Emergency Radiology (2000) 7: 7±13
Ó American Society of Emergency Radiology 2000
O R I GI N A L A RT IC LE
Mel E. Herbert
Issues in minor head injury: to CT or not to CT, that is the question
Abstract Each year, approximately five million patients present to US Emergency Departments (EDs) with minor head injuries (MHI). Most of these patients have insignificant injuries that require no specific therapy. However, a small fraction have serious injuries that are amenable to treatment. Neuroimaging can reliably detect clinically significant injuries, but making a logical decision about the need for neuroimaging in a given patient is a complex process that involves multiple clinical variables. To date, no one has identified exactly which characteristics are reliable in predicting (or excluding) the need for emergent radiographic imaging. This review highlights the issues surrounding the decision to perform neuroimaging in MHI, including modality of choice, consequences of ªmissed injuries,º and current data on risk stratification and attempts to identify ªlow-riskº patients.
Issues in deciding to perform CT scanning in minor head trauma Each year, approximately five million patients present to US Emergency Departments (EDs) with minor head injury (MHI) [1]. Most of these patients have insignificant injuries that require no specific therapy. However, a small fraction of them have serious injuries that are amenable to treatment. Many authorities, noting the low incidence of injury requiring intervention, are reluctant to advocate neuroimaging of all MHI cases [2, 3, 4]. Others, concerned about the potentially grave consequences of a ªmissedº injury, have encouraged liberal computed tomographic (CT) scanning [5, 6]. It is curM. E. Herbert Department of Emergency Medicine, Olive View ± UCLA Medical Center, 14445 Olive View Dr. ± North Annex, UCLA School of Medicine, Sylmar, CA 91342, USA e-mail:
[email protected] Tel.: + 1-8 18-3 64 39 95 Fax: + 1±818±3643268)
rently unclear which of these recommendations, if any, is most useful. Making a logical decision about the need for neuroimaging in a given patient is likely to involve multiple characteristics of the presentation. Unfortunately, no one has identified exactly which characteristics are the most important in determining the need for emergent radiographic imaging. Add to this the fact that no has determined an acceptable ªmiss rate,º and it becomes clear that little in this area is clear. This review will highlight the issues regarding the decision to perform CT scanning on patients with MHI and the current state of the literature.
Steps in the decision making process Developing a rational algorithm to guide the use of CT scanning of patients with MHI involves a number of important data elements. These include the prevalence of serious intracranial pathology among MHI patients, the consequences of a ªmissedº injury, the chance of detecting the injury on CT, the cost and risk of CT scanning, and the role of alternatives to CT scanning (e. g., other imaging modalities, observation, etc.). Further, one may reasonably ask whether a set of clinical criteria can be used to define a very low risk subgroup of MHI patients who do not, because of an extremely low prevalence of relevant injury, require CT scanning.
Prevalence of significant intracranial injuries among patients with MHI The reported prevalence of intracranial injury among patients with MHI varies greatly. The reasons for this variation include differing definitions of MHI, selection bias and inconsistent definitions of significant intracranial injury. Each of these issues will be addressed individually.
8
Definition of MHI MHI has often been defined as blunt head trauma in a patient with a Glasgow coma score (GCS) of 13 or above in the ED [5]. This definition is now rejected by many authors [2, 3, 4, 7, 8], who note that the prevalence of surgically important injury in patients with GCS less than 15 is consistently found to be very high [5, 6, 9]. The reported prevalence of significant injury among patients with GCS of 13 or 14 is between 6 and 40 % [3, 5, 6, 10], and as many as 20 % require neurosurgical intervention [5, 6]. Even if the lower prevalence rates are accurate, they represent unacceptable ªmissº rates. Thus, virtually all patients with blunt head trauma and a GCS less than 15 in the ED should be viewed as candidates for emergent neuroimaging. In consequence, most authors now consider the definition of MHI to include only those patients with a GCS of 15 in the ED [2, 3, 7, 8]. Selection bias Selection bias is present in all studies of MHI, but is particularly severe in some analyses. It is impossible to determine the prevalence of a disease if one cannot determine the denominator from which the study population is derived. As a result, studies of patients referred to trauma services [6] or of admitted patients [4] substantially overestimate the true prevalence of significant injuries among MHI populations. Even studies from consecutive ED presentations may overestimate the prevalence of injury if they count only patients who undergo CT scanning. Published series frequently include only patients who underwent neuroimaging and often fail to obtain follow-up information on unscanned patients [2, 3, 4, 5, 6, 7, 8, 9, 10]. Failure to recognize significant injuries among the unscanned patients may lead to underestimation of injury prevalence, while failure to count the unscanned patients as MHI victims tends to produce overestimation. Many studies fail to state the criteria for neuroimaging, or give insufficient detail for the study to be replicated. In addition, the issue of compliance with a stated department CT guideline is never addressed [2, 5, 6, 7, 10, 11, 12, 13]. Indeed, the true prevalence of serious injuries among patients with MHI could be determined only if all patients with blunt head injury underwent serial CT scanning. This type of study will almost certainly never be performed. Definition of ¹significant injuryª The definition of ªsignificant injuryº is another factor that makes it difficult to interpret findings from many studies. A consistent definition is neither accepted nor used by all investigators. For example, some studies have counted patients with simple linear skull fractures among their ªsignificant injuryº victims [5, 6], while oth-
ers have used the need for surgical intervention as a principal outcome measure [2, 7]. Simple linear skull fractures are correlated with significant head injury, but their predictive value for intracranial bleeding and the need for surgery is extremely poor [14]. Consequently, most authorities do not include isolated linear skull fractures among their list of serious injuries. Designating relatively minor injuries (e. g., skull fracture) or any CT abnormality as significant increases the apparent prevalence of serious injuries among MHI victims. The best definitions of serious injury focus on conditions that have potential for remedial intervention, in particular, ªthe need for surgery.º Reasonable definitions of serious injury usually include any intracranial blood, air, contusion, or edema, presence of depressed or diastatic skull fracture, or other surgically remedial condition. Even injuries like intracranial blood or contusion, that may be apparent on CT scan, do not always require surgical intervention. In some cases, these may not be ªsignificantº injuries. Even the ªneed for surgeryº as an endpoint is somewhat problematic. The indications for surgical intervention of intracranial lesions have changed over the years, with an increasing trend towards nonsurgical management of smaller lesions in clinically stable patients [15, 16, 17]. Some intracranial lesions are routinely observed in the inpatient setting, generally because of concern about potential rapid deterioration. It is unclear what percentage of patients admitted for observation under these circumstances actually require the services of an inpatient unit [15, 16, 17].
Estimating the prevalence of injury using currently accepted MHI definitions Using the endpoints noted above, it is possible to estimate the prevalence of significant injury among MHI victims from the current literature. Tables 1 and 2 summarize the case series involving patients with MHI with an initial GCS of 15 in the ED. The total number of MHI patients in these studies numbers 7,996. The total number of patients undergoing craniotomies is 59. Thus, the prevalence of injury requiring neurosurgical intervention is 0.74 %, or less than 1 %. In comparison, a total of 4,519 MHI patients were selected for CT imaging, and 302 were found to have abnormal findings (excluding simple skull fractures where possible), yielding a 6.68 % prevalence of abnormal CT scans. Although these figures should be interpreted with caution (due to problems in defining the underlying populations), they suggest that the prevalence of CT abnormalities among blunt head injury victims is nearly ten times greater than the number of surgically important lesions. It is encouraging to note that the best studies, as well as most recent studies, all suggest a very low prevalence of surgical injury in patients with MHI and GCS of 15 [2, 3, 4, 7]. In fact, these same studies suggest that the prevalence of surgically important lesions in patients with
Yes: Admit all concussion (no definition)
Taheri et al. 1993 [13], retrospective data collection Fulton et al. 1993 [12], retrospective data collection Schynoll et al. 1993 [8], prospective data collection Harad and Kerstien 1992 [10], retrospective data collection GCS > ? only admitted patients
For definition of low-risk group, see text
a
All patients with head CT done for blunt trauma within last 2 weeks N/A GCS ³ 13, LOC/amnesia, focal deficits, depressed skull fracture, declining mental status Stein and Ross 1990 Yes. per standard CT GCS > 13, LOC/ampolicy nesia, CT in 6 h not [5], related to clinical retrospective data deterioration collection Feuerman et al. 1988 No standard CT pol- GCS 13, < 24 hours icy, ? standard admit from injury [11], policy retrospective data collection Dacey et al. 1986 [9], Standard admit poli- GCS ³ 13, LOC or loss of speech, vision cy: all admitted for retrospective data or memory at time 12 h observation collection of event
No: CT at physician's discretion
No standard policy
Yes: LOC or amnesia
GCS 15, LOC/amnesia
Yes: LOC or amnesia even if normal neuro in ED Yes: LOC or amnesia
Jeret et al. 1993 [7], Prospective data collection
GCS ³ 13
N/A
Culotta et al. 1996 [4], retrospective data collection Miller et al. 1996 [2], prospective data collection Borczuk 1995 [3], retrospective chart review GCS ³ 13 and underwent head CT (generally for LOC/ amnesia, neuro deficit, persistent or progressive symptoms, intoxication) GUS 15, LOC/amnesia (any duration), < 24 h since event, age > 18 ? Exact GCS, LOC/ amnesia
Study population
Standardized policy for CT or observation?
Reference
N/A
Yes
N/A
> 18
14±90
15±87
All ages
Not stated
Not stated
18±80
Yes
No, only those admitted, & ¹sickerª patients All admitted patients Yes
Yes
Yes
No, only those admitted & ¹sickerª patients
No, only those dete- 1±90 riorating during 12 h observation period
No
> 16 only
N/A
No
No
Yes
Yes
Yes
Yes
No
14±91 All ages
Skull fracture counted as ¹abnormal CTª?
Ages (years)
92 % of study population scanned Yes
All patients scanned?
2
533 Need for invasive neurosurgery procedure
N/A
N/A
0
0
0
454
251
213
296
N/A
N/A
N/A
66 (4.8 %)
95 (4 %)
N/A
N/A
N/A
34 (7.5 %)
N/A
N/A
59 (13 %)
43 (17 %) 43 (17 %)
15 (7.0 %)
4 (0.8 %) craniotomy
2 (0.8 %)
17 (3.7 %)
5±11 (2±4.4 %)
3 (1.4 %)
0
<5
3 craniotomy
1 craniotomy
3 craniotomy
10 (0.4 %)
CT abnor- Surgeries malities ex- performed cepting simple skull fracture
5 (1.7 %) N/A
1 (0.3 %) N/A
310
67 (9.4 %)
72 (5.9 %)
84 (6.1 %)
95 (4 %)
0
0
1
N/A
Surgeries CT abin ¹low- normaliriskª ties groupa
712
1211
1382
2398
Patients
Results: total numbers of
236
Surgery
CT abnormality, surgery
CT abnormalily, surgery
CT abnormality
Surgery
Surgery, deterioration
(CT abnormality, surgery
CT abnormality, surgery
CT abnormality, surgery
CT abnormality, surgery
End point
Table 1 Summary of case series in minor head trauma patients with CGS 15 (N/A not available, LOC loss of consciousness)
9
10 Table 2 Intracranial injuries requiring surgical intervention among minor head trauma patients. (SDH subdural hematoma, EDH epidural hematoma, M male, ICP intracranial pressure, N/A not available, MVA motor vehicle accident, MI myocardial infarction) Case no.
Age, sex, mechanism
CT findings
Procedure
Interval: injury to operation
Outcome
Comments
Miller et al. [2], 1
N/A
Depressed skull fracture
Elevation of fracture
N/A
Good recovery
Palpable
Miller et al. [2], 2
N/A
Depressed skull fracture
Elevation of fracture
N/A
Good recovery
Miller et al. [2], 3
N/A
Contusion and SDH
Partial lobectomy day 13
N/A
Moderate deficits Severe headache, nausea, hematoma
Borczuk [3], 1
18, fall
Epidural
Craniotomy
2 days later
Good recovery
Initial 15 min LOC, then presented 2 days later with headache, nausea, and vomiting
Jeret et al. [7], 71
49, M, assault
Contusion, SAH
Craniotomy
Several hours
Good recovery
Initial GCS of 15, then progressive lethargy and agitation in the ED
Jeret et al. [7], 297
41, M, assault
SDH
Craniotomy
< 1 day
Good recovery
Jeret et al. [7], 530
79, M, assault
Intracranial free air
I. v. antibiotics and observation
< 1 day
Taheri et al. [13], 1
29, F
EDH
? Intervention
N/A
Taheri et al. [13], 2
18, M
EDH
? Intervention
N/A
Non-focal exam
Taheri et al. [13], 3 Taheri et al. [13], 4
63, M 36, M
Contusion EDH, SDH
? Intervention ? Intervention
N/A N/A
Focal exam Focal exam
Taheri et al. [13], 5
72, M
SDH
? Intervention
N/A
Focal exam
Dacey et al. [9], 17
50, M, assault
Normal CT
ICP monitoring
< 1 day
Good recovery
No further information
Dacey et al. [9], 18
35, M, fall
R frontal contusion
ICP monitoring
< 1 day
Good recovery
No further information
Feuerman et al. [11], 1 34, M, assault
EDH
Craniotomy
72 h after injury presented with headache
Good recovery
Headache, poor memory, difficult speech
Feuerman et al. [11], 2 46, M, MVA
EDH
Craniotomy
Deteriorated in Good recovery ED, ? time frame
Up-going plantar on one side
MHI is significantly less than 1 % and maybe as low as one in a thousand.
Health consequences of a missed injury The value of CT scanning is directly related to the cost of a ªmissedº injury. These consequences may be small if missed lesions produce gradually increasing symptoms that prompt patients to seek timely re-evaluation. Alternatively, missed injuries that produce rapid onset of coma, disability, or death may be associated with much greater costs. Because of ethical concerns, few
Normal exam, vomiting three times, headache, dizziness Died of unrelated Normal exam, injuries mild headache, intracranial lesion resolved. Had an MI on hospital day 5. Focal exam
studies specifically investigate the natural history of intracranial injury after MHI. However, review of the series listed in Table 1 reveals information on ten cases requiring craniotomy and the time frame associated with deterioration. Four of these ten patients deteriorated ªwithin hoursº in the ED. Two patients deteriorated within 1 day of their presentation, and four others deteriorated after 48 h or more. A British study by Voss et al. [23] found that of 11,700 patients that presented with head injury, 606 returned for re-evaluation. Of these 606 patients, 5 % needed operative intervention, most presenting more than 3 days after the initial injury with worsening headache. It is not clear from the paper
11
what percentage of these patients had MHI on initial presentation. Based on this best evidence, it appears that patients who ultimately deteriorate to the point of requiring surgery do so in three peaks. The first peak occurs within a few hours of the insult. A second peak, consisting of patients who generally present with slowly progressive headache, occurs a day or two later, and a third peak occurs some weeks to months later and involves patients who present with progressive headache and/or confusion. It is important to note that performing CT in the ED at initial presentation may not necessarily prevent some of these outcomes, particularly in cases where initial scanning fails to demonstrate any evidence of injury [4]. Review of older neurosurgical literature, from the era before CT, reveals that significant injuries were probably missed quite frequently. Presumably most of these patients did fine. In the last 10 years there has been a trend towards more liberal scanning of MHI victims, and a simultaneous move to treat asymptomatic or minimally symptomatic patients without surgery [15, 16, 17]. Papers documenting this trend provide good evidence that the mental status of the patient prior to surgery is the most important prognostic factor [24, 25, 26, 27]. Comatose patients do poorly, while those awake and alert do well [24, 25, 28]. It is possible to extrapolate these findings to the patient who presents with slowly progressive symptoms a day or more after their injury. Such patients are likely to have a relatively good prognosis, and ªmissingº their injury on the initial ED visit probably does little to affect ultimate outcome. Alternatively, patients who present with rapid deterioration shortly after their initial evaluations may be expected to have relatively poor prognoses, and experience dire consequences as a direct result of their missed injuries.
The utility of CT scanning in detecting injury Generally, noncontrast CT scanning is considered the modality of choice for detecting significant intracranial injuries in blunt head injury victims. However, the sensitivity of CT is very time-dependent, particularly for some types of intracranial injury. Very early CT scans occasionally produce false negative scans, and repeat imaging may be necessary in patients exhibiting persistent or progressive symptoms. For example, small epidural hematomas may be missed on initial scans, but as the lesions expand and become clinically significant they may become evident on later CT scans [4]. In a study by Culotta et al. of over 2,300 consecutive ED patients with MHI, ten patients required surgery. Five of these ten patients had no injury noted on initial CT scan [4]. The false positive rate for CT scanning in MHI is very low, and CT rarely identifies lesions that are not present at operation. More frequently, and of greater concern, is the ability of CT to detect injuries that do not require
treatment. These lesions are not false positives in the usual sense, but their detection may trigger invasive or expensive interventions than would not have occurred without the technology, and with little or no clinical gain. This effect becomes increasingly important as clinicians move toward nonsurgical management of many intracranial injuries [16, 17, 18]. Improved detection of small lesions, along with changes in the indications for surgery, may explain, to a significant degree, the apparent decreasing rate of surgery in MHI over the last 10±15 years [19, 20].
Health consequences of CT scanning If CT scanning was rapid, inexpensive, widely available, and without health effects, any discussion of the utility of CT would be moot. Unfortunately, CT scanning does cost time and money, and is associated with potentially dangerous health effects, particularly when applied to large populations. CT scanning involves low level exposure of the head and neck to ionizing radiation. The average CT scan exposes thyroid tissue to 0.5 mGy [30]. Thirty-year-olds exposed to this level of radiation can be expected, during their lifetime, to develop two new thyroid cancers for every 10,000 radiated individuals [31]. Extrapolating these figures to the estimated one million CT examinations performed for minor head injury each year reveals that such imaging may generate up to 200 new thyroid cancers over the lifetime of the population scanned. Radiation health effects place a natural limit on the utility of CT. Imaging is only beneficial when the chance of detecting a correctable lesion equals or exceeds the potential of inducing a lethal malignant transformation.
Costs of CT scanning The true cost of scanning is difficult to estimate and varies by practice setting. At a minimum, CT imaging requires use of technician time, machine wear and tear, and image processing costs. In contrast, charges for a CT scan can be considerable, and often range from several hundred to over one thousand dollars at some institutions. If these charges are applied to a large population of low-risk patients, the charges may easily exceed tens of thousands of dollars for each significant lesion detected. Consider, for example, an institution that charges $500 for a noncontrast head CT. At an intracranial injury prevalence of 1 %, CT charges alone amount to $50,000 for each lesion identified. This estimate is conservative, because head CT charges frequently exceed $500, while the prevalence of significant injury is probably much lower [2, 3, 4, 7]. CT scanning also takes time, although the amount of time varies by institution. At some facilities, delays of several hours may be added to already significant patient waiting times [29]. Furthermore, CT scanning may
12
interfere with or delay other care urgently needed by blunt trauma victims [12]. It is important to balance the potential benefits of scanning against the risks associated with delaying other aspects of care.
Using clinical criteria to define a very low risk group The low incidence of significant injury among MHI victims, coupled with medical and financial costs, makes it desirable to develop clinical criteria that reliably identify a subset of patients having essentially no risk of significant intracranial injury, and hence no need for neuroimaging. As mentioned previously, many MHI case series have used GCS as a surrogate form of neurologic examination. These series inevitably find significant injuries among patients exhibiting normal GCS scores [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]. The GCS was developed as a means to estimate the long-term prognosis of head injury victims based on clinical findings 6 h after injury. It was never meant to be a thorough neurologic screen, and many patients with profound neurologic deficits exhibit perfect GCS scores [21]. While GCS scores are clearly not reliable, it may be possible to stratify risk on the basis of more detailed examinations or historical factors [22]. Potential risk criteria include: abnormal mental status (as defined by treating clinician); memory deficits; loss of consciousness and prolonged loss of consciousness; focal neurologic deficits (defined as: abnormal gait, strength, or reflexes in upper and lower limbs, abnormal cranial nerve exam, abnormal cerebellar function); intoxication; coagulopathy; recurrent or projectile vomiting; progressive or severe headache; high-force small surface area injuries; scalp hematoma (particularly in young children with limited verbal capacity); and basilar, depressed, or diastatic skull fractures. The case series summarized in Tables 1 and 2 reveal that the above criteria may reliably identify a set of ªlow-riskº patients whose risk of surgical injury is of the order of 0.1 %. It must be emphasized that these suggested criteria are only hypothetical and must be validated before being used in the clinical setting. These risk criteria do not apply to pediatric populations, where the existing MHI literature is very difficult to analyze due to frequent study design problems, small sample sizes, and difficulties in performing reliable evaluations on infants and children [38].
Alternatives to CT scanning A number of alternate strategies have been suggested for evaluating patients with MHI. Historically, skull films have been used to identify high-risk patients (i. e., those having skull fractures) [33]. This approach was abandoned when skull fractures were found to be poor predictors of important injury [14]. Most patients with skull fracture have no intracranial injury, and many pa-
tients with significant injury have no skull fracture. Furthermore, isolated linear skull fractures need no therapy, and skull radiography exposes patients to significant radiation and its attendant risk of malignant transformation. Magnetic resonance imaging (MRI) is excellent at visualizing many types of cranial pathology and is able to detect many injuries not seen on CT [34]. However, nearly all of these ªnewfound injuriesº are clinically insignificant and do not influence the overall care of MHI victims. Additionally, MRI is slow, expensive, and frequently unavailable. Thus, its use as a screening device cannot be recommended at this time. Observation has been proposed as an alternate strategy for some patients. Personal experience reveals that 4 h ªneuro-observationº is a reliable means of excluding significant injury in patients who sustain loss of consciousness but are without focal signs. Patients who remain without findings at the end of this timeframe are sent home with a caretaker who can provide further observation and return if the victim develops signs of deterioration. This type of observation occurs in many busy or inefficient EDs as a matter of course. There is some validity to this strategy, and it may be appropriate in some practice settings. The practice of routine admission for MHI, as an alternative to CT scanning, is not cost-effective in most institutions, as the cost of a 1-day hospital admission typically exceeds the cost of a CT scan [35, 36, 37].
Summary A large number of patients with mild head injuries present to EDs each year, but the prevalence of injuries requiring intervention is very small (perhaps as low as one in a thousand). Using CT imaging to screen these patients generates significant costs and is associated with potentially serious health effects. Risk stratification could improve our ability to use CT more efficiently. However, it is currently impossible to assign risk classification to individual patients, and the decision to image patients remains fraught with complexity. Large multicenter prospective trials to develop low risk criteria are required and are currently underway in both Canada and the US. Until the results of these studies are available, the decision to perform CT imaging will continue to be a clinical dilemma.
References 1. Marshall LF, Toole BM, Bowers SA (1983) The National Traumatic Coma Data Bank: Part 2. Patients who talk and deteriorate: implications for treatment. J Neurosurg 59:285±288 2. Miller EC, Derlet RW, Kinser D (1996) Minor head trauma: is computed tomography always necessary? Ann Emerg Med 27:290±294 3. Borczuk P (1995) Predictors of intracranial injury in patients with mild head trauma. Ann Emerg Med 25:731±736
13 4. Culotta VP, Sementilli MF, Gerold K, et al (1996) Clinicopathological heterogeneity in the classification of mild head injury. Neurosurgery 38:245±250 5. Stein SC, Ross SE (1990) The value of computed tomographic scans in patients with lowrisk head injuries. Neurosurgery 26:638±640 6. Stein SC, Ross E (1992) Mild head injury: a plea for routine early CT scanning. J Trauma 33:11±13 7. Jeret JS, Mandell M, Anziska B, et al (1993) Clinical predictors of abnormality disclosed by computed tomography after mild head trauma. Neurosurgery 32:9±16 8. Schynoll W, Overton D, Krome R, et al (1993) A prospective study to identify high yield criteria associated with acute intracranial computed tomography finding in head-injury patients. Am J Emerg Med 11:321±326 9. Dacey RG, Alves WM, Rimel R, et al (1986) Neurosurgical complications after apparently minor head injury: assessment of risk in a series of 610 patients. J Neurosurg 65:203±210 10. Harad T, Kerstien MD (1992) Inadequacy of bedside clinical indicators in identifying significant intracranial injury in trauma patients. J Trauma 32:359±363 11. Feuerman T, Wackym PA, Gade GF, et al (1988) Value of skull radiography, head computed tomographic scanning and admission for observation in cases of minor head injury. Neurosurgery 22:449±453 12. Fulton RL, Everman D, Mancino M, et al (1993) Ritual head computed tomography may unnecessarily delay lifesaving trauma care. Surg Gynecol Obstet 176:3 327±332 13. Taheri PA, Karamanoukian, Gibbons K, et al (1993) Can patients with minor head injuries be safely discharged home? Arch Surg 128:289±292 14. Masters SJ, McClean PM, Arcarese JS, et al (1987) Skull X-ray examinations after head truama. Recommendations by a multidisciplinary panel and validation study. N Engl J Med 316:84±91 15. Dent DL, Croce MA, Menke PG, et al (1995) Prognostic factors after acute subdural hematoma. J Trauma 39:36±43 16. Croce MA, Dent DL, Menke PG, et al (1994) Acute subdural hematoma: nonsurgical management of selected patients. J Trauma 36:820±827 17. Bender MB, Christoff N (1974) Nonsurgical treatment of subdural hematomas. Arch Neurol 31:73±79 18. Singounas EG, Sfakianos G, Sourtzis I, et al (1990) ªBenignº acute subdural haematomas: analysis of 12 cases. Acta Neurochir 196:140±144 19. Lobato RD, Rivas JJ, Cordobes F, et al (1988) Acute epidural hematoma: an analysis of factors influencing the outcome of patients undergoing surgery in coma. J Neurosurg 68:48±57 20. Servadei F, Piazza G, Seracchioli A, et al (1988) Extradural haematomas: an analysis of the changing characteristics of patients admitted from 1980 to 1986. Diagnostic and therapeutic implications in 158 cases. Brain Injury 2:87±100
21. Teasdale G, Jennet B (1974) Assessment of coma and impaired consciousness. A practical scale. Lancet 2:81±84 22. Moran SG, McCarthy MC, Uddin DE (1994) Predictors of positive CT scans in the trauma patients with minor head injury. Am Surg 60:533±535 23. Voss M, Knottenbelt JD, Peden MM (1995) Patients who reattend after head injury: a high risk group. BMJ 311: 1395±1398 24. Klauber MR, Marshall LF, Luerssen TG, et al (1989) Determinants of head injury mortality: importance of the low risk patient. Neurosurgery 24:31±36 25. Paterniti S, Fiore P, Macri E, et al (1994) Extradural haematoma: report of 37 consecutive cases with survival. Acta Neurochir 131:207±210 26. Servadei F, Faccani G, Roccella P, et al (1989) Asymptomatic extradural haematomas. Results of a multicenter study of 158 cases in minor head injury. Acta Neurochir 96:39±45 27. Phuenpathom N, Choomuang M, Ratanalert S (1993) Outcome and outcome prediction in acute subdural hematoma. Surg Neurol 40:22±25 28. Bricolo AP, Pasut LM (1984) Extradural hematoma: toward zero mortality. A prospective study. Neurosurgery 14:8±12 29. Kyriacou DN, Ricketts V, Dyne PL, McCollough, MD, Talan DT (1999) A 5-year time study analysis of emergency department patient care efficiency. Ann Emerg Med 34:326±335 30. Nishizawa K, Maruyama T, Takayama M et al (1991) Determinations of organ doses and effective dose equivalents from computed tomographic examination. Br J Radiol 64:20±28 31. Bier V (1990) Health effects of exposure to low levels of radiation. National Research Council, National Academy Press, Washington DC 32. Cooper PR, Ho V (1983) Role of the emergency skull X-ray films in the evaluation of the head injured patient: a retrospective study. Neurosurgery 13:136±140 33. Thornbury JR, Campbell JA, Masters SJ, et al (1984) Skull fracture and the low risk of intracranial sequelae in minor head trauma. AJR Am J Roentgenol 143:661±664 34. Orrison WW, Gentry LR, Stimac GK, et al (1994) Blinded comparison of cranial CT and MR in closed head injury evaluation. Am J Neuroradiol 15:351±356 35. Sainsbury CP, Sibert JR (1984) How long do we need to observe head injuries in hospital? Arch Dis Child 59:856±859 36. Mendelow AD, Campbell DA, Jeffrey RR, et al (1982) Admission after mild head injury: benefits and costs. BMJ 285:1530±1532 37. Ros SP, Ros MA (1989) Should patients with normal cranial CT scans following minor head injury be hospitalized for observation? Pediatr Emerg Care 5:216±218 38. Savitsky EA, Votey SR (1999) Current controversies in the management of pediatric minor head injuries. J Emerg Med (in press)