al. (9) described 10 high-yield variables to screen patients for the need for CT scans. This method led to a 96% sensitivity and 21% specificity for detecting abnormal CT scans. However, only 16% of CT scans were 10 found unnecessary. a small reward for employinp
atAbnormalCTScanrlnPatientsWithandWithoutCHnicPlRiskFactors
CT Scan Abnormality
Abnormal CT Scan Findings in Patients With No Clinical Risk Factors
Contusion/SAH/IVH Epidural Hernatoma Subdural Hematoma Skull Fracture only Total
36 0 3 7 46
SAH = sub-arachnoid
oi lksd
hemorrhage,
IVH = intraventricular
hemorrhage.
Abnormal CT Scan Findings in Patients With One or More Clinical Risk Factor 40 4 24 22 90 .- _.--_
E. C. Miller et al.
456 criteria. Borczuk (6) described four variables (“cranial trauma,” focal neurological deficit, signs of basilar skull fracture, age > 60) to lower CT scan requests (6). These variables had a sensitivity of 92% and a specificity of 46% for abnormal CT scans. This method reduced the number CT scans by only 21%, however, and the presence of “cranial trauma” was not completely defined. In the present study, we used four simple clinical variables that are fairly objective and quickly assessible. The use of these four variables had a sensitivity of 65% for detecting patients with an abnormal CT scan, which some clinicians might feel is too low. However, this sensitivity is somewhat misleading. All patients who required neurosurgical intervention were detected with these clinical risk factors (95% confidence interval = O-0.03%). Thus, the risk of missing someone with a surgically correctable lesion by using our criteria is 1 in 3333. In addition, utilization of these variables identified 90% of patients with an extra-axial hematoma and 75% of patients with a skull fracture, those patients most likely to require surgical intervention. Some clinicians may also feel uncomfortable with the fact that 41 of the 1302 patients (3%) without risk factors had an intracranial lesion identified on CT scan because they may feel that it is important to identify and treat all patients with abnormal CT scans regardless of the need for surgery. We do not feel this is necessary for the following reasons. First, none of these 41 patients had any change in mental status or required any intervention during hospital observation (2 d average). Second, CT scan is a relatively poor instrument to detect minor brain injury; Levin et al. (lo), in a study of 50 patients with mild head injury, found that 25% had an abnormal CT scan vs. 80% with abnormalities on magnetic resonance imaging (MRI). Third, an abnormal CT scan is a poor predictor of eventual neurologic outcome. In the study by Levin et al., both CT and MRI abnormalities did not accurately predict which patients would have persistent neurologic impairment. Eide and Tysnes (11) compared patients with concussion and normal CT scans with patients with concussion and focal contusions on CT scans with respect to late neurologic impairment (11). They found very poor correlation with CT scan abnormality and increased neurologic problems; equal numbers in both groups had persistent headache, dizziness, sleep problems, etc. Only speech and memory problems were higher in the group with abnormal CT scans and then
only in those patients with multiple contusions on CT scan. Thus, instead of trying to identify all patients with abnormal CT scans, it would better serve the patient to make sure that follow-up is available for any person with minor head trauma and persistent neurologic problems regardless of CT scan findings. We achieved a specificity of 63% with these criteria, which is much higher than most of the previous studies. This specificity could reduce the number of head CT scans requested in minor trauma patients by 61% in comparison with a protocol to obtain a head CT scan on all patients with LOC or amnesia, something our institution and many others currently practice. During the 2-yr study period, we could have reduced the number of requested CT scans by 1302. This reduction would have led to a decrease in patient charges of $1.7 million for minor head trauma care at our institution alone. It has been estimated that 400,000-450,000 patients are seen each year in the nation’s EDs for head trauma (12,13). The majority of these patients present with a GCS of 15, and one study found that 80% of all head injured patients in a year’s time presented with a GCS of 15 (14). Using a conservative figure of 65%, 275,000 patients with head trauma present with a GCS of 15. If our criteria were utilized nationwide, this method could result in a reduction of up to 61% (167,750) of the requested head CT scans, resulting in enormous cost savings.
CONCLUSIONS We prospectively determined that four simple clinical variables (nausea, vomiting, severe headache, skull depression) can be utilized to determine which patients with minor head trauma and a GCS of 15 require head CT scan. These clinical variables safely identified all patients who required neurosurgical intervention and the vast majority of patients with potentially serious intracranial abnormalities. The variables are objective, easily remembered, and quickly assessible in ED patients. Use of these clinical predictors can reduce the number of necessary CT scans by up to 61%, potentially leading to a large reduction in patient charges and use of the emergency CT scanner. Prospective studies using these criteria may be useful to confirm these findings.
REFERENCES 1. Harad FT, Kerstein MD. Inadequacy of bedside clinical indicators in identifying significant intracranial injury in trauma patients. J Trauma. 1992;32:359-63.
2. Stein SC, Ross SE. Mild head injury: A plea for routine early CT scanning. J Trauma. 1992,33:1 l-3. 3. Shackford SR, Wald SL, Ross SE, et al. The clinical utility of
CT Scan for Minor Head Trauma computed tomographic scanning and neurologic examination in the managementof patients with minor head injuries. J Trauma. 1992;33:385-94. 4. Stein SC, Ross SE. The value of computed tomographic scansin patients with low-risk head injuries. Neurosurgery. 1990;26: 638-40.
5. Jeret JS, Mandell M, Anziska B, Lipitz M, Vilceus AP, Ware JA, Zesiewicz TA. Clinical predictors of abnormality disclosed by computed tomography after mild head trauma. Neurosurgery. 1993;32:9-15. 6. Borczuk P. Predictors of intracranial injury in patients with mild head trauma. Ann Emerg Med. 1995;25:731-6. 7. Miller EC, Derlet RW, Kinser D. Minor head trauma: Is computed tomography always necessary?Ann Emerg Med. 1996;27:290-4. 8. Reinus WR, Wippold FJ, Erickson KK. Practical selection criteria for noncontrastcranial computedtomography in patients with head trauma. Ann Emerg Med. 1993;22:1148-55. 9. Madden C. Witzke DB, Sanders AB, Valente J, Fritz M. High-
457 yield selectioncriteria for cranial computedtomography after acute trauma. Acad Emerg Med. 1995;2:248-52. 10. Levin HS, Williams DH, Eisenberg HM, High WM Jr. Guinto FC Jr. Serial MRI and neurobehavioural findings after mild to moderate closed head injury. J Neurol Neurosurg Psychiatr. 1992;55: 255-62.
11. Eide PK, Tysnes OB. Early and late outcome in head Injury patients with radiological evidence of brain damage. Acta Neurol Stand. 1992;86:194-8. 12. FederleMP, Brant-Zawadzki M. Trauma: CT in emergency.Diagn Imaging. 1982;51:34-8. 13. Kraus JF. Epidemiology of head injury. In: Cooper PR, ed. Head injury, 2nd ed. Baltimore: Williams & Wilkins; 1987:l-19. 14. Schynoll W, Overton D, Krome R, Wesolowski D, Wang AM, Wilson A, Coffey M. A prospective study to identify high-yield criteria associatedwith acute intracranial computed tomography findings in head-injured patients. Am J Emerg Med. 1993:ll: 321-6.
UTILIZING
CLlNlCAL
FACTORS TO REDUCE HEAD CT SCAN MlNOR liEAD TRAUMA PATfEW’S
Erik C. Miller, MD, James F. Holmes, Reprint Address:
and Robert W. Derlet, MD
Division of Emergency Medicine, University of California Davis Medical Center, Davis, Californra Erik C. Miller, MO, Emergency Department, Yakima Valley Memorial Hospital, 2811 Tieton Drive, Yakima. WA 98902
INTRODUCTION
Cl Abstract-Our study objective was to determine whether simple clinical criteria can be used to safely reduce the number of patients who require cranial computed tomography (CT) scan after sustaining minor head trauma. Awake patients (Glascow Coma Scale = 15) who presented to the emergency department with acute head injury associated with a loss of consciousness were evaluated for clinical predictors of head injury prior to CT scan. The studied risk factors included severe headache, nausea, vomiting, and depressed skull fracture on physical examination. Patients with no risk factors present were compared with patients with one or more risk factors with respect to abnormal CT rate and rate of operative intervention for head injury. Of the 2143 patients entered into the study, 1302 (61%) had no risk factor for head injury, whereas 841 (39%) had one or more risk factors present. A total of 138 (6.4%) of those studied had an abnormal CT scan. This number included 3.7% of those patients with no risk factors vs. 11% in patients with one or more risk factors. The CT scan abnormalities in the no-risk-factor group were not clinicaNy significant. AU 5 patients who required operative intervention had at least one of the risk factors present. The use of four simple clinical criteria in minor head trauma patients would allow a 61% reduction in the number of head CT scans performed and still identify all patients who require neurosurgical intervention and the majority of patients with an abnormal CT scan. This method could lead to a large savings in patient charges nationwide. Further studies may be helpful in confirming these findings. 0 1997 Elsevier Science Inc. Z Keywords-trauma; head trauma; phy: head CT scan; cost savings
RECEIVED: ACCEPTED:
MD,
There continues to be debate in the medical literature about which patients with head trauma require computed tomography (CT) in the emergency department (ED). The debate is driven by the need to immediately identify all patients with potentially serious injury vs. increasing demands to cut escalating medical costs. Prior studies and clinical experience in patients with altered mental status or sustained confusion after blunt head trauma have shown the need for immediate head CT scan to rule out a neurosurgical lesion (I-3). These studies have demonstrated that 20-35s of patients with a Glasglow Coma Scale (GCS) of 13 or 14 have abnormal CT scan results, and many of these patients require intervention. The work-up of patients with a normal GCS and neurologic examination after head trauma is much less clear. Previous studies have suggested that all patients with a loss of consciousness (LOC) or amnesia after head trauma should have a head CT scan performed in the ED (l-4). Recent studies, with better design, have shown that very few patients in this group require intervention, and most of the abnormalities seen on CT scan are minor. These studies have led to the possibility of a less conservative approach (5-7). Another matter that is clear from previous studies is that of all patients with minor or moderate head injury, those with a GCS =- IS represent the vast majority seen in the ED (65-X5%), and it is from this group that medical costs could be substantially re-. duced (2,6.X).
computed tomogra-
24 July 1996; FINAL SUBMISSION RECEIVED: 28 January 1997: I 1 February 1997 453
_I_...
--_II_ -_
454
E. C. Miller et al.
Table 1. Comparison of Risk Factors in Normal and Abnormal CT Scan Groups
Characteristic Severe headache Nausea Vomiting Skull depression on exam None of aboverisk factors TOTAL
Number With Normal CT Scan N = 2005 (%) 617 190 101 0 1097 2005
(31) (9) (5) (0) (55) (100%)
In a previous study, we found that there were four clinical risk factors that identified the majority of patients with abnormal CT scansafter minor head trauma (7). In the present study, we examined these risk factors to see whether we could safely determine which patients with a GCS = 15 after head trauma would require a CT scan in the ED. In addition, we attemptedto estimate the potential savings that could be made in the medical costs of treating these patients utilizing our clinical criteria.
METHODS
The study was conducted at a large, urban, Level 1 trauma center from October 1993 to October 1995. The trauma protocol at our institution required patients who have a history of LOC or amnesia after head trauma to have a head CT scan performed, even when they have a normal mental status in the ED. For the study, LOC was defined as a witness viewing the patient in a state of unconsciousnessand reporting this fact to emergency medical service personnel or the senior emergency resident or attending physician in charge of the patient’s care. Amnesia was defined as a patient being unable to remember or describe the incident that led to the head trauma in the history directly taken by the examing physician. All patients with a normal mental status, a history of LOC or amnesia,and who presentedless than 24 h after blunt headtrauma were eligible for enrollment. Patients were not excluded for drug or alcohol use, multisystem trauma, or use of anticoagulants or known coagulopathy as long as mental status was normal and GCS was 15. The patients were enrolled prospectively and consecutively. To be included in the study, patients must have had their injury less than 2 h prior to presenting to the ED. All patients had their CT scan performed within 8 h of presentation to the ED. A data sheet was completed on all patients prior to the CT scan by the senior emergency resident or attending physician in charge of the patient’s care. Normal mental status was defined as a GCS = 15 in adults and age-appropriatebehavior in children after
Number With Abnormal CT Scan N = 138 (%) 75 28 21 3 11 138
p-Value <.OOOl <.OOOl <.OOOl <.OOOl
(55) (20) (15) (2)
03) (100%)
arrival in the ED. The four key risk factors for head injury were severeheadache,nausea,vomiting, and skull depressionon examination. Severeheadachewas defined by comparison to mild or moderate headachesduring questioning of the patient. All patients were askedabout the subjective feeling of nausea,and a yes/no answer was recorded. Vomiting was recorded if witnessed prior to ED presentation or in the ED. Skull depression was defined as a palpable defect in the outer table of the skull on examination by a resident or attending physician. All CT scanswere performed on a model TCT 900s scannerfrom Toshiba America Medical Systems(Tustin, CA). Bone, brain, and epidural windows were obtained from all patients. The attending neuroradiologist’s final reading was used as the official result. There were several radiologists who read the CT scans,and no attempt was made to compare interrater reliability. An abnormal CT scan was defined as one showing an acute traumatic intracranial lesion (contusion, parenchymal hematoma, epidural hematoma, subdural hematoma, subarachnoid hemorrhage)or a skull fracture. Patientswith none of the key risk factors studied for head injury were compared with patients with one or more risk factors present with respect to the incidence of abnormal CT scans and the need for neurosurgical intervention. Neurosurgical intervention was defined as craniotomy to repair an acute traumatic injury or placement of a monitoring bolt. All patients with abnormal CT scanswere followed to hospital discharge to determine whether any intervention or change in mental statusoccurred. Data were analyzed with chi-square tests. Patients were not discharged until they had been monitored in the ED a total of 3 h from initial presentation.
RESULTS
A total of 2143 patients were enrolled in the study over the 2-yr time period. There were 138 patients (6.4%) who had an abnormal CT scan and only 5 patients (0.2%) who required neurosurgical intervention. Table 1 compares patients with abnormal and
-155
CT Scan for Minor Head Trauma Table 2. Abnormal CT Scan and Neuroswgkal
Abnormal CT scans Neurosurgical intervention
Rates in Patients With and WwIout Ctidcal -
Patients With No Risk Factors, n = 1302
Patients With One Or More Risk Factor, n = 641
46 (3.7%) 0 (0)
90 (11%) 5 (0.6%)
normal CT scans with respect to the four key risk factors studied. Table 1 shows that the risk factors are present significantly more often in patients with abnormal CT scans. Applying the four studied criteria to the patients revealed that 1302 (61%) had none of the key risk factors present, whereas 841 (39%) had at least one of the risk factors present. Table 2 shows a comparison of these two groups with respect to rate of abnormal CT scans and rate of neurosurgical intervention. Patients with no clinical risk factor had a significantly lower rate of abnormal CT scans, and no patient in this group required surgical intervention (95% confidence interval = O-0.03%). In addition, those patients with risk factors present tended to have more potentially serious CT scan abnormalities than patients without a risk factor (Table 3). All 5 patients who required neurosurgical intervention had at least one risk factor present, including 3 with depressed skull fracture on examination, 1 with severe headache, and 1 with both nausea and severe headache. The 3 patients with skull depression underwent elevation of a depressed skull fracture. The patient with a severe headache had an epidural hematoma that was evacuated. The final patient had multiple contusions on initial CT scan and required partial lobectomy during prolonged hospitalization for massive cerebral edema. Utilizing the four criteria for head injury would have lead to a 65% sensitivity and 63% specificity in detecting patients with an abnormal CT scan. There would have been a 100% sensitivity, however, for detecting patients who required surgical intervention.
Table 3. Dlatdbuth
Injury --p-Value x: .Ooool c; .Ooool _ _--__-
DISCUSSION Minor head trauma is an extremely common entity seen in most EDs, yet there are widely varying recommendations on the care of these patients (l-9). Many trauma centers obtain head CT scans on all patients with LOC or amnesia after head trauma, regardless of the level of consciousness in the ED (l-3). At our institution, this practice was very time consuming and could lead to a 3-4-h delay for other patients needing urgent CT scans. This approach is also very costly because the CT scan is still relatively expensive and the clinical management of these patients is almost never changed by the result of the CT scan. This problem has led to a search for clinical variables that would allow physicians to screen patients for the necessity of an emergent CT scan. This search has met with limited success. Reinus et al. (8) proposed four clinical variables that could decrease head CT scans by 60%. The variables proposed (abnormal neurological examination, intoxication, history of amnesia or focal neurological deficit) had a sensitivity of 90% and a specificity of 66% for detecting patients with an abnormal CT scan and detected all patients requiring operative intervention. However, there were only 373 patients in the study, which limited its conclusions, and the investigators reported that they had insufficient data to perform all the calculations. Madden et al. (9) described 10 high-yield variables to screen patients for the need for CT scans. This method led to a 96% sensitivity and 21% specificity for detecting abnormal CT scans. However, only 16% of CT scans were 10 found unnecessary. a small reward for employinp
atAbnormalCTScanrlnPatientsWithandWithoutCHnicPlRiskFactors
CT Scan Abnormality
Abnormal CT Scan Findings in Patients With No Clinical Risk Factors
Contusion/SAH/IVH Epidural Hernatoma Subdural Hematoma Skull Fracture only Total
36 0 3 7 46
SAH = sub-arachnoid
oi lksd
hemorrhage,
IVH = intraventricular
hemorrhage.
Abnormal CT Scan Findings in Patients With One or More Clinical Risk Factor 40 4 24 22 90 .- _.--_
E. C. Miller et al.
456 criteria. Borczuk (6) described four variables (“cranial trauma,” focal neurological deficit, signs of basilar skull fracture, age > 60) to lower CT scan requests (6). These variables had a sensitivity of 92% and a specificity of 46% for abnormal CT scans. This method reduced the number CT scans by only 21%, however, and the presence of “cranial trauma” was not completely defined. In the present study, we used four simple clinical variables that are fairly objective and quickly assessible. The use of these four variables had a sensitivity of 65% for detecting patients with an abnormal CT scan, which some clinicians might feel is too low. However, this sensitivity is somewhat misleading. All patients who required neurosurgical intervention were detected with these clinical risk factors (95% confidence interval = O-0.03%). Thus, the risk of missing someone with a surgically correctable lesion by using our criteria is 1 in 3333. In addition, utilization of these variables identified 90% of patients with an extra-axial hematoma and 75% of patients with a skull fracture, those patients most likely to require surgical intervention. Some clinicians may also feel uncomfortable with the fact that 41 of the 1302 patients (3%) without risk factors had an intracranial lesion identified on CT scan because they may feel that it is important to identify and treat all patients with abnormal CT scans regardless of the need for surgery. We do not feel this is necessary for the following reasons. First, none of these 41 patients had any change in mental status or required any intervention during hospital observation (2 d average). Second, CT scan is a relatively poor instrument to detect minor brain injury; Levin et al. (lo), in a study of 50 patients with mild head injury, found that 25% had an abnormal CT scan vs. 80% with abnormalities on magnetic resonance imaging (MRI). Third, an abnormal CT scan is a poor predictor of eventual neurologic outcome. In the study by Levin et al., both CT and MRI abnormalities did not accurately predict which patients would have persistent neurologic impairment. Eide and Tysnes (11) compared patients with concussion and normal CT scans with patients with concussion and focal contusions on CT scans with respect to late neurologic impairment (11). They found very poor correlation with CT scan abnormality and increased neurologic problems; equal numbers in both groups had persistent headache, dizziness, sleep problems, etc. Only speech and memory problems were higher in the group with abnormal CT scans and then
only in those patients with multiple contusions on CT scan. Thus, instead of trying to identify all patients with abnormal CT scans, it would better serve the patient to make sure that follow-up is available for any person with minor head trauma and persistent neurologic problems regardless of CT scan findings. We achieved a specificity of 63% with these criteria, which is much higher than most of the previous studies. This specificity could reduce the number of head CT scans requested in minor trauma patients by 61% in comparison with a protocol to obtain a head CT scan on all patients with LOC or amnesia, something our institution and many others currently practice. During the 2-yr study period, we could have reduced the number of requested CT scans by 1302. This reduction would have led to a decrease in patient charges of $1.7 million for minor head trauma care at our institution alone. It has been estimated that 400,000-450,000 patients are seen each year in the nation’s EDs for head trauma (12,13). The majority of these patients present with a GCS of 15, and one study found that 80% of all head injured patients in a year’s time presented with a GCS of 15 (14). Using a conservative figure of 65%, 275,000 patients with head trauma present with a GCS of 15. If our criteria were utilized nationwide, this method could result in a reduction of up to 61% (167,750) of the requested head CT scans, resulting in enormous cost savings.
CONCLUSIONS We prospectively determined that four simple clinical variables (nausea, vomiting, severe headache, skull depression) can be utilized to determine which patients with minor head trauma and a GCS of 15 require head CT scan. These clinical variables safely identified all patients who required neurosurgical intervention and the vast majority of patients with potentially serious intracranial abnormalities. The variables are objective, easily remembered, and quickly assessible in ED patients. Use of these clinical predictors can reduce the number of necessary CT scans by up to 61%, potentially leading to a large reduction in patient charges and use of the emergency CT scanner. Prospective studies using these criteria may be useful to confirm these findings.
REFERENCES 1. Harad FT, Kerstein MD. Inadequacy of bedside clinical indicators in identifying significant intracranial injury in trauma patients. J Trauma. 1992;32:359-63.
2. Stein SC, Ross SE. Mild head injury: A plea for routine early CT scanning. J Trauma. 1992,33:1 l-3. 3. Shackford SR, Wald SL, Ross SE, et al. The clinical utility of
CT Scan for Minor Head Trauma computed tomographic scanning and neurologic examination in the managementof patients with minor head injuries. J Trauma. 1992;33:385-94. 4. Stein SC, Ross SE. The value of computed tomographic scansin patients with low-risk head injuries. Neurosurgery. 1990;26: 638-40.
5. Jeret JS, Mandell M, Anziska B, Lipitz M, Vilceus AP, Ware JA, Zesiewicz TA. Clinical predictors of abnormality disclosed by computed tomography after mild head trauma. Neurosurgery. 1993;32:9-15. 6. Borczuk P. Predictors of intracranial injury in patients with mild head trauma. Ann Emerg Med. 1995;25:731-6. 7. Miller EC, Derlet RW, Kinser D. Minor head trauma: Is computed tomography always necessary?Ann Emerg Med. 1996;27:290-4. 8. Reinus WR, Wippold FJ, Erickson KK. Practical selection criteria for noncontrastcranial computedtomography in patients with head trauma. Ann Emerg Med. 1993;22:1148-55. 9. Madden C. Witzke DB, Sanders AB, Valente J, Fritz M. High-
457 yield selectioncriteria for cranial computedtomography after acute trauma. Acad Emerg Med. 1995;2:248-52. 10. Levin HS, Williams DH, Eisenberg HM, High WM Jr. Guinto FC Jr. Serial MRI and neurobehavioural findings after mild to moderate closed head injury. J Neurol Neurosurg Psychiatr. 1992;55: 255-62.
11. Eide PK, Tysnes OB. Early and late outcome in head Injury patients with radiological evidence of brain damage. Acta Neurol Stand. 1992;86:194-8. 12. FederleMP, Brant-Zawadzki M. Trauma: CT in emergency.Diagn Imaging. 1982;51:34-8. 13. Kraus JF. Epidemiology of head injury. In: Cooper PR, ed. Head injury, 2nd ed. Baltimore: Williams & Wilkins; 1987:l-19. 14. Schynoll W, Overton D, Krome R, Wesolowski D, Wang AM, Wilson A, Coffey M. A prospective study to identify high-yield criteria associatedwith acute intracranial computed tomography findings in head-injured patients. Am J Emerg Med. 1993:ll: 321-6.