- Email: [email protected]
A national survey of neurosurgical care for penetrating head injury
370
Surg Neurol 1991;36:370-7
A National Survey of Neurosurgical Care for Penetrating Head Injury Howard H. Kaufman, M.D., Karen Schwab, M.A., and Andres M. Salazar, M.D. Department of Neurosurgery, West Virginia University, Morgantown, West Virginia, and Head Injury Unit, Uniformed Services University of the Health Sciences, Bethesda, Maryland
Kaufman HH, Schwab K, Salazar AM. A national survey of neurological care for penetrating head injury. Surg Neurol 1991;36:370-7.
We report results of a survey on the management patterns of penetrating head injury (PHI). American neurosurgeons (N = 2969) were asked to participate in a mail survey. One thousand one hundred twenty-eight responded, providing detailed information about their practices, their opinions concerning diagnostic testing, nonoperative therapy, and surgical debridement for PHI. Although there was agreement on some areas of management an~l care, including the use of computed tomography scanning, antibiotics, anticonvulsants, and a few surgical indications/contraindications, there was wide variation on a significant number of points. These included testing for coagulopathy; use of corticosteroids, intracranial pressure monitoring, and barbiturate coma; and surgical indications including debridement of contusions, removal of fragments, location of injury, and Glasgow Coma Score 5-8. Many nonoperative decisions were influenced by the presence of neurosurgical residents, whereas the decision to operate was based primarily on judgment of patient salvageability. This survey uncovered disagreement on several crucial issues in the care of patients with penetrating head injury and the need for continued research in this area of patient care. KEY WORDS; Survey; Neurosurgeons; Surgical debridement; Penetrating head injury; Gunshot wounds; Military
If trauma can be considered "the hidden epidemic," gunshot wounds are probably its most obscure subcategory, despite the fact that firearms are among the 12 leading causes of death in the United States and the top 10 causes of accidental death. Firearms are the etiology of 58% of homicides and 57% of suicides [2]. T h e fre-
Address reprint requests to: Howard H. Kaufman, M.D., Professor and Chairman, Department of Neurosurgery, Health Sciences Center, West Virginia University, Morgantown, West Virginia 26506. Received August 20, 1990; accepted May 20, 1991.
© 1991 by Elsevier Science Publishing Co., Inc.
quency of nonfatal injuries is not available. A large proportion of the injuries involve the brain [23], and an overwhelming percentage of brain injuries immediately or rapidly result in death [19,23]. Informal discussions over the last several years led us to believe that there might be an important variability in some aspects of the care in gunshot wounds to the head, including surgical decision making, medical management, and ethical issues. To identify the variations, if any, in the care of penetrating head injuries, and as part of a larger, ongoing US Army-sponsored multicenter study of acute penetrating head injury (PHI), we carried out a national survey of neurosurgeons, asking about their approach to various aspects of the care o f victims of gunshot wounds to the head. W h e r e we found variability, we tried to determine what factors might have led to that variability. Methods The questionnaire we used was designed initially by the three investigators to examine demographic information about the neurosurgeons and their approach to the care of patients with P H I (Table 1). It was informally pretested on approximately 20 colleagues. After adjustment, it was pretested formally on three independent random samples of 100 neurosurgeons each. These neurosurgeons' names were obtained f r o m lists supplied by the American Association of Neurological Surgeons and the Congress of Neurosurgeons. Questionnaires were sent only to neurosurgeons living in the United States and its territories. T h e pretest revealed only minor difficulties, and appropriate adjustments to the questionnaires were again made. T h e final questionnaire consisted of 43 questions, although several questions had a n u m b e r of subquestions. Most required closed-end responses, but some were open ended. W e then sent the new questionnaires to the remaining neurosurgeons on the lists. Those that did not return the mailings were sent a second questionnaire. W e eventually received a total of 1128 replies from 2969 neurosurgeons (38%); 966 (from neurosurgeons who had been in active prac0090-3019/91/$3.50
Penetrating Head Injury Care Survey
Surg Neurol 1991;36:370-7
Table 1. Areas Investigated in Questionnaire Demographics
Age Time from residency Military/combat experience Size of city Section of country Size/type of hospital Trauma-center level Number of PHIs/year Assault/suicide/accident Type of weapon Neurosurgery residents
Patient care Screening for coagulopathy Computed tomography
Antibiotics Anticonvulsants Steroids Reason to operate/not operate Surgical timing/technique Intracranial pressure monitoring Brain-death criteria Organ retrieval Social/ethical issues Legal issues
371
teaching centers. The average neurosurgeon saw nine penetrating head-injury patients a year, but this was highly variable (SD = 16.6). Most saw mainly hand-gun injuries (81% -+ 23) as opposed to rifles (10% -+ 16). In the average experience, assaults (42% -+ 29) and suicides (45% -+ 28) were seen with equal frequency, whereas accidents were more unusual (15% -+ 20). N e u rosurgeons in larger cities, at larger hospitals, or at hospitals with level I trauma centers saw larger numbers of patients. The smaller the city, the more likely that the injury was related to suicide. Conversely, homicides were more c o m m o n in larger cities.
Surgical Treatment tice within the past 2 years and had seen patients with P H I ) were used for this analysis. The responses were double-entry key punched and analyzed with an I B M 370 using SAS Institute Software, Version 5, 1985. Bivariate and more complex analyses were carried out and tested, using logistic regression. We defined consensus in treatment as occurring when 75~, or more of neurosurgeons agreed on a particular point in the care of patients with P H I . For example, we asked how strongly hematomas influenced them to do or not to do surgery. Because over 75% (ie, 9 2 % ) said they are influenced to do surgery in such patients, we considered operations in these cases to be the typical approach taken by neurosurgeons today. We also noted in which categories there was a 60% to 74% concurrence. W h e n we found variation (lack of consensus) we analyzed whether certain characteristics of the neurosurgeons and/or their practices might account for their diagnostic or treatment choices. We determined whether variation in the use of tests/treatments was associated with length of time since the neurosurgeon had finished his or her residency, n u m b e r of P H I cases seen annually, whether or not the hospital was a level I trauma center, percent of P H I cases due to assault, and whether there were neurosurgery residents in the hospital.
One of the principal questions studied was what determined the decision to treat or not to treat with intracranial surgical debridement. To demonstrate the responses, we have reproduced the questions, collapsed the data from seven categories to three (influence against surgery, neutral, influence for surgery), and indicated which questions received consensus answers, that is, 75% or m o r e agreement that they were influenced for or against surgery, or were neutral (Table 3). O f the 29 reasons presented in the questionnaire for operating/not operating on patients with P H I s , neurosurgeons reached consensus on eight, and 6 0 % to 74% agreed on another eight (Table 3). Consensus was seen about the significance of low Glasgow C o m a Score (GCS) categories, bilateral pupillary dilation, cerebrospinal fluid leaks, depressed fractures, removal of bone fragments, and removal of hematomas. T h e r e was disagreement on the need for surgery in those patients with the middle range of GCS. M o r e than half of surgeons did not feel it was necessary to operate a second time to r e m o v e retained bone or bullet frag-
Table 2. Demographic Characteristics of Respondents City s i z e < i00,000 100,000-500,000
Results The average age of the respondents was 48 _+ 9 years. They had been out of training 15 -+ 9 years. Twenty-six percent had been in the military, but only 6 % had seen combat. Respondents tended to practice in larger cities compared with the location of all neurosurgeons, but otherwise spread with the population geographically (Table 2). Respondents generally practiced in larger hospitals (Table 2 ) 6 6 % in private hospitals and 2 6 % in resident
500,000-1,000,000 > 1,000,000 Region Northeast South Central West Beds <400 400-800 >800
Neurosurgeons (%)
Population(%)
20
75
34
13
18 28
5 8
Neurosurgeons (%)
Population(%)
18 36 23 23
21 34 25 20
Neurosurgeons (%)
All hospitals (%)
39 49 12
88 10 2
372
Surg Neurol 1991;36:370-7
T a b l e 3.
Kaufman et al
Percent of Neurosurgeons Indicating the Effect of Different Parameters on Their Decisions to Operate a No surgery
Neutral
Surgery
Consensus
Hematoma Contusion Edema with mass effect
1 39 46
7 32 23
92 29 31
+
Remove bone fragments Bone fragments remaining after first debridement Remove bullet or metal fragments Metal fragments remaining after first debridement Copper-jacketed bullets
4 31
6 21
90 49
+
45 71
24 19
31 10
_+
39
26
36
Medicolegal pressure To reassure family
45 49
34 33
21 18
To prevent delayed edema or infection To prevent or repair CSF leak Elevation of depressed fracture
16 4 3
14 8 7
69 88 90
-+ + +
GCS GCS GCS GCS GCS GCS
82 73 50 12 7 10
10 16 29 34 16 14
8 12 21 55 77 77
+ _+
One pupil dilated Both pupils dilated
10 77
19 14
72 9
___ +
Location Dominant hemisphere Deep brain Cross-midline Cross-ventricle
32 70 60 52
39 20 26 32
29 10 13 17
_+ -+
Did not respond to resuscitation Hyperventilation Mannitol
68 68
20 19
12 13
_+ -+
Age (eg, 65 and older) Suicide
44 32
40 52
16 16
3 4 5 6-8 9-12 13-15
+ +
Abbreviations: +, consensus reached by >75% in one category; -+, 60%-74% agreement; GCS, Glasgow Coma Score. Respondents were asked to make the following evaluation: "Reasons why you would or would not operate on penetrating wounds to the head. Please consider each item in isolation from the others. How strongly does this influence you to operate?" The answers were labeled as 0-2 = no surgery, 3 = neutral, 4-6 = surgery.
ments. There was no agreement on the importance of the location of the bullet in the dominant hemisphere or its having crossed the ventricles, although most neurosurgeons do not operate on deep wounds or those that have crossed the midline. Likewise, the patient's age, eg, 65 years or older, or whether the patient was a suicide attempt did not usually influence them on the question of operating or not operating. We wondered whether the neurosurgeons' practice characteristics or their general view on the salvageability or nonsalvageability of PHI patients explained some of the variation that occurs in surgical decisions. Nonsalvageable patients were defined by 68% of respondents as those expected to die, whereas 30% added those expected to survive in a vegetative state, and only 2%
added those expected to survive with a severe disability. We asked, "In your experience, what percentage of the following penetrating head injured patients are ultimately salvageable?" In general, we found that outlook for P H I patients would vary by admission GCS level. Salvageability is generally not expected with GCS 3 and 4, but it is in patients with 9 - 1 2 and 1 3 - 1 5 , although perhaps not to the extent reported in the literature (Figure 1). However, there was considerable variation around the average responses for patients in the GCS 5 - 8 range. Assessments o f salvageability did not vary by number of patients seen, military service of the neurosurgeon, or level of the trauma center. The infection rate of the reporting neurosurgeon correlated with his or her per-
Penetrating Head Injury Care Survey
T a b l e 4. Percent of Responding Neurosurgeons Using Various Neurosurgical Evaluations and Treatment for Penetrating Brain Wounds
100
80
~h
go
373
Surg Neurol 1991;36:370-7
Respondents
60
40
20
3
4 Glasgow
5
9-12
6-8 Coma
13-15
Score
Figure 1. The meanpercent of patients felt to be salvageableand the mean percent treated surgically for a given Glasgow Coma Scoreor group of coma
scores. N = 1015 neurosurgeons +- 1 SD.
ception of the ultimate salvageability of patients with admission GCS 9-15. Multiple logistic regression analysis showed that the neurosurgeons' perceptions of patient salvageability was an important predictor of tendency to operate. In fact, of the variables we examined, it was the only factor significantly associated with tendency to operate on severely injured patients (GCS 8 or less). The larger the percentage of patients seen as ultimately salvageable within an injury class (ie, GCS levels), the more likely the neurosurgeon was to report a tendency to operate on that group of patients. The variable most significantly related to tendency to operate in the less severely injured patient was length of time since the neurosurgeon's residency. That relationship was negative, meaning the longer the neurosurgeon had been in practice, the less likely he or she was to operate on patients with admission GCS 9-15. Surgical technique is fairly standardized. This includes craniotomy (83%) or craniectomy (82%), following the track deep into the brain (75%), and patching the dura (86%), especially with autogenous tissue (79%). Over half (532) do not use artificial dura, whereas 48% never use human prepared dura. As mentioned before, 36% use ultrasound fairly frequently. According to 68%, surgery should be performed as soon as possible, and another 15% felt it should be performed within a day.
Nonsurgical Treatment and Testing Another set of questions probed the incidence of use of various tests and treatments (Table 4).
Do Do Do Do
you usually or always test for coagulopathy you evaluate with CT you use prophylactic antibiotics you use anticonvulsants For 6 months For 1 year For 2 years Longer Do you use corticosteroids routinely Do you use ultrasound Do you usually or always use ICP monitoring
65 97 87 84 30 43 19 9 42 36
If GCS 3-5 If GCS 6 - 8 If GCS 9-12 I f G C S 13-15 Do you use barbiturate coma Threshold pressure (mm Hg) 15-20 21-25 26-30 31-35 36-40 ~40
30 40 25 16 37 6 23 31 17 10 12
Abbreviations: CT, computed tomography; GCS, Glasgow Coma Score; ICP, intracranial pressure.
Testing for coagulopathy [6,26,27] was done frequently (65% usually or always). The most common tests were the prothrombin time, activated partial thromboplastin time, and platelet count. Most did not use thrombin time, fibrinogen split products, or fibrinogen. Computed tomography (CT) scanning [8] was used by 97%, although a few seemed to use it less to make the decision of whether or not to operate (ie, they used it more for surgical planning). Prophylactic antibiotics were generally used (87%), but the duration of treatment varied considerably (Table 5). Multiple antibiotics were used by 438 respondents: two by 320 respondents, three by 80 respondents, and four or more by 32 respondents. The most commonly used antibiotics were cepha-
T a b l e 5. Distribution of Neurosurgeons Who Routinely Use Antibiotics in Patients Expected to Survive and Number of Days
Used ( N -- 750) Neurosurgeons Days
(%)
1-3 4-5 6-7 8-10 ~10
24 19 20 25 13
374
Surg Neurol 1991;36:370-7
Kaufman et al
Table 6. EstimatedInfection Rates (N = 812) Infection(%)
Respondentswho estimate this infectionrate (%)
0
21
1-5 6-10 11-25 -->26
33 28 12 6
losporins (440), chloramphenicol (176), penicillins (117), aminoglycosides (88), vancomycin (47), miconazole (26), tetracycline (20), and erythromycin (19). Although they were asked about routine coverage, it is not clear that each respondent used multiple antibiotics in each patient. Estimation of infection rates for penetrating wounds also varied (Table 6). Anticonvulsants were used prophylactically by 87% of respondents, and 98% of these used phenytoin. Of these, 88% loaded the drug, 60% adjusted by blood level, while 20% gave 300 mg/day and 20% gave 400 mg/day. Thirty-five percent used phenobarbital, generally in combination with phenytoin. Medications were used most often for 1 year or less (Table 4). Twenty-two percent switched to carbamazepine. Corticosteroids were used routinely by 42% of respondents. Thirty-six percent used ultrasound at the time of surgery. Intracranial pressure (ICP) monitoring remained controversial. A few respondents indicated that they might use it without debridement (13 % usually or always), while a few (7% usually or always) used it to determine whether or not to operate. Forty percent of neurosurgeons reported using ICP monitoring "usually" or "always" in patients with a GCS of 6-8 (Table 4). Only 37% ever used barbiturate coma, of whom 83% used it with ICP monitoring. The threshold pressure at which it was used was variable (Table 4). Multivariate regression analysis of which physician/ practice characteristics determined the decision to use tests/treatments where there was no consensus showed that the presence of neurosurgical residents was the most important (Table 7). Neurosurgeons practicing in centers with neurosurgery residents report more frequent testing for coagulopathy; less frequent use of steroids; and more frequent use of intraoperative ultrasound, ICP monitoring of patients with GCS 6-8, and barbiturate coma. Management of "Unsalvageable" Patient~Brain Death Criteria Eighty percent of neurosurgeons will sometimes to always put "unsalvageable" patients into an intensive care
unit, although only 70% will sometimes to always ventilate them. Decisions to carry out such treatment are influenced by questions of legal liability (66% sometimes to always), to show the family something is being done (66% sometimes to always), and for organ procurement (89% sometimes to always). Most respondents (85%) say they do not treat suicides differently. Of the many brain-death criteria [24], 48% used the Harvard criteria of 1986, 6% used the National Institutes of Health Combined Study criteria of 1977, and 12% used the criteria of the President's Commission of 1981. No particular characteristics of physicians or their practices distinguished those who used the most recently proposed criteria, those of the President's Commission. Thirty-four percent used other criteria. Eighty-one percent of respondents' hospitals had specific brain-death criteria. Seventy-five percent stated they had a hospital policy regarding organ retrieval. Ninety-one percent felt their medical examiner is "sometimes to always" cooperative with retrieval. Sixty-six percent "sometimes to always" request donation in homicide, and 70% feel they are successful in half or less of these cases; 86% "sometimes to always" request retrieval in suicide, and 65% feel they are successful in half or less of these cases. Discussion
The advantages of the survey approach are considerable: we were able to obtain a national sample of physician practices (very difficult to obtain in record review studies); the standardized items facilitated the testing of our hypotheses about variation in care; and we have the physicians' own judgments concerning current practices in the treatment of PHIs. The neurosurgeons responded at rates found in mail surveys of other populations and the questionnaires appeared to have been thoughtfully completed. The heavy time commitments of neurosurgeons and the length and complexity of the sur~ey (five pages of detailed questions) had led us to expect lower return rates. On the contrary, many surgeons provided lengthy additional assessments. Although the survey results are not necessarily representative of all neurosurgeons, we are confident that the data reflect the variation of opinion concerning the care to be given PHI patients. Neurosurgeons differ in their opinions of what constitutes appropriate management of various aspects of PHI, such as surgical debridement, use of ICP monitoring, and nonoperative treatment modalities. On the other hand, there was consensus on other issues, including various surgical indications. The presence of neurosurgery residents was the most significant demographic factor affecting their choices. However, the neurosurgeons' assessment of a patient's salvageability determined the decision of whether or not to operate.
Penetrating Head Injury Care Survey
Surg Neurol 1991;36:370-7
375
Table 7. Influences on Management of Penetrating Head Injury
Intercept T e s t for c o a g u l o p a t h y U s e o f steroids
.491 -.372
Experience~
PHI patients/year h
Level I trauma center
NS
NS
-.395
- .031 (.0004) NS
PHI assault (%y
NS
.009 (.001) .010
NS
(.0004) NS
Use of ultrasound
-.893
NS
(.0185 NS
Use of ICP monitors on GCS 6-8 Use of barbiturate coma
-.720
NS
NS
NS
NS
-.688
NS
NS
NS
NS
Nsurg resident ~
Model yl"z (p value)
1.206 (.0001) - 1.002
70.20,3 df (.0001) 46.77,3 df
(.00015 1.083 (.00015 1.017 (.00015 .678 (.OOl)
(.0001) 41.43,1 df (.0001) 34.57,1 df (.0001) 16.74,1 df (.OOl)
Abbreviation: df, degrees of freedom; NS, not significant. Analysis was carried out using logistic regression analysis. Beta ( P value~ Number of years since end of residency. b 1-9 versus 10 + PHI patients treated/year. Percent of PHI patients who were victims of assault. d Neurosurgical residents present in practice.
The variation in the care provided to PHI patients may be partly explained by several factors: 1. The traditional military standard of care for PHI emphasizes early surgical debridement, whereas the modern civilian approach seems derived from the treatment of closed head injuries and is based on the medical control of ICP and the use of newer antibiotics for contamination. 2. Many treatments have not been subjected to rigorous testing; thus, neurosurgeons differ in their opinions regarding different treatments, and they vary in practice as well. 3. Residents find it easier to provide more aggressive and innovative care and are encouraged to do so. The fact that there is such variation in the approach to the treatment of gunshot wounds to the head is not unreasonable when one considers certain historical facts. Current attitudes, including the military's emphasis on early debridement, are likely based by some on an extensive experience going back to World War I [3,5-7,9, 10,12-14,17-20,26,29,34-39,42-46,48]. However, such attitudes, based on opinions generated many years ago, have been modified by current experience. For example, early resuscitation and rapid transport to trauma centers, as well as prevention and treatment of complications, permit salvage of certain patients who would previously have died. We have argued that the common view and teaching about PHI is overly pessimistic and that failure to debride mass lesions on some occasions may lead to the inappropriate fulfillment of the prediction of a poor outcome [2 2,2 3,2 5 ]. Although difficulty in comparing different hospital experiences made detailed analysis impossible, we did perceive a trend toward bet-
ter survival rates in series with higher operative rates [5,6,14,19-21,26,34,36,42,48]. Our data suggest that this may in turn reflect opinions regarding salvageability. The recent literature has described optimal treatment of severe head injuries as including early resuscitation with intubation and rapid transport to a center with the capability for definitive treatment [47], use of a CT scanner to detect clots, even in the presence o f artifacts from bullets [8], early removal of mass lesions [41], and management of ICP elevations [ 1,4,31,40] that includes the use o f barbiturate coma [11]. N e w antibiotics may be more effective in preventing infections. The use of anticonvulsants has been improved by the ability to measure blood levels. At the time of this survey, steroids had been found inefficacious, and indeed in some respects harmful [ 15,16,32]; 5 8% o f respondents, particularly those in academic centers, do not use them. The situation may change again in light of the recent demonstration of the potential value of large doses of methylprednisolone in spinal cord injury [30]. The survey indicated the neurosurgeons' concern about their own ability to make educated decisions. A number of respondents discussed their opinions about the treatment o f PHI. Some expressed frustration over the lack of information about the outcome o f P H I treatments and pointed to the need for research. Even though many of the severely injured cases are not seen as very salvageable, many neurosurgeons feel treatment is the only hope and they worry about patients who might have had good survival if full treatment had been given. On the other hand, others worry about the devastating effects upon family members, society, and the patient if he or she survives with severe handicaps. Finally, the existence of consensus does not necessarily indicate optimum treatment. For example, although
376
Surg N e u r o l 1991;36:370-7
our survey shows consensus on the removal of bone fragments, recent research indicates that this may not be needed [28,33]. We feel strongly that results of this survey should not be used to imply or suggest a particular "optimum" or "standardized" approach to PHI management. Properly conducted clinical studies remain the principal guide for establishing optimal care. The views expressed are those of the authors and do not necessarily represent those of the Department of Defense. This study was sponsored by USAMR&DC contract #87PP7824. We would like to thank the responding neurosurgeons for their willingness to participate and their thoughtful comments. Master Chief (Ret.) H. Brown provided technical help. Susan Budinsky, Joyce Herschberger, and Robin Metheny provided secretarial support. The American Association of Neurological Surgeons and the Congress of Neurological Surgeons kindly supplied the list of neurosurgeons and their addresses.
References 1. Becker DP, Miller JD, Ward JD, Greenberg RP, Young HF, Sakalas R. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 1977;47:491-502. 2. Bohigan GM. Firearms as a public health problem in the United States: injuries and deaths. Action of the AMA House of Delegates Interim 1987 Session. American Medical Association, Council on Scientific Affairs Reports: A(I-87), 1987, 20 pp. 3. Brierbrauer K, Tindall SC. Gunshot wounds to the head and spine. Contemp Neurosurg 1986;9(20):1-5; 9(21):1-6. 4. Bruce DA, Schut L, Bruno LA, Wood JH, Sutton LN. Outcome following severe head injuries in children. J Neurosurg 1978;48:679-88. 5. Byrnes DP, Crockard HA, Gordon DS, Gleadhill OA. Penetrating craniocerebrai missile injuries in the civil disturbances in Northern Ireland. B r J Surg 1974;61:169-76. 6. Clark WC, Muhlbauer MS, Watridge CB, Morris WR. Analysis of 76 civilian craniocerebral gunshot wounds. J Neurosurg 1986;65:9-14. 7. Cooper PR. Gunshot wounds of the brain. In: Cooper PR, ed. Head injury. New York: Williams & Wilkins, 1987:313-26. 8. Cooper PR, Maravilla K, ConeJ. Computerized tomographic scan and gunshot wounds of the head: Indications and radiographic findings. Neurosurgery 1979;4:373-80. 9. Crockard HA. Bullet injuries of the brain. Ann Roy Coil Surg Engl 1974;55:111-23. 10. Crockard HA. Early intracranial pressure studies in gunshot wounds of the brain. J Trauma 1975;15:339-47. 11. Eisenberg HM, Frankowski RF, Contant CF, Marshall LF, Walker MD, and the Comprehensive Central Nervous System Trauma Centers. High-dose barbiturate control of elevated intracranial pressure in patients with severe head injury. J Neurosurg 1988;69:15-23. 12. George ED, Dagi TF. Penetrating missile injuries of the head. In: Schmidek HH, Sweet WH, eds. Operative neurosurgical techniques: indications, methods, and results. Orlando: Grune & Stratton, 1988:49-56. 13. Gianotta SL, Weiss MH, Apuzzo MJ, Martin E. High dose glucocorticoids in the management of severe head injury. Neurosurgery 1984;15:497-501. 14. Goodman JM, Kalsbeck J. Outcome of self-inflicted gunshot wounds of the head. J Trauma 1965;5:636-42.
Kaufman et al
15. Gordon DS. Missile wounds of the head and spine. Br Med J 1975;1:614-16. 16. Gudeman SK, MillerJD, Becker DP. Failure of high-dose steroid therapy to influence intracranial pressure in patients with severe head injury. J Neurosurg 1979;51:301-6. 17. Haddad FS. Nature and management of penetrating head injuries during the civil war in Lebanon. CanJ Surg 1978;21:233-8. 18. Harsh GRIII, Harsh GR IV. Penetrating wounds of the head. In: Wilkins RH, Rengachary SS, eds. Neurosurgery. Vol 2. New York: McGraw Hill, 1985:1670-8. 19. Hernesniemi J. Penetrating craniocerebral gunshot wounds in civilians. Acta Neurochir 1979;49:199-205. 20. Hubschmann O, Shapiro K, Baden M, Shulman K. Craniocerebral gunshot injuries in civilian practice: prognostic criteria and surgical management: Experience with 82 cases. J Trauma 1979;19: 6-12. 21. Kaufman HH. Treatment of civilian gunshot wounds to the head. In: Dagi TF, ed. Gunshot wounds to the head. Boston: Little Brown: in press. 22. Kaufman HH, Levin HS, High WM Jr, Childs TL, Wagner KA, Gildenberg PL. Neurobehavioral outcome after gunshot wounds to the head in civilians. Neurosurgery 1985;16:754-8. 23. Kaufman HH, Loyola WP, Makela ME, Frankowski RF, Wagner KA, Bernstein DP, Gildenberg PL. Civilian gunshot wounds: the limits of salvageability. Acta Neurochir 1983 ;67:115-25. 24. Kaufman HH, Lynn J. Brain death. Neurosurgery 1986;19: 850-6. 25. Kaufman HH, Makela ME, Lee KF, Haid RW, Gildenberg PL. Gunshot wounds to the head: a perspective. Neurosurgery 1986;18:689-95. 26. Kaufman HH, Mattson JC. Coagulopathy in head injury. In: Becker DP, ed. Central nervous system trauma and research status report. Richmond: William Byrd Press, 1985:187-206. 27. Kaufman HH, MoakeJL, OlsonJD, Miner ME, deCret RP, Pruessner JL, Gildenberg PL. Delayed and recurrent intracranial hematomas related to disseminated intravascular clotting and fibrinolysis in head injury. Neurosurgery 1980;7:445-9. 28. Brandvold B, Levi L, Feinsod M, George E. Penetrating craniocerebral injuries in the Israeli involvement in the Lebanese conflict, 1982-1985: analysis of a less aggressive surgical approach. J Neurosurg 1990;72:15-21. 29. Lillard PL. Five years experience with penetrating craniocerebral gunshot wounds. Surg Neurol 1978;9:79-83. 30. Bracken MB. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. N Engl J Med 1990;332:1405-11. 31. Miller JD, Butterworth JF, Gudeman SK, Faulkner JE, Choi SC, Selhorst JB, Harbison JW, Lutz HA, Young HF, Becker DP. Further experience in the management of severe head injury. J Neurosurg 1981;54:289-99. 32. Molofsky WJ. Steroids and head trauma. Neurosurgery 1984;15: 424-6. 33. Myers PW, Brophy J, Salazar AM, Jonas B. Retained bone fragments after penetrating head injury. Read before the American Association of Neurological Surgeons, Washington, DC, 1989. 34. Nagib MG, Rockswold GL, Sherman RS, Lagaard MW. Civilian gunshot wounds to the brain: prognosis and management. Neurosurgery 1986;18:533-7. 35. Owen-Smith MS. High-velocity and military gunshot wounds. In: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988:61-94. 36. Raimondi AJ, Samuelson GH. Craniocerebral gunshot wounds in civilian practice. J Neurosurg 1970;32:647-53.
Penetrating Head Injury Care Survey
37. Robertson CS, Clifton GL, Grossman RG. Oxygen utilization and cardiovascular function in head-injured patients. Neurosurgery 1984;15:307-14. 38. Rybek B. Missile wounding and hemodynamic effects of energy absorption. Acta Chir Scand 1974;450:1-32. 39. Saba MI. Surgical management of gunshot wounds of the head. In: Schmidek HH, Sweet WH, eds. Operative neurosurgical techniques: indications, methods, and results. Orlando: Grune & Stratton, 1988:37-48. 40. Saul T. Is ICP monitoring worthwhile? Clin Neurosurg 1988;34:560-71. 41. SeeligJM, Becker DP, MillerJD, Greenberg RP, Ward JD, Choi SC. Traumatic acute subdural hematoma. N Engl J Med 1981;304:1511-18. 42. Sherman WD, Apuzzo MLJ, Heiden JS, Petersons VT, Weiss MH. Gunshot wounds to the brain: a civilian experience. WestJ Med 1980;132:99-105.
Surg Neurol 1991;36:370-7
377
43. Shoung HM, SichezJP, Pertuiset B. The early prognosis of craniocerebral gunshot wounds in civilian practice as an aid to the choice of treatment. Acta Neurochir 1985;74:27-30. 44. Swan KG, Swan RC. The head. In: Swan KG, Swan RC, eds. Gunshot wounds: pathophysiology and management. Littleton, MA: PSG Publishing Company, 1980:29-55. 45. Villanueva PA. Cranial gunshot wounds, in: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988: 257-66. 46. West CG. A short history of the management of penetrating missile injuries of the head. Surg Neurol 1981;16:145-9. 47. West JG. The role of trauma centers in treating victims with gunshot wounds. In: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988:13-24. 48. Yashon D, Jane JA, Martonffy D, White RJ. Management of civilian craniocerebral bullet injuries. Am Surg 1972;346-51.
Surg Neurol 1991;36:370-7
A National Survey of Neurosurgical Care for Penetrating Head Injury Howard H. Kaufman, M.D., Karen Schwab, M.A., and Andres M. Salazar, M.D. Department of Neurosurgery, West Virginia University, Morgantown, West Virginia, and Head Injury Unit, Uniformed Services University of the Health Sciences, Bethesda, Maryland
Kaufman HH, Schwab K, Salazar AM. A national survey of neurological care for penetrating head injury. Surg Neurol 1991;36:370-7.
We report results of a survey on the management patterns of penetrating head injury (PHI). American neurosurgeons (N = 2969) were asked to participate in a mail survey. One thousand one hundred twenty-eight responded, providing detailed information about their practices, their opinions concerning diagnostic testing, nonoperative therapy, and surgical debridement for PHI. Although there was agreement on some areas of management an~l care, including the use of computed tomography scanning, antibiotics, anticonvulsants, and a few surgical indications/contraindications, there was wide variation on a significant number of points. These included testing for coagulopathy; use of corticosteroids, intracranial pressure monitoring, and barbiturate coma; and surgical indications including debridement of contusions, removal of fragments, location of injury, and Glasgow Coma Score 5-8. Many nonoperative decisions were influenced by the presence of neurosurgical residents, whereas the decision to operate was based primarily on judgment of patient salvageability. This survey uncovered disagreement on several crucial issues in the care of patients with penetrating head injury and the need for continued research in this area of patient care. KEY WORDS; Survey; Neurosurgeons; Surgical debridement; Penetrating head injury; Gunshot wounds; Military
If trauma can be considered "the hidden epidemic," gunshot wounds are probably its most obscure subcategory, despite the fact that firearms are among the 12 leading causes of death in the United States and the top 10 causes of accidental death. Firearms are the etiology of 58% of homicides and 57% of suicides [2]. T h e fre-
Address reprint requests to: Howard H. Kaufman, M.D., Professor and Chairman, Department of Neurosurgery, Health Sciences Center, West Virginia University, Morgantown, West Virginia 26506. Received August 20, 1990; accepted May 20, 1991.
© 1991 by Elsevier Science Publishing Co., Inc.
quency of nonfatal injuries is not available. A large proportion of the injuries involve the brain [23], and an overwhelming percentage of brain injuries immediately or rapidly result in death [19,23]. Informal discussions over the last several years led us to believe that there might be an important variability in some aspects of the care in gunshot wounds to the head, including surgical decision making, medical management, and ethical issues. To identify the variations, if any, in the care of penetrating head injuries, and as part of a larger, ongoing US Army-sponsored multicenter study of acute penetrating head injury (PHI), we carried out a national survey of neurosurgeons, asking about their approach to various aspects of the care o f victims of gunshot wounds to the head. W h e r e we found variability, we tried to determine what factors might have led to that variability. Methods The questionnaire we used was designed initially by the three investigators to examine demographic information about the neurosurgeons and their approach to the care of patients with P H I (Table 1). It was informally pretested on approximately 20 colleagues. After adjustment, it was pretested formally on three independent random samples of 100 neurosurgeons each. These neurosurgeons' names were obtained f r o m lists supplied by the American Association of Neurological Surgeons and the Congress of Neurosurgeons. Questionnaires were sent only to neurosurgeons living in the United States and its territories. T h e pretest revealed only minor difficulties, and appropriate adjustments to the questionnaires were again made. T h e final questionnaire consisted of 43 questions, although several questions had a n u m b e r of subquestions. Most required closed-end responses, but some were open ended. W e then sent the new questionnaires to the remaining neurosurgeons on the lists. Those that did not return the mailings were sent a second questionnaire. W e eventually received a total of 1128 replies from 2969 neurosurgeons (38%); 966 (from neurosurgeons who had been in active prac0090-3019/91/$3.50
Penetrating Head Injury Care Survey
Surg Neurol 1991;36:370-7
Table 1. Areas Investigated in Questionnaire Demographics
Age Time from residency Military/combat experience Size of city Section of country Size/type of hospital Trauma-center level Number of PHIs/year Assault/suicide/accident Type of weapon Neurosurgery residents
Patient care Screening for coagulopathy Computed tomography
Antibiotics Anticonvulsants Steroids Reason to operate/not operate Surgical timing/technique Intracranial pressure monitoring Brain-death criteria Organ retrieval Social/ethical issues Legal issues
371
teaching centers. The average neurosurgeon saw nine penetrating head-injury patients a year, but this was highly variable (SD = 16.6). Most saw mainly hand-gun injuries (81% -+ 23) as opposed to rifles (10% -+ 16). In the average experience, assaults (42% -+ 29) and suicides (45% -+ 28) were seen with equal frequency, whereas accidents were more unusual (15% -+ 20). N e u rosurgeons in larger cities, at larger hospitals, or at hospitals with level I trauma centers saw larger numbers of patients. The smaller the city, the more likely that the injury was related to suicide. Conversely, homicides were more c o m m o n in larger cities.
Surgical Treatment tice within the past 2 years and had seen patients with P H I ) were used for this analysis. The responses were double-entry key punched and analyzed with an I B M 370 using SAS Institute Software, Version 5, 1985. Bivariate and more complex analyses were carried out and tested, using logistic regression. We defined consensus in treatment as occurring when 75~, or more of neurosurgeons agreed on a particular point in the care of patients with P H I . For example, we asked how strongly hematomas influenced them to do or not to do surgery. Because over 75% (ie, 9 2 % ) said they are influenced to do surgery in such patients, we considered operations in these cases to be the typical approach taken by neurosurgeons today. We also noted in which categories there was a 60% to 74% concurrence. W h e n we found variation (lack of consensus) we analyzed whether certain characteristics of the neurosurgeons and/or their practices might account for their diagnostic or treatment choices. We determined whether variation in the use of tests/treatments was associated with length of time since the neurosurgeon had finished his or her residency, n u m b e r of P H I cases seen annually, whether or not the hospital was a level I trauma center, percent of P H I cases due to assault, and whether there were neurosurgery residents in the hospital.
One of the principal questions studied was what determined the decision to treat or not to treat with intracranial surgical debridement. To demonstrate the responses, we have reproduced the questions, collapsed the data from seven categories to three (influence against surgery, neutral, influence for surgery), and indicated which questions received consensus answers, that is, 75% or m o r e agreement that they were influenced for or against surgery, or were neutral (Table 3). O f the 29 reasons presented in the questionnaire for operating/not operating on patients with P H I s , neurosurgeons reached consensus on eight, and 6 0 % to 74% agreed on another eight (Table 3). Consensus was seen about the significance of low Glasgow C o m a Score (GCS) categories, bilateral pupillary dilation, cerebrospinal fluid leaks, depressed fractures, removal of bone fragments, and removal of hematomas. T h e r e was disagreement on the need for surgery in those patients with the middle range of GCS. M o r e than half of surgeons did not feel it was necessary to operate a second time to r e m o v e retained bone or bullet frag-
Table 2. Demographic Characteristics of Respondents City s i z e < i00,000 100,000-500,000
Results The average age of the respondents was 48 _+ 9 years. They had been out of training 15 -+ 9 years. Twenty-six percent had been in the military, but only 6 % had seen combat. Respondents tended to practice in larger cities compared with the location of all neurosurgeons, but otherwise spread with the population geographically (Table 2). Respondents generally practiced in larger hospitals (Table 2 ) 6 6 % in private hospitals and 2 6 % in resident
500,000-1,000,000 > 1,000,000 Region Northeast South Central West Beds <400 400-800 >800
Neurosurgeons (%)
Population(%)
20
75
34
13
18 28
5 8
Neurosurgeons (%)
Population(%)
18 36 23 23
21 34 25 20
Neurosurgeons (%)
All hospitals (%)
39 49 12
88 10 2
372
Surg Neurol 1991;36:370-7
T a b l e 3.
Kaufman et al
Percent of Neurosurgeons Indicating the Effect of Different Parameters on Their Decisions to Operate a No surgery
Neutral
Surgery
Consensus
Hematoma Contusion Edema with mass effect
1 39 46
7 32 23
92 29 31
+
Remove bone fragments Bone fragments remaining after first debridement Remove bullet or metal fragments Metal fragments remaining after first debridement Copper-jacketed bullets
4 31
6 21
90 49
+
45 71
24 19
31 10
_+
39
26
36
Medicolegal pressure To reassure family
45 49
34 33
21 18
To prevent delayed edema or infection To prevent or repair CSF leak Elevation of depressed fracture
16 4 3
14 8 7
69 88 90
-+ + +
GCS GCS GCS GCS GCS GCS
82 73 50 12 7 10
10 16 29 34 16 14
8 12 21 55 77 77
+ _+
One pupil dilated Both pupils dilated
10 77
19 14
72 9
___ +
Location Dominant hemisphere Deep brain Cross-midline Cross-ventricle
32 70 60 52
39 20 26 32
29 10 13 17
_+ -+
Did not respond to resuscitation Hyperventilation Mannitol
68 68
20 19
12 13
_+ -+
Age (eg, 65 and older) Suicide
44 32
40 52
16 16
3 4 5 6-8 9-12 13-15
+ +
Abbreviations: +, consensus reached by >75% in one category; -+, 60%-74% agreement; GCS, Glasgow Coma Score. Respondents were asked to make the following evaluation: "Reasons why you would or would not operate on penetrating wounds to the head. Please consider each item in isolation from the others. How strongly does this influence you to operate?" The answers were labeled as 0-2 = no surgery, 3 = neutral, 4-6 = surgery.
ments. There was no agreement on the importance of the location of the bullet in the dominant hemisphere or its having crossed the ventricles, although most neurosurgeons do not operate on deep wounds or those that have crossed the midline. Likewise, the patient's age, eg, 65 years or older, or whether the patient was a suicide attempt did not usually influence them on the question of operating or not operating. We wondered whether the neurosurgeons' practice characteristics or their general view on the salvageability or nonsalvageability of PHI patients explained some of the variation that occurs in surgical decisions. Nonsalvageable patients were defined by 68% of respondents as those expected to die, whereas 30% added those expected to survive in a vegetative state, and only 2%
added those expected to survive with a severe disability. We asked, "In your experience, what percentage of the following penetrating head injured patients are ultimately salvageable?" In general, we found that outlook for P H I patients would vary by admission GCS level. Salvageability is generally not expected with GCS 3 and 4, but it is in patients with 9 - 1 2 and 1 3 - 1 5 , although perhaps not to the extent reported in the literature (Figure 1). However, there was considerable variation around the average responses for patients in the GCS 5 - 8 range. Assessments o f salvageability did not vary by number of patients seen, military service of the neurosurgeon, or level of the trauma center. The infection rate of the reporting neurosurgeon correlated with his or her per-
Penetrating Head Injury Care Survey
T a b l e 4. Percent of Responding Neurosurgeons Using Various Neurosurgical Evaluations and Treatment for Penetrating Brain Wounds
100
80
~h
go
373
Surg Neurol 1991;36:370-7
Respondents
60
40
20
3
4 Glasgow
5
9-12
6-8 Coma
13-15
Score
Figure 1. The meanpercent of patients felt to be salvageableand the mean percent treated surgically for a given Glasgow Coma Scoreor group of coma
scores. N = 1015 neurosurgeons +- 1 SD.
ception of the ultimate salvageability of patients with admission GCS 9-15. Multiple logistic regression analysis showed that the neurosurgeons' perceptions of patient salvageability was an important predictor of tendency to operate. In fact, of the variables we examined, it was the only factor significantly associated with tendency to operate on severely injured patients (GCS 8 or less). The larger the percentage of patients seen as ultimately salvageable within an injury class (ie, GCS levels), the more likely the neurosurgeon was to report a tendency to operate on that group of patients. The variable most significantly related to tendency to operate in the less severely injured patient was length of time since the neurosurgeon's residency. That relationship was negative, meaning the longer the neurosurgeon had been in practice, the less likely he or she was to operate on patients with admission GCS 9-15. Surgical technique is fairly standardized. This includes craniotomy (83%) or craniectomy (82%), following the track deep into the brain (75%), and patching the dura (86%), especially with autogenous tissue (79%). Over half (532) do not use artificial dura, whereas 48% never use human prepared dura. As mentioned before, 36% use ultrasound fairly frequently. According to 68%, surgery should be performed as soon as possible, and another 15% felt it should be performed within a day.
Nonsurgical Treatment and Testing Another set of questions probed the incidence of use of various tests and treatments (Table 4).
Do Do Do Do
you usually or always test for coagulopathy you evaluate with CT you use prophylactic antibiotics you use anticonvulsants For 6 months For 1 year For 2 years Longer Do you use corticosteroids routinely Do you use ultrasound Do you usually or always use ICP monitoring
65 97 87 84 30 43 19 9 42 36
If GCS 3-5 If GCS 6 - 8 If GCS 9-12 I f G C S 13-15 Do you use barbiturate coma Threshold pressure (mm Hg) 15-20 21-25 26-30 31-35 36-40 ~40
30 40 25 16 37 6 23 31 17 10 12
Abbreviations: CT, computed tomography; GCS, Glasgow Coma Score; ICP, intracranial pressure.
Testing for coagulopathy [6,26,27] was done frequently (65% usually or always). The most common tests were the prothrombin time, activated partial thromboplastin time, and platelet count. Most did not use thrombin time, fibrinogen split products, or fibrinogen. Computed tomography (CT) scanning [8] was used by 97%, although a few seemed to use it less to make the decision of whether or not to operate (ie, they used it more for surgical planning). Prophylactic antibiotics were generally used (87%), but the duration of treatment varied considerably (Table 5). Multiple antibiotics were used by 438 respondents: two by 320 respondents, three by 80 respondents, and four or more by 32 respondents. The most commonly used antibiotics were cepha-
T a b l e 5. Distribution of Neurosurgeons Who Routinely Use Antibiotics in Patients Expected to Survive and Number of Days
Used ( N -- 750) Neurosurgeons Days
(%)
1-3 4-5 6-7 8-10 ~10
24 19 20 25 13
374
Surg Neurol 1991;36:370-7
Kaufman et al
Table 6. EstimatedInfection Rates (N = 812) Infection(%)
Respondentswho estimate this infectionrate (%)
0
21
1-5 6-10 11-25 -->26
33 28 12 6
losporins (440), chloramphenicol (176), penicillins (117), aminoglycosides (88), vancomycin (47), miconazole (26), tetracycline (20), and erythromycin (19). Although they were asked about routine coverage, it is not clear that each respondent used multiple antibiotics in each patient. Estimation of infection rates for penetrating wounds also varied (Table 6). Anticonvulsants were used prophylactically by 87% of respondents, and 98% of these used phenytoin. Of these, 88% loaded the drug, 60% adjusted by blood level, while 20% gave 300 mg/day and 20% gave 400 mg/day. Thirty-five percent used phenobarbital, generally in combination with phenytoin. Medications were used most often for 1 year or less (Table 4). Twenty-two percent switched to carbamazepine. Corticosteroids were used routinely by 42% of respondents. Thirty-six percent used ultrasound at the time of surgery. Intracranial pressure (ICP) monitoring remained controversial. A few respondents indicated that they might use it without debridement (13 % usually or always), while a few (7% usually or always) used it to determine whether or not to operate. Forty percent of neurosurgeons reported using ICP monitoring "usually" or "always" in patients with a GCS of 6-8 (Table 4). Only 37% ever used barbiturate coma, of whom 83% used it with ICP monitoring. The threshold pressure at which it was used was variable (Table 4). Multivariate regression analysis of which physician/ practice characteristics determined the decision to use tests/treatments where there was no consensus showed that the presence of neurosurgical residents was the most important (Table 7). Neurosurgeons practicing in centers with neurosurgery residents report more frequent testing for coagulopathy; less frequent use of steroids; and more frequent use of intraoperative ultrasound, ICP monitoring of patients with GCS 6-8, and barbiturate coma. Management of "Unsalvageable" Patient~Brain Death Criteria Eighty percent of neurosurgeons will sometimes to always put "unsalvageable" patients into an intensive care
unit, although only 70% will sometimes to always ventilate them. Decisions to carry out such treatment are influenced by questions of legal liability (66% sometimes to always), to show the family something is being done (66% sometimes to always), and for organ procurement (89% sometimes to always). Most respondents (85%) say they do not treat suicides differently. Of the many brain-death criteria [24], 48% used the Harvard criteria of 1986, 6% used the National Institutes of Health Combined Study criteria of 1977, and 12% used the criteria of the President's Commission of 1981. No particular characteristics of physicians or their practices distinguished those who used the most recently proposed criteria, those of the President's Commission. Thirty-four percent used other criteria. Eighty-one percent of respondents' hospitals had specific brain-death criteria. Seventy-five percent stated they had a hospital policy regarding organ retrieval. Ninety-one percent felt their medical examiner is "sometimes to always" cooperative with retrieval. Sixty-six percent "sometimes to always" request donation in homicide, and 70% feel they are successful in half or less of these cases; 86% "sometimes to always" request retrieval in suicide, and 65% feel they are successful in half or less of these cases. Discussion
The advantages of the survey approach are considerable: we were able to obtain a national sample of physician practices (very difficult to obtain in record review studies); the standardized items facilitated the testing of our hypotheses about variation in care; and we have the physicians' own judgments concerning current practices in the treatment of PHIs. The neurosurgeons responded at rates found in mail surveys of other populations and the questionnaires appeared to have been thoughtfully completed. The heavy time commitments of neurosurgeons and the length and complexity of the sur~ey (five pages of detailed questions) had led us to expect lower return rates. On the contrary, many surgeons provided lengthy additional assessments. Although the survey results are not necessarily representative of all neurosurgeons, we are confident that the data reflect the variation of opinion concerning the care to be given PHI patients. Neurosurgeons differ in their opinions of what constitutes appropriate management of various aspects of PHI, such as surgical debridement, use of ICP monitoring, and nonoperative treatment modalities. On the other hand, there was consensus on other issues, including various surgical indications. The presence of neurosurgery residents was the most significant demographic factor affecting their choices. However, the neurosurgeons' assessment of a patient's salvageability determined the decision of whether or not to operate.
Penetrating Head Injury Care Survey
Surg Neurol 1991;36:370-7
375
Table 7. Influences on Management of Penetrating Head Injury
Intercept T e s t for c o a g u l o p a t h y U s e o f steroids
.491 -.372
Experience~
PHI patients/year h
Level I trauma center
NS
NS
-.395
- .031 (.0004) NS
PHI assault (%y
NS
.009 (.001) .010
NS
(.0004) NS
Use of ultrasound
-.893
NS
(.0185 NS
Use of ICP monitors on GCS 6-8 Use of barbiturate coma
-.720
NS
NS
NS
NS
-.688
NS
NS
NS
NS
Nsurg resident ~
Model yl"z (p value)
1.206 (.0001) - 1.002
70.20,3 df (.0001) 46.77,3 df
(.00015 1.083 (.00015 1.017 (.00015 .678 (.OOl)
(.0001) 41.43,1 df (.0001) 34.57,1 df (.0001) 16.74,1 df (.OOl)
Abbreviation: df, degrees of freedom; NS, not significant. Analysis was carried out using logistic regression analysis. Beta ( P value~ Number of years since end of residency. b 1-9 versus 10 + PHI patients treated/year. Percent of PHI patients who were victims of assault. d Neurosurgical residents present in practice.
The variation in the care provided to PHI patients may be partly explained by several factors: 1. The traditional military standard of care for PHI emphasizes early surgical debridement, whereas the modern civilian approach seems derived from the treatment of closed head injuries and is based on the medical control of ICP and the use of newer antibiotics for contamination. 2. Many treatments have not been subjected to rigorous testing; thus, neurosurgeons differ in their opinions regarding different treatments, and they vary in practice as well. 3. Residents find it easier to provide more aggressive and innovative care and are encouraged to do so. The fact that there is such variation in the approach to the treatment of gunshot wounds to the head is not unreasonable when one considers certain historical facts. Current attitudes, including the military's emphasis on early debridement, are likely based by some on an extensive experience going back to World War I [3,5-7,9, 10,12-14,17-20,26,29,34-39,42-46,48]. However, such attitudes, based on opinions generated many years ago, have been modified by current experience. For example, early resuscitation and rapid transport to trauma centers, as well as prevention and treatment of complications, permit salvage of certain patients who would previously have died. We have argued that the common view and teaching about PHI is overly pessimistic and that failure to debride mass lesions on some occasions may lead to the inappropriate fulfillment of the prediction of a poor outcome [2 2,2 3,2 5 ]. Although difficulty in comparing different hospital experiences made detailed analysis impossible, we did perceive a trend toward bet-
ter survival rates in series with higher operative rates [5,6,14,19-21,26,34,36,42,48]. Our data suggest that this may in turn reflect opinions regarding salvageability. The recent literature has described optimal treatment of severe head injuries as including early resuscitation with intubation and rapid transport to a center with the capability for definitive treatment [47], use of a CT scanner to detect clots, even in the presence o f artifacts from bullets [8], early removal of mass lesions [41], and management of ICP elevations [ 1,4,31,40] that includes the use o f barbiturate coma [11]. N e w antibiotics may be more effective in preventing infections. The use of anticonvulsants has been improved by the ability to measure blood levels. At the time of this survey, steroids had been found inefficacious, and indeed in some respects harmful [ 15,16,32]; 5 8% o f respondents, particularly those in academic centers, do not use them. The situation may change again in light of the recent demonstration of the potential value of large doses of methylprednisolone in spinal cord injury [30]. The survey indicated the neurosurgeons' concern about their own ability to make educated decisions. A number of respondents discussed their opinions about the treatment o f PHI. Some expressed frustration over the lack of information about the outcome o f P H I treatments and pointed to the need for research. Even though many of the severely injured cases are not seen as very salvageable, many neurosurgeons feel treatment is the only hope and they worry about patients who might have had good survival if full treatment had been given. On the other hand, others worry about the devastating effects upon family members, society, and the patient if he or she survives with severe handicaps. Finally, the existence of consensus does not necessarily indicate optimum treatment. For example, although
376
Surg N e u r o l 1991;36:370-7
our survey shows consensus on the removal of bone fragments, recent research indicates that this may not be needed [28,33]. We feel strongly that results of this survey should not be used to imply or suggest a particular "optimum" or "standardized" approach to PHI management. Properly conducted clinical studies remain the principal guide for establishing optimal care. The views expressed are those of the authors and do not necessarily represent those of the Department of Defense. This study was sponsored by USAMR&DC contract #87PP7824. We would like to thank the responding neurosurgeons for their willingness to participate and their thoughtful comments. Master Chief (Ret.) H. Brown provided technical help. Susan Budinsky, Joyce Herschberger, and Robin Metheny provided secretarial support. The American Association of Neurological Surgeons and the Congress of Neurological Surgeons kindly supplied the list of neurosurgeons and their addresses.
References 1. Becker DP, Miller JD, Ward JD, Greenberg RP, Young HF, Sakalas R. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 1977;47:491-502. 2. Bohigan GM. Firearms as a public health problem in the United States: injuries and deaths. Action of the AMA House of Delegates Interim 1987 Session. American Medical Association, Council on Scientific Affairs Reports: A(I-87), 1987, 20 pp. 3. Brierbrauer K, Tindall SC. Gunshot wounds to the head and spine. Contemp Neurosurg 1986;9(20):1-5; 9(21):1-6. 4. Bruce DA, Schut L, Bruno LA, Wood JH, Sutton LN. Outcome following severe head injuries in children. J Neurosurg 1978;48:679-88. 5. Byrnes DP, Crockard HA, Gordon DS, Gleadhill OA. Penetrating craniocerebrai missile injuries in the civil disturbances in Northern Ireland. B r J Surg 1974;61:169-76. 6. Clark WC, Muhlbauer MS, Watridge CB, Morris WR. Analysis of 76 civilian craniocerebral gunshot wounds. J Neurosurg 1986;65:9-14. 7. Cooper PR. Gunshot wounds of the brain. In: Cooper PR, ed. Head injury. New York: Williams & Wilkins, 1987:313-26. 8. Cooper PR, Maravilla K, ConeJ. Computerized tomographic scan and gunshot wounds of the head: Indications and radiographic findings. Neurosurgery 1979;4:373-80. 9. Crockard HA. Bullet injuries of the brain. Ann Roy Coil Surg Engl 1974;55:111-23. 10. Crockard HA. Early intracranial pressure studies in gunshot wounds of the brain. J Trauma 1975;15:339-47. 11. Eisenberg HM, Frankowski RF, Contant CF, Marshall LF, Walker MD, and the Comprehensive Central Nervous System Trauma Centers. High-dose barbiturate control of elevated intracranial pressure in patients with severe head injury. J Neurosurg 1988;69:15-23. 12. George ED, Dagi TF. Penetrating missile injuries of the head. In: Schmidek HH, Sweet WH, eds. Operative neurosurgical techniques: indications, methods, and results. Orlando: Grune & Stratton, 1988:49-56. 13. Gianotta SL, Weiss MH, Apuzzo MJ, Martin E. High dose glucocorticoids in the management of severe head injury. Neurosurgery 1984;15:497-501. 14. Goodman JM, Kalsbeck J. Outcome of self-inflicted gunshot wounds of the head. J Trauma 1965;5:636-42.
Kaufman et al
15. Gordon DS. Missile wounds of the head and spine. Br Med J 1975;1:614-16. 16. Gudeman SK, MillerJD, Becker DP. Failure of high-dose steroid therapy to influence intracranial pressure in patients with severe head injury. J Neurosurg 1979;51:301-6. 17. Haddad FS. Nature and management of penetrating head injuries during the civil war in Lebanon. CanJ Surg 1978;21:233-8. 18. Harsh GRIII, Harsh GR IV. Penetrating wounds of the head. In: Wilkins RH, Rengachary SS, eds. Neurosurgery. Vol 2. New York: McGraw Hill, 1985:1670-8. 19. Hernesniemi J. Penetrating craniocerebral gunshot wounds in civilians. Acta Neurochir 1979;49:199-205. 20. Hubschmann O, Shapiro K, Baden M, Shulman K. Craniocerebral gunshot injuries in civilian practice: prognostic criteria and surgical management: Experience with 82 cases. J Trauma 1979;19: 6-12. 21. Kaufman HH. Treatment of civilian gunshot wounds to the head. In: Dagi TF, ed. Gunshot wounds to the head. Boston: Little Brown: in press. 22. Kaufman HH, Levin HS, High WM Jr, Childs TL, Wagner KA, Gildenberg PL. Neurobehavioral outcome after gunshot wounds to the head in civilians. Neurosurgery 1985;16:754-8. 23. Kaufman HH, Loyola WP, Makela ME, Frankowski RF, Wagner KA, Bernstein DP, Gildenberg PL. Civilian gunshot wounds: the limits of salvageability. Acta Neurochir 1983 ;67:115-25. 24. Kaufman HH, Lynn J. Brain death. Neurosurgery 1986;19: 850-6. 25. Kaufman HH, Makela ME, Lee KF, Haid RW, Gildenberg PL. Gunshot wounds to the head: a perspective. Neurosurgery 1986;18:689-95. 26. Kaufman HH, Mattson JC. Coagulopathy in head injury. In: Becker DP, ed. Central nervous system trauma and research status report. Richmond: William Byrd Press, 1985:187-206. 27. Kaufman HH, MoakeJL, OlsonJD, Miner ME, deCret RP, Pruessner JL, Gildenberg PL. Delayed and recurrent intracranial hematomas related to disseminated intravascular clotting and fibrinolysis in head injury. Neurosurgery 1980;7:445-9. 28. Brandvold B, Levi L, Feinsod M, George E. Penetrating craniocerebral injuries in the Israeli involvement in the Lebanese conflict, 1982-1985: analysis of a less aggressive surgical approach. J Neurosurg 1990;72:15-21. 29. Lillard PL. Five years experience with penetrating craniocerebral gunshot wounds. Surg Neurol 1978;9:79-83. 30. Bracken MB. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. N Engl J Med 1990;332:1405-11. 31. Miller JD, Butterworth JF, Gudeman SK, Faulkner JE, Choi SC, Selhorst JB, Harbison JW, Lutz HA, Young HF, Becker DP. Further experience in the management of severe head injury. J Neurosurg 1981;54:289-99. 32. Molofsky WJ. Steroids and head trauma. Neurosurgery 1984;15: 424-6. 33. Myers PW, Brophy J, Salazar AM, Jonas B. Retained bone fragments after penetrating head injury. Read before the American Association of Neurological Surgeons, Washington, DC, 1989. 34. Nagib MG, Rockswold GL, Sherman RS, Lagaard MW. Civilian gunshot wounds to the brain: prognosis and management. Neurosurgery 1986;18:533-7. 35. Owen-Smith MS. High-velocity and military gunshot wounds. In: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988:61-94. 36. Raimondi AJ, Samuelson GH. Craniocerebral gunshot wounds in civilian practice. J Neurosurg 1970;32:647-53.
Penetrating Head Injury Care Survey
37. Robertson CS, Clifton GL, Grossman RG. Oxygen utilization and cardiovascular function in head-injured patients. Neurosurgery 1984;15:307-14. 38. Rybek B. Missile wounding and hemodynamic effects of energy absorption. Acta Chir Scand 1974;450:1-32. 39. Saba MI. Surgical management of gunshot wounds of the head. In: Schmidek HH, Sweet WH, eds. Operative neurosurgical techniques: indications, methods, and results. Orlando: Grune & Stratton, 1988:37-48. 40. Saul T. Is ICP monitoring worthwhile? Clin Neurosurg 1988;34:560-71. 41. SeeligJM, Becker DP, MillerJD, Greenberg RP, Ward JD, Choi SC. Traumatic acute subdural hematoma. N Engl J Med 1981;304:1511-18. 42. Sherman WD, Apuzzo MLJ, Heiden JS, Petersons VT, Weiss MH. Gunshot wounds to the brain: a civilian experience. WestJ Med 1980;132:99-105.
Surg Neurol 1991;36:370-7
377
43. Shoung HM, SichezJP, Pertuiset B. The early prognosis of craniocerebral gunshot wounds in civilian practice as an aid to the choice of treatment. Acta Neurochir 1985;74:27-30. 44. Swan KG, Swan RC. The head. In: Swan KG, Swan RC, eds. Gunshot wounds: pathophysiology and management. Littleton, MA: PSG Publishing Company, 1980:29-55. 45. Villanueva PA. Cranial gunshot wounds, in: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988: 257-66. 46. West CG. A short history of the management of penetrating missile injuries of the head. Surg Neurol 1981;16:145-9. 47. West JG. The role of trauma centers in treating victims with gunshot wounds. In: Ordog GJ, ed. Management of gunshot wounds. New York: Elsevier, 1988:13-24. 48. Yashon D, Jane JA, Martonffy D, White RJ. Management of civilian craniocerebral bullet injuries. Am Surg 1972;346-51.