|
|
| |||||||||||||||||||||||||||||||||||||||||||||||
Background Computed tomography (CT) is widely used as a screening test in patients with minor head injury, although the results are often normal. We performed a study to develop and validate a set of clinical criteria that could be used to identify patients with minor head injury who do not need to undergo CT.
Methods In the first phase of the study, we recorded clinical findings in 520 consecutive patients with minor head injury who had a normal score on the Glasgow Coma Scale and normal findings on a brief neurologic examination; the patients then underwent CT. Using recursive partitioning, we derived a set of criteria to identify all patients who had abnormalities on CT scanning. In the second phase, the sensitivity and specificity of the criteria for predicting a positive scan were evaluated in a group of 909 patients.
Results Of the 520 patients in the first phase, 36 (6.9 percent) had positive scans. All patients with positive CT scans had one or more of seven findings: headache, vomiting, an age over 60 years, drug or alcohol intoxication, deficits in short-term memory, physical evidence of trauma above the clavicles, and seizure. Among the 909 patients in the second phase, 57 (6.3 percent) had positive scans. In this group of patients, the sensitivity of the seven findings combined was 100 percent (95 percent confidence interval, 95 to 100 percent). All patients with positive CT scans had at least one of the findings.
Conclusions For the evaluation of patients with minor head injury, the use of CT can be safely limited to those who have certain clinical findings.
The use of clinical findings as predictors of intracranial lesions in patients with minor head injury has been evaluated in several studies.4,5,7 In two studies, selective use of CT on the basis of clinical findings identified 96 percent and 98 percent of patients with abnormalities on CT scanning,5,7 and none of the patients with abnormalities who did not have the specified clinical findings required neurosurgery.4,5 However, some physicians are not willing to accept the risk of missing an abnormality. In a survey of emergency physicians, more than half insisted that a clinical decision rule for minor head injury must have a sensitivity of 100 percent.8 Thus, the use of CT to screen patients with minor head injury for intracranial lesions has become routine, but such screening is expensive. According to one estimate, even a 10 percent reduction in the number of CT scans in patients with minor head injury would save more than $20 million per year.9 We conducted a study to derive and validate a set of clinical criteria that could be used to identify patients with minor head injury in whom CT could be forgone. The study was conducted in two phases at a large, inner-city, level 1 trauma center from December 1997 to June 1999.
Methods
Phase 1
In phase 1 of the study, 520 consecutive patients who had minor head injury, who were at least three years old, and who presented within 24 hours after the injury were prospectively evaluated to determine which clinical findings identified patients with positive findings on CT studies of the head. In phase 2, the set of findings found to have predictive value in phase 1 was prospectively evaluated in a separate group of 909 consecutive patients with minor head injury who were similar to the patients in phase 1 with respect to age and the interval between injury and presentation. Since all patients with minor head injury and loss of consciousness or amnesia for the traumatic event, regardless of other symptoms or signs, routinely undergo CT at our center, no additional CT scans were ordered for the purpose of the study. The institutional review board approved the study and waived the requirement to obtain informed consent because we planned to record data ordinarily obtained in the evaluation of patients with minor head injury.
Minor head injury was defined as loss of consciousness in patients with normal findings on a brief neurologic examination (normal cranial nerves and normal strength and sensation in the arms and legs) and a score of 15 on the Glasgow Coma Scale, as determined by a physician on the patient's arrival at the emergency department. Advanced emergency-medicine residents, under faculty supervision, performed the initial evaluation of all patients, including the determination of whether the patient had lost consciousness and of the score on the Glasgow Coma Scale and the neurologic examination. Patients were considered to have lost consciousness if a witness or the patient reported loss of consciousness by the patient or if the patient could not remember the traumatic event. Patients with isolated deficits in short-term memory and an otherwise normal score on the Glasgow Coma Scale were considered to have a normal score on the scale. All patients included in the study underwent CT scanning. Patients who declined CT, had concurrent injuries that precluded the use of CT, or reported no loss of consciousness or amnesia for the traumatic event were excluded from the study. The CT scan was considered to be positive if it showed the presence of an acute traumatic intracranial lesion (a subdural, epidural, or parenchymal hematoma; subarachnoid hemorrhage; cerebral contusion; or depressed skull fracture). All patients with positive CT scans were admitted by the neurosurgical team and were followed until discharge in order to document any neurosurgical interventions.
All questionnaires and clinical assessments were completed before the CT studies were performed. All physicians participating in the study received an explanation of the questionnaire, including the definition of each criterion. The items included in the phase 1 questionnaire, which were selected after an extensive review of the literature on minor head injury, included age and the presence or absence of headache, vomiting, drug or alcohol intoxication, deficits in short-term memory, post-traumatic seizure, history of coagulopathy, and physical evidence of trauma above the clavicles.
Headache was defined as any head pain, whether diffuse or local. Vomiting was defined as any emesis after the traumatic event. Drug or alcohol intoxication was determined on the basis of the history obtained from the patient or a witness and suggestive findings on physical examination, such as slurred speech or the odor of alcohol on the breath. Measurements of blood alcohol and toxicologic tests were ordered at the discretion of the physician. A deficit in short-term memory was defined as persistent anterograde amnesia in a patient with an otherwise normal score on the Glasgow Coma Scale. Physical evidence of trauma above the clavicles was defined as any external evidence of injury, including contusions, abrasions, lacerations, deformities, and signs of facial or skull fracture. Seizure was defined as a suspected or witnessed seizure after the traumatic event. Coagulopathy was defined as a history of bleeding or a clotting disorder or current treatment with warfarin. We did not ask about aspirin therapy or order studies of coagulation factors or platelet counts. Questionnaire responses, examination data, and CT results from phase 1 were entered into a data base. The patients who had normal CT findings were compared with those who had abnormalities on CT. Chi-square analysis and likelihood ratios were determined for each criterion, and sensitivity and specificity were calculated for the best combinations of criteria. Chi-square recursive-partitioning analysis was performed with the use of SPSS software, version 8.0,10 in order to select a set of criteria that identified all patients with positive CT scans.
Phase 2
In phase 2 of the study, the seven findings that had predictive value in phase 1 were prospectively validated in a separate group of 909 consecutive patients with minor head injury. The phase 2 questionnaire, also completed before CT was performed, included items on headache, vomiting, age, drug or alcohol intoxication, deficits in short-term memory, physical evidence of trauma above the clavicles, and seizure. The patients were separated into two groups: those who had at least one of the seven findings, and those who had none. The frequency of positive CT scans was determined for each group and then entered in a two-by-two table. The sensitivity, specificity, and negative predictive value of the criteria, with 95 percent confidence intervals, were then calculated.
Interpretation of CT Scans
In both phases of the study, the CT scanning was performed with the use of a Somatom Plus 4 or Somatom Plus S scanner (Siemens), and the results were interpreted by staff neuroradiologists. An independent staff radiologist who was unaware of the original interpretations reviewed 50 randomly selected CT scans. Agreement between the two sets of readings was analyzed with the use of Cohen's kappa test and SPSS software, version 8.0.10 There was agreement between the two interpretations for 49 scans (98 percent, 
=0.94). The finding on one scan was interpreted as a contusion or an infarct initially and as an artifact by the independent radiologist. To determine the reproducibility of the clinical data, 50 patients were interviewed and examined by a second physician at the time of the initial evaluation. There was agreement between the two sets of evaluations for 46 patients (92 percent, =0.78).
Results
The mean age of the 520 patients in phase 1 was 36 years (range, 3 to 97), and 65 percent were male. Thirty-six patients (6.9 percent; 95 percent confidence interval, 4.2 to 9.6 percent) had positive CT scans (Table 1). Three clinical findings (deficits in short-term memory, drug or alcohol intoxication, and physical evidence of trauma above the clavicles) were significantly associated with a positive CT scan. If patients without the combination of these three findings had not undergone CT scanning, the number of scans would have been reduced by 31 percent, although abnormalities would have been missed in two patients (sensitivity, 94 percent) (Table 2). Recursive-partitioning analysis yielded a set of seven findings that identified all patients with positive CT scans: headache, vomiting, an age over 60 years, drug or alcohol intoxication, deficits in short-term memory, physical evidence of trauma above the clavicles, and seizure (Figure 1).
|
|
|
|
|
Approximately two thirds of patients with head trauma in the United States are classified as having minor head injury11; less than 10 percent of patients with minor head injury have positive findings on CT scanning, and less than 1 percent require neurosurgical intervention.5,6 Historically, the options for evaluation have included skull radiography, CT scanning, and observation in the emergency department or hospital. Several studies have concluded that patients with normal findings on neurologic examination and CT scanning can be safely discharged from the emergency department.7,12,13,14 CT is the preferred imaging study for patients with minor head injury, although magnetic resonance imaging (MRI) is more sensitive in detecting subtle lesions.15,16 As MRI becomes more widely available, it may have a greater role in the evaluation of patients with minor head injury.
The goal of our study was to develop and validate a simple set of clinical criteria for identifying patients with minor head injury who should undergo CT scanning. In phase 1 of the study, demographic data, symptoms, and the results of physical examination were prospectively recorded. In prior studies, the demographic data associated with abnormalities on CT in patients with minor head injury have included an age over 60 years,6,17,18 alcohol intoxication,9,19,20 and coagulopathy.21,22 Symptoms associated with positive CT scans have included headache,5 nausea and vomiting,5,7 amnesia,9,19,22 and seizures.7,23 The findings with respect to minimal confusion are controversial.24 Although most patients with loss of consciousness do not subsequently remember the traumatic event, some patients also have persistent anterograde amnesia, as evidenced by deficits in short-term memory, but are oriented and can follow commands. Such patients are considered to have a score of 15 on the Glasgow Coma Scale.6,25 We included deficits in short-term memory as a criterion in order to account for patients with persistent anterograde amnesia and an otherwise normal score on the Glasgow Coma Scale. Findings on physical examination that have been associated with a positive CT scan include signs of linear, basilar, or depressed skull fracture6,18,26 and scalp hematoma or soft-tissue injury.18
Of the 520 patients in phase 1, 36 (6.9 percent) had positive CT scans, a finding that is consistent with the results of other studies.4,5,6 Recursive partitioning revealed that the presence of one or more of seven findings — headache, vomiting, an age over 60 years, drug or alcohol intoxication, deficits in short-term memory, physical evidence of trauma above the clavicles, and seizure — was associated with a positive CT scan. In previous studies, the presence of coagulopathy has been associated with a positive CT scan and a poor outcome.21,22 Since patients with coagulopathy were underrepresented in our study, we could not evaluate this criterion.
In phase 2, the criteria derived from the findings in phase 1 were applied to a separate group of patients with minor head injury to determine their sensitivity in predicting the results of CT. Several studies have evaluated various combinations of clinical findings as predictors of positive CT scans in patients with minor head injury, but none of the combinations had a sensitivity of 100 percent.4,5,6,7,9,18,27,28,29 In a study of 2143 patients, 96 percent of those with positive scans were identified on the basis of a finding of nausea, vomiting, or severe headache or evidence of a depressed skull fracture.5 In a study of 1448 patients, 92 percent of those with positive CT scans were identified on the basis of the presence of cranial soft-tissue injury, evidence of a basilar skull fracture, abnormalities on neurologic examination, or an age over 60 years.18 In a study of 373 patients, 90 percent of those with positive CT scans were identified on the basis of the presence of alcohol intoxication, abnormal findings on neurologic examination, or amnesia.9 Some authors have argued that the failure to identify intracranial lesions in some patients with minor head injury who do not require neurosurgical intervention is an acceptable risk of selective CT scanning,5,30 but others consider the risk unacceptable because of the potential morbidity associated with missed lesions.2 In our study, we considered a lower specificity acceptable in order to obtain 100 percent sensitivity in identifying patients with positive CT scans.
In phase 2 of our study, all 57 patients with a positive CT scan had one or more of the seven findings; all 212 patients in whom none of the findings were present had negative scans. The presence of all seven findings had a sensitivity of 100 percent (95 percent confidence interval, 95 to 100 percent) for identifying patients with positive CT scans, and the absence of all seven findings had a negative predictive value of 100 percent. Despite this high sensitivity, the lower limit of the confidence interval indicates the possibility of missing an intracranial lesion that would be detected by CT scanning. Application of the criteria to the patients in phase 2 would have reduced the use of CT by 22 percent. Even a 10 percent reduction in the number of CT scans ordered for patients with minor head injury would result in a substantial decrease in health care expenditures.9
Of the 1429 patients in phases 1 and 2 combined, 93 had positive CT scans. Cerebral contusions and subdural hematomas were the most common injuries, a finding that is consistent with the results of other studies.5,18 Only 6 patients (0.4 percent) underwent neurosurgery, a finding that is also consistent with the results of other studies.5,6 Since the patients in our study were not followed after discharge, data on possible delayed complications were not available. There have been a few reported cases in which patients with initially normal CT findings were subsequently found to have intracranial lesions.31,32 However, several groups have concluded that patients with normal findings on both neurologic examination and CT scanning can be safely discharged from the emergency department.7,12,13,14,33
The goal of our study was to derive and validate a set of clinical findings that could be used to identify patients with minor head injury who should undergo CT. Our results suggest that such patients can be identified by the presence of one or more of the following seven findings: headache, vomiting, an age over 60 years, drug or alcohol intoxication, deficits in short-term memory, physical evidence of trauma above the clavicles, and seizure. Although the combination of these findings had a sensitivity of 100 percent for identifying patients with positive CT scans, they should be validated at other centers.
We are indebted to Hilary Thompson, Ph.D., Ted Glassman, M.D., Michael Volner, M.D., Roy Brandhurst, M.D., and Dauphinot Piper, M.D., for their help with the study.
Source Information
From the Sections of Emergency Medicine (M.J.H., C.A.P., T.J.M., S.L., P.M.C.D.) and Emergency Radiology (E.B.), Louisiana State University Health Science Center at New Orleans, New Orleans.
Address reprint requests to Dr. Haydel at the Emergency Medicine Offices, Charity Hospital, 13th Fl., 1532 Tulane Ave., New Orleans, LA 70112, or at micellehaydel{at}yahoo.com.
References
| |||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Indications for Computed Tomography after Minor Head Injury
Stein S. C., Stiell I. G., Laupacis A., Wells G. A., The Canadian CT Head and Cervical-Spine Study Group , Kohn M. A., Newman T. B., Haydel M. J.
Extract |
Full Text
N Engl J Med 2000;
343:1570-1571, Nov 23, 2000.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
