The concept of formulating a set of recommendations for patient management has its basis in the advent of scientific investigation in clinical medicine. Along with this came the realization that deriving recommendations from our own experience as clinicians, although tempting, has limitations related to the fact that humans remember the best or the worst of their experiences, and cannot easily objectify their results. What clinical experience does do for us is to help with the formulation of hypotheses that can be tested through human clinical trials experimentation. Rather than seeing clinical experience and clinical trials as dichotomous, polar sources of information regarding outcomes of patient care, it is probably more useful to think of the gamut of clinical testing along a continuum. On one end, we have the more subjective impressions of care delivery that are attractive because they are grounded in personal experience in the real world of caring for patients. This is expressed well by one of the fathers of clinical epidemiology, Alvan Feinstein. Dr. Feinstein, in his seminal work, Clinical Judgment2, writes: ". . .any decent doctor reflects on alternatives, is aware of uncertainties, modifies judgments on the basis of accumulated evidence, balances risks of various kinds, considers the potential consequences of his or her diagnoses or treatments, and synthesizes all of this in making a reasoned decision that he or she decrees right for the patient." He further goes on to underscore the value of clinical experience when he writes, "In caring for patients, clinicians constantly perform experiments. During a single week of active practice, a busy clinician conducts more experiments than most of his laboratory colleagues do in a year."

This philosophy underscores and supports the notion that we can make recommendations from the wealth of clinical experience provided by direct, day-to-day clinical care. However, a close look at the process of true experimentation yields a long list of potential errors that can be made in the gathering and interpretation of data, even in highly controlled circumstances. Sources of error have been classified into two broad categories: systematic error (bias) and random error. In general, the former source of error is controlled by careful study design, and the latter is dealt with by randomization. Without these protectors against error, the chances of making mistakes in deriving recommendations from practice are great. For this reason, the notion of categorizing recommendations into groups associated with various strengths of clinical studies on the basis of scientific rigor was introduced in the last decade-and-a-half of the 20th century.

In 1990, the Institute of Medicine published a landmark work entitled Clinical Practice Guidelines: Directions for a New Program1. The instigating factors for this publication were the observation of wide variability of practice throughout American healthcare and the development of mechanisms for establishing quality assurance and review within Medicare, along with the increasing use of technology and concomitant costs of care delivery. In this document, the authors made recommendations regarding attributes defining the validity of recommendations. These include the following: 1. The available scientific literature should be searched using appropriate and comprehensive search terminology. 2. A thorough review of the scientific literature should precede guideline development. 3. The evidence should be evaluated and weighted, reflecting the scientific validity of the methodology used to generate the evidence. 4. There should be a link between the available evidence and the recommendations, with the strength of the evidence being reflected in the strength of the recommendations, reflecting scientific certainty (or lack thereof). 5. Empirical evidence should take precedence over expert judgment in the development of guidelines. 6. Expert judgment should be used to evaluate the quality of the literature and to formulate guidelines when the evidence is weak or nonexistent. 7. Guideline development should be a multidisciplinary process, involving key groups affected by the recommendations.

Search of the Literature

The literature is generally searched using the National Library of Medicine computerized database, either through individual search engines in academic medical centers or through the internet. This, by definition, will limit the scope of the search to 1966 and forward. In addition, if modern neuroimaging techniques are pertinent, as in this document, the search should be limited to the years since the modality was available. Therefore, our searches were limited to the period of 1975 to the present. Because guidelines are applicable to patient care, the literature is limited to human studies in the area of traumatic brain injury and surgical management, along with the imaging characteristics in patients with surgically amenable lesions. In addition, further limitations in the search were imposed by the fact that most of the literature is in English, and limited ability existed for reading the useful articles that appear in different languages. Therefore, most searches were limited to the English language. Appropriate search terms were chosen, as demonstrated in each of the individual Guideline sections.

Evaluation and Weighting of the Literature

The journal articles found have been carefully read and evaluated, including an assessment of the methodology used in the studies. This not only includes the establishment of the clinical question addressed (e.g., therapeutic effectiveness, diagnostic tests, prognostic studies, etc.) and type of study (randomized controlled trial, case-control study, case series, etc.), but also the quality of the study with respect to potential errors in design, execution, or conclusions reached. Therefore, studies that might, on the surface, represent evidence supporting one level of recommendation, may instead be flawed enough to be devalued to support a recommendation of lesser strength. The quality of the literature was evaluated in this way according to well-established criteria3. All articles were cross-reviewed and disagreements were resolved by consensus.

Link Between Evidence and Guidelines

The general concept of relating strength of recommendations to strength of evidence reflecting varying degrees of clinical certainty was formalized into a scheme that has been followed by medical societies, including organized neurosurgery, from the inception of the Guideline development process. Despite problems with the strict application of this paradigm (some of which are displayed and discussed in this supplement), the scheme has the benefit of using scientific evidence rather than expert opinion for the substrate of the recommendations, although expert opinion is used to formulate the recommendations themselves, as well as to make judgments regarding the quality of the evidence. The evidence-based scheme used in these and all Guidelines regarding therapeutic effectiveness endorsed by the American Association of Neurological Surgeons and the Congress of Neurological Surgeons begins with classification of the literature into three categories of evidence, as outlined next and in Table 1.

The classification of evidence into these three categories leads to the formulation of recommendations called Standards, Guidelines, and Options. Class I evidence is used to support treatment recommendations of the strongest type, practice Standards, reflecting a high degree of clinical certainty. Class II evidence is used to support Guidelines, reflecting a moderate degree of clinical certainty. Class III evidence supports practice Options reflecting unclear clinical certainty. This terminology was developed to indicate, in normal vocabulary, the strength of the recommendations on the basis of strong to weak medical evidence. In neurosurgery, this scheme has been used to formulate Guidelines, rather than a scheme that uses letters or numbers that have no grounding in language and are, therefore, more easily misinterpreted. The link between scientific evidence and recommendations has been highlighted in these Guidelines by presenting those studies in the scientific foundation that support the stated recommendation in boldface type.

Recommendations are not necessarily weak in cases in which the evidence is weak. Examples of this appear in many of the Guidelines developed in neurosurgery to date, and authors and readers alike feel frustrated at the impotency of the recommendations when the evidence is weak, especially when the logic of the recommendation and all of the evidence supporting it, however weak, support the recommendation. One example of this, among many that can be found in this supplement, is in the recommendation regarding timing of evacuation of epidural hematomas. According to the paradigm embraced and used in this set of Guidelines, case series indicating that patients who have a Glasgow Coma Scale score of 8 or below with evidence of a "blown pupil" and who are operated on early achieve better outcomes can only support a practice "Option." However, no competent neurosurgeon would allow a patient in this clinical scenario to be neglected when the need for surgical relief of brain compression is so clear. It is fairly certain to say that there will never be a randomized controlled trial for this circumstance, and, thus, never a practice "Standard." However, a patient database could be used to generate a case-control study, thus, yielding a higher recommendation. However, there is no evidence that waiting to operate on such a patient is beneficial, and, therefore, an "Option" to delay surgical evacuation will also probably never be promulgated. If, indeed, such a recommendation were put forward, it would never be accepted by the profession, and rightfully so.

Expert Judgment and Empirical Evidence

There are two ways in which expert judgment comes into Guideline development. The most common use of expert opinion is in developing recommendations for practice. This has been a usual method in the past (as well as the present, in the form of textbook chapters), but has more recently given way to more formalized approaches embraced by evidence-based medicine methodology, such as that used in this supplement. However, even in evidence-based methodology, expert opinion is used to evaluate the literature as well as to frame the concepts and wording of the recommendations. In addition, if the evidence is weak and conflicting, expert opinion is used to derive recommendations. This use is unavoidable, but the expert opinion is guided by the evidence published in the literature, rather than from personal experience alone.

Option Level Recommendation

All of the recommendations in the Surgical Management of Traumatic Brain Injury are at the option level, supported only by Class III scientific evidence. Unfortunately, option can mean choice and, as discussed above, it would be unethical not to operate on surgical mass lesions in salvageable patients. Currently, all neurological guidelines approved by the American Association of Neurological Surgeons and the Congress of Neurological Surgeons use option recommendations for Class III evidence. In the future, this may change because these surgical guidelines put maximized strain on using such terminology.

Given that the recommendations in these surgical guidelines are at the option level (Class III evidence) we state the recommendation at the beginning of the chapter on each topic, and refer the reader to this methodology section for an explanation of the level of evidence.

Multidisciplinary Process

In all Guidelines published under the auspices of the Brain Trauma Foundation and the American Association of Neurological Surgeons, other professional organizations were involved in either developing the Guidelines or reviewed and approved them. In these Surgical Management of Traumatic Brain Injury Guidelines, however, only neurosurgeons were involved. These neurosurgeons represent a wide range of organizations. There were representatives from the American Association of Neurological Surgeons, the Congress of Neurological Surgeons, the European Brain Injury Consortium, the American College of Surgeons (Committee of Trauma) and the World Federation of Neurological Surgeons (Neurotrauma section) involved in the development of these Surgical Management of Traumatic Brain Injury Guidelines.