- Comments
The Surgical Guidelines are an important and long-awaited addition to the series of brain injury related evidencebased guidelines dealing with the medical management and prognosis of traumatic brain injury (TBI), pre-hospital care, penetrating injuries, and pediatric brain injury. Most of the authors are veterans at producing brain injury guidelines and were authors for one or more of the previous publications. This supplement contains chapters on each of the five most important post-traumatic surgical lesions (epidural, subdural, intracerebral, and posterior fossa hematomas/contusions, and depressed cranial fractures). Chronic subdural hematomas and gunshot wounds are not addressed; the latter is the focus of a separate publication. For each chapter, an extensive review of the English language literature provided that clearly defines specific clinical and radiological prognostic indicators. In summarizing their recommendations, the authors have deviated from previous conventions. Rather than listing "standards," "guidelines," and "options," they provide "recommendations" in terms of indications for surgery, timing, and methods. This change in format was intentional: according to guidelines methodology, standards and guidelines must be supported by prospective studies that compare patients given the study treatment with those who do not receive that treatment. For the recommendation to rise to the level of a standard, there must be one or more high quality large prospective randomized clinical trials supporting it. If there are no studies with a comparative element, the strongest recommendation that can be made for the treatment is that of an option. But, for large post-traumatic intracranial lesions, there have not been any prospective randomized clinical trials in which a control group was assigned to a no surgery arm, and there is not likely to be such a study in the future. As a result, there are no, and will likely never be, studies sufficient to elevate recommendations for surgery above the level of an option. Yet, as pointed out in the methodology section of the paper, the need for emergent evacuation of a large hematoma that is causing neurological deterioration is recognized by all neurosurgeons. Surgery is not considered optional, particularly if the patient is young and not brain dead. Therefore, although ranking surgical evacuation of a large post-traumatic hematoma as an "option" would be consistent with guidelines methodology, it may not be consistent with best medical practice.
Noninvasive or minimally invasive approaches are increasingly popular for a wide range of surgical diseases, including trauma. Many neurosurgeons are advocating a nonsurgical approach for increasingly larger post-traumatic intracranial lesions, especially epidural hematomas, and particularly if the patient is awake and relatively neurologically intact. With close neurological and physiological monitoring in an intensive care unit, deterioration can be rapidly detected and evacuation of the lesion quickly accomplished. However, those who advocate a more conservative (nonoperative) approach are encouraged to carefully consider that the studies cited in most of the chapters found increased neurological morbidity and mortality in patients with similarly sized lesions who were not operated on until there was neurological deterioration compared with those who underwent early surgery. There is also concern about the effect of prolonged focal compression of the cortex underlying the hematoma. In order to fully understand the potential for ischemic compression injury, one would need to systematically study these patients with magnetic resonance imaging and neuropsychological tests at 3 to 6 months, or longer, after injury. Given the absence of studies that demonstrate we are not causing ischemic cortical injury by leaving a large clot in place, as well as the substantial body of evidence documenting worse outcomes for subgroups of patients undergoing late versus early surgery, the recommendations for surgical intervention as outlined in this document are certainly appropriate.
Donald W. Marion
Boston, Massachusetts
It is obvious that the authors of this review have invested a great deal of time in identifying the relevant literature, reviewing and studying those reports, and organizing them into the framework presented here. We owe them a debt of gratitude. In this era of skyrocketing practice expenses, decreasing reimbursements, and reduced availability of residents to help carry the load, projects of this type may become increasingly rare. An especially nice feature of this work is the concise, but thorough, Methodology section, which can be easily reviewed both by those desiring an introduction to this whole process and by those who simply wish a refresher course.
What do we make of the fact that all of the recommendations in this work are presented only as options, which are the weakest type of recommendations? Unfortunately, the existing evidence does not support the creation of recommendations at the more substantial levels of standards or guidelines. Just about every reference in this document consists of only Class III data. Some readers may be tempted to go down the path of nihilism, throwing up their hands and complaining that no data exist to support any surgical intervention. However, at least three rebuttals can be offered to such a reaction. First, within the broad category of Class III data, some studies are more valuable than others. Stated in another way, not all Class III studies are equal. It may be true that the available data do not reach the standards of Class I or Class II evidence, but it is incorrect to assume that there exists absolutely no guidance to help us navigate these waters.
Second, evidence-based medicine should be about more than blindly following experts' recommendations, even those based on well-conducted Class I studies. The literature can give us a good idea of how we should proceed, but these recommendations must always be tempered by such factors as a patient's preferences, the overall condition of a patient, a physician's skill and experience, the resources available to a patient and a physician, etc. Analysis of these variables used to be called the "art" of medicine, a term which is used less and less frequently in this era of growing emphasis on adherence to standardized protocols.
The third reason to oppose the path of nihilism and cynicism is that such responses are not fair to our patients. They deserve better.
To the casual reader, some of the recommendations in this document may seem far too picky and detailed, perhaps reminding some readers of complex legal decisions. However, those who review the literature can work only with the available evidence, and if the relevant studies couch their conclusions in specific terms, reviewers should not alter those terms for the sake of convenience or simplicity. To their credit, the authors of this document generally respect the limitations of the supporting data, but this adherence to the limitations of the original references leads to some obvious problems.
One example of such a problem, which is discussed in the Methodology section, is the lack of Class I evidence to support the need for immediate evacuation of an epidural hematoma in a patient with rapid neurological deterioration. Although widespread standards of care call for immediate evacuation of such a hematoma, this review cannot recommend this type of surgery as a standard or guideline because no Class I or II studies have been conducted to validate immediate surgery. Obviously, randomizing such patients to the "nonoperative" arm of such a study would be unethical, making it impossible to conduct a trial of this type.
Similar questions could be asked about the other recommendations in this supplement. For example, the chapter on epidural hematomas describes specific indications under which some hematomas could be managed nonoperatively. What about a patient at a small hospital that may not have the ability to repeat a computed tomographic scan or take a patient to the operating room abruptly if he deteriorates in the middle of the night, while, at the same time, the nearby trauma centers and tertiary care centers are saturated or otherwise unable to accept the patient in transfer? Those situations seem to be occurring with increasing frequency. In such cases, it may be appropriate to evacuate such hematomas promptly so that these patients are not subjected to the risk of neurological deterioration at a time when immediate surgery may not be possible.
On the other hand, the chapter on subdural hematomas provides specific indications for surgery, but most neurosurgeons have seen elderly patients or others with sufficient cerebral atrophy to allow them to accommodate relatively large hematomas without any deficits.
Similarly, one could find reasonable exceptions to the recommendations for surgical management of traumatic parenchymal lesions. Two sentences in that chapter apply to every pathological entity covered in this supplement: "Surgical indications are, in fact, related to many factors, including CT parameters (i.e., volume, midline shift, basal cistern compression), clinical status, and the occurrence of clinical deterioration, among others," and "These studies highlight the dynamic nature of parenchymal injuries and the dangers inherent in placing too much emphasis on a single, static CT scan for management decisions."
The literature review and discussion of decompressive craniectomy as a treatment for intracranial hypertension seems to be skewed in favor of such an operation. For example, in the fourth paragraph of the Introduction, the authors state that a report from the Medical College of Virginia in 1977 found that wide decompressive craniectomy with duraplasty was an effective treatment for raised ICP. However, that report simply states that a wide craniotomy was performed to expose the frontal and anterior temporal lobe on the appropriate side
1 . The "decompression" mentioned in that article referred to removal of clot and/or necrotic brain. Duraplasty was not mentioned, and the bone flap was usually replaced.Another example consists of the discussion of the study by Taylor et al.
2 at the end of the chapter on traumatic parenchymal lesions. The authors of this review classify Taylor's study as Class II. However, management protocols continually evolved and changed during the course of that study, including regimens for administration of intravenous fluid, use of hyperventilation and hypothermia, and definition of adequate cerebral perfusion pressure. The decompressive craniectomies consisted of only limited bitemporal bone removal without dural opening. But, in certain cases, the dura was "scarified in a crisscross pattern." Only 27 patients were enrolled over a 7-year period. The study's authors themselves discuss some potential problems with outcome evaluation and statistical analysis of results. This list of shortcomings suggests that these results likely represent Class III evidence. These criticisms are not meant to attack the investigators, who deserve nothing but praise for carrying out a difficult study in a challenging group of patients. Instead, these comments are merely intended to emphasize that the enthusiasm for decompressive craniectomy that seems to characterize these surgical guidelines may someday prove to be justified. But, based on the evidence that is presently available, such enthusiasm seems to be out of place in a guidelines document that must be based on an objective review of the literature.It seems fair to summarize the literature on decompressive craniectomy as follows: decompressive craniectomy seems to be useful for lowering intracranial pressure in many patients, but no high-quality study has demonstrated an improvement in outcomes. Decompressive craniectomy is probably the procedure of choice to control intracranial pressure and improve outcome in specific types of patients, but until those patients can be identified prospectively, the exact indications for this procedure remain unclear.
The recommendations for posterior fossa mass lesions and for depressed cranial fractures are fairly straightforward and seem to reflect common clinical practice.
A curious feature about the introduction and dissemination of clinical guidelines is the unpredictability with which specific recommendations are accepted and implemented by practitioners in the relevant specialties. Solidly supported recommendations often seem to be ignored, whereas relatively weak options may become accepted as gospel. To avoid misinterpreting the recommendations in this document, readers should remember that they are all options, supported only by Class III data.
Another interesting aspect of the guidelines movement is that it seems as if more and more clinicians are constructing guidelines instead of conducting clinical research. Guidelines are best thought of as ways of reviewing where we have been and where we stand right now. They are ways to stop and check the pulse of a specialty. However, it is much harder to create and publish original data that advance the state of our knowledge. Unquestionably, most clinical researchers are being forced to work harder and harder in the current environment of decreasing reimbursements, increasing expenses, and increasing demands on our professional time. Perhaps advances in information technology may offset some of these pressures by making it easier to capture the wealth of clinical material that neurosurgeons see every day and by including that information in well-conceived trials that address some of the important questions posed in reviews like this one. In this regard, the new emphasis of the National Institutes of Health on large-scale translational research may also facilitate the creation of appropriate organizational frameworks for investigating these issues.
Alex B. Valadka
Houston, Texas
The Brain Trauma Foundation has sponsored a monograph on the development of guidelines for the surgical management of TBI. This is a landmark accomplishment and follows the initial development of guidelines for the treatment of TBI, which were published some years ago through a cooperative effort of the American Association of Neurological Surgeons and the Brain Trauma Foundation. The authors are to be congratulated for their meticulous review of this, overall, extremely difficult topic. The various topics are beautifully put together and well thought through, and the recommendations appropriate. Clearly well crafted is the initial description of evidentiary technique. The authors have taken to heart the Institute of Medicine's 1990 report on Clinical Practice Guidelines. This runs throughout the entire document, and the neurosurgeons involved with this effort are owed a substantial debt by those of us care for such patients.
One important feature of this document is the recognition that randomized trials cannot be applied in a number of circumstances, particularly in patients with substantial mass lesions who are deteriorating. This principle applies particularly in their discussion of management of intraparenchymal hemorrhages. Judgment depends on a variety of factors, including size of the mass, degree of midline shift, the age of the patient, and clinical status.
It is important to remember that good results are often reported in a variety of mass lesions in patients who develop unilateral anisocoria. I have never seen a patient older than 18 years of age who has been entirely normal when having such an experience. This does not mean that the patients cannot make an excellent recovery but that herniation should be avoided if at all possible, and that newer technologies than those we have traditionally relied upon are required to give us an earlier hint of a patient's decline.
In summary, Surgical Management of Traumatic Brain Injury will take its place with its antecedent on the overall management of TBI and represents a seminal accomplishment to assist us in the management of this heterogeneous disorder, which remains a worldwide scourge.
Lawrence F. Marshall
San Diego, California
For those neurosurgeons still involved in the care of patients sustaining TBI, this evidence-based text on the surgical treatment of traumatic hematomas and depressed cranial fractures is a must read. Written by recognized experts in the field, each chapter provides an excellent overview of the topic, highlights common sense management recommendations supported by the literature from 1975 through 2001, and suggests keys areas for future study. As in previous evidencebased guideline efforts, the authors adhered to the relatively rigid format of categorizing the strength of the evidence as Class I, II, or III, with the resultant treatment recommendations based on this categorization. Not surprisingly, given the absence of controlled clinical trials in the surgical management of traumatic hematomas or cranial fractures, the available evidence does not support any treatment "standards" or "guidelines," but does support numerous treatment "options." For example, surgical thresholds for epidural, subdural, and intraparenchymal hematomas are offered based on clot thickness, clot volume, and amount of midline shift. Although such "options" may seem obvious to most surgeons and excessively rigid to others, when taken in context of a given clinical scenario, these recommendations will provide a useful "reality check" to help ensure appropriate clinical management. Given that much of the data and treatment recommendations are based on hematoma volume, one should be familiar with the simple "ellipsoid" or "ABC/2" method to calculate clot volume, as shown in Appendix I. Despite numerous areas where the data is inconclusive, such as the indications for use of decompressive craniectomy, the authors have done an excellent job in synthesizing but not over-interpreting the available data. The end result is a highly pragmatic set of recommendations that should become part of day-to-day management of TBI patients and part of the core curriculum at all neurosurgical training programs and trauma centers. Although admittedly a work in progress, the authors are to be congratulated for this extensive contribution to the neurosurgical literature.
Daniel F. Kelly
Los Angeles, California
The authors have done a great service by compiling much of the published English language literature on surgical management of TBI. They have highlighted the limited quality of that evidence while identifying a great deal of information that has made it possible to gradually improve the care of these patients. They have wisely avoided identifying these as evidence-based guidelines, because of both methodological limitations of the review and the primary data. However, as a repository of the best evidence that is available to guide the surgical treatment of patients with TBI, this document is a major step forward. As a guide to the need for further, higher quality, clinical research in TBI, it is unsurpassed.
There are important methodological issues that must be kept in mind when interpreting and applying the information the authors have assembled. They have forthrightly discussed a number of these points in the section on methodology, which should be revisited by the reader frequently. Some points should be emphasized.
It should be recognized that limiting literature searches to the English language may introduce bias, particularly against studies that show no treatment effect
4 .The authors have used a practical scheme for classifying individual articles developed for the American Association of Neurological Surgeons practice guidelines development process. This three-class scheme is the same as the classification of summarized evidence and has made the concept of evidence classification easier to explain and remember. There are more complex systems of classification that allow more nuanced distinctions in evidence quality
3 . Rather than being what the authors call "a scheme that uses letters or numbers that have no grounding in language and are, therefore, more easily misinterpreted," this grading scale is more sensitive in identifying graded quality of evidence, allowing flawed randomized trials to be downgraded and well conducted nonrandomized controlled studies to take their proper place in the hierarchy of evidence. The three-category scale used by the authors was developed and intended for use in summarizing evidence from multiple studies and, for the reasons stated by the authors, does a good, but not perfect, job in this regard. By substituting this scale for the more appropriate five-level scale in the evaluation of single articles, the authors may have inadvertently downgraded the quality of much of the evidence they reviewed (because no randomized controlled trial data was available) and, thereby, downgraded the levels of recommendations they could support. In the end, it is unlikely that these methodological issues have affected the final conclusion that the evidence quality does not support either "standards" or "guidelines."The problem of relegating well established principles of care to the status of an "option" is, on its surface, difficult. As the authors state, no competent neurosurgeon would fail to remove a large epidural hematoma in a patient with neurological compromise. This has been clinical dogma for decades for good reasons. However, 30 years ago many small epidural hematomas were surgically treated based on logical extrapolation from the understanding that enlarging epidural hematomas could be fatal. A more sophisticated approach that allows for nonsurgical management of small hematomas has evolved in recent years. Had the methods of evidence-based medicine existed 30 years ago, and had experts elevated the dogma "epidural hematomas must be evacuated" to "standard" status without evidence of appropriate quality, the development of today's more nuanced approach could have been prevented. Describing a recommendation as an "option" need not prevent it from being the dominant course of action based on the best current evidence, but it does allow, and encourages, development of better evidence leading to better treatment. We should not be afraid to be forthright about the inadequacy of our evidence. For these reasons, I believe it would have been preferable for the authors to identify each of the recommendations as "options" while stating, where appropriate, that expert consensus supports a particular course of action.
The lack of high quality evidence to support the authors' recommendations has two important implications. First, it must not be forgotten that evidence of this quality leaves room for, and indeed requires, flexibility in individual physician interpretation and application to specific clinical situations. It would be as bad a mistake to interpret the recommendation to evacuate subdurals with midline shift greater than 5 mm as an absolute requirement as it would be to fail to remove an epidural hematoma in a patient with neurological deterioration simply because it was less than 30 cc in volume. Second, this compilation provides an eloquent and urgent plea for more cooperation among neurosurgeons to carry out the kind of high quality clinical research that can answer questions as well as raise them.
Stephen J. Haines
Minneapolis, Minnesota