JOURNAL CLUB JOURNAL CLUB
Journal Club: National Variability in Intracranial Pressure Monitoring and Craniotomy for Children With Moderate to Severe Traumatic Brian Injury Brian L. Anderson, MD Einar Bogason, MD Nicholas Brandmeir, MD, MS Ephraim Church, MD Jonathon Cooke, MD* Justin Davanzo, MD Gareth Davies, MD Namath Hussain, MD, MBA Akshal Patel, MD Russell Payne, MD Pratik Rohatgi, MD, MSE Emily Sieg, MD Omar Zalatimo, MD Department of Neurosurgery, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania *This author is a military service member. This work was prepared as part of his official duties. Title 17, USC, §105 provides that, Copyright protection under this title is not available for any work of the US Government. Title 17, USC, §101 defines a US Government work as a work prepared by a military service member or employee of the US Government as part of that person’s official duties. The views expressed in this presentation are those of the author and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or the United States government. Correspondence: Brian L. Anderson, MD, 30 Hope Drive, P.O. Box 859 Hershey, PA 17033. E-mail: [email protected] Copyright © 2014 by the Congress of Neurological Surgeons.
NEUROSURGERY
SIGNIFICANCE AND IMPORTANCE OF STUDY
T
raumatic injury is the number one cause of morbidity and mortality in the pediatric population. Of the 14 million injuries occurring each year in the United States to children younger than 15 years of age, head injury is the leading cause of death and permanent disability.1 Although significant progress in the evidence-based evaluation and treatment of traumatic brain injury (TBI) has been made, important knowledge gaps remain, and current practices are frequently supported by low-quality evidence derived from retrospective studies. This is particularly true in the setting of pediatric TBI, where current practice is largely based on extrapolation from the adult literature, despite important differences in circumstance and pathophysiology of injury and the clinical evaluation of TBI in infants, toddlers, and adolescents. In this context, the study by Van Cleve et al2 provides a “snapshot” of national practices regarding the use of intracranial pressure monitoring and craniectomy/craniotomy in the treatment of significant pediatric TBI. Theirs is a large retrospective cohort study that highlights dramatic variations in the management of head trauma in the pediatric population despite widely publicized evidence-based practice parameters published in 2003. Substantial practice variation is generally seen in situations in which compelling evidence-based guidelines are lacking or are not accepted by the majority of relevant practitioners. This is not surprising because the evidence-based guidelines published by the Brain Trauma Foundation in 2003 are almost exclusively based on Level III evidence and consequently can offer only treatment options rather than formal recommendations. These guidelines were revised in 2012, emphasizing even more clearly the low quality of evidence that exists.3 Similarly, only weak evidence exists for the use of decompressive craniectomy in
patients with evidence of herniation.3 With such a lack of evidence-based guidelines, substantial practice variability is not surprising. The findings by Van Cleve et al provide a quantitative measure of this variability, a stark and urgent reminder of how desperately high-quality prospective research is needed and a clear ethical foundation for such research. As such, this is a critically important study.
ORIGINALITY OF THE WORK The use of large retrospective registry studies has become increasingly common in the medical literature, as technology has allowed ready access to and easy analysis of large-volume databases at very low cost. The popularity of this approach to the neurosurgical community is illustrated by the observation that this type of study has been most frequently selected for this Journal Club competition. Although previous studies have addressed issues of initially evaluation and management, with a focus on imaging decisions and predictors of poor outcome,4 no previous study has addressed the issues of neurosurgical intervention. This approach also opens the door to similar investigations of other aspects of pediatric neurotrauma.
APPROPRIATENESS OF STUDY DESIGN Patients younger than 18 years of age who experience significant TBI as determined by the Centers for Disease Control and Prevention criteria, an Abbreviated Injury Score of 3 or higher, and a Glasgow Coma Scale motor score of 3 or lower are defined as the population of interest. Data were analyzed from the American College of Surgeons National Trauma Databank from 2002 through 2008. Although such a database supplies a large volume of data, the lack of quality-control standards and inherent sources of error compromises their integrity with regard to
VOLUME 75 | NUMBER 2 | AUGUST 2014 | 191
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
ANDERSON ET AL
developing reliable and consistent findings necessary in evidencebased research.5 Data entry errors, missing data, and no effort to standardize data collection are among the most common deficiencies encountered. This is evident in the large number of missing data points reported in this trial, which accounted for 10% to 15% of each comparison group. Retrospective registries are limited to the data points collected which are often coarse, leaving many questions of interest unanswerable, and many potential confounders unaccounted for. In 2009, the National Trauma Databank included data collection from more than 550 voluntary centers across the United States. The database includes patient demographic information, International Classification of Diseases Ninth Revision procedural and discharge codes, as well as hospital trauma designation, region, and case volumes. The voluntary nature of participation in this database results in considerable risk of lack of generalizability (external validity). Two other exclusion criteria further increase the risks to external validity. First, all centers with fewer than 10 moderate/severe TBI cases were excluded. No information was provided to give insight into the magnitude of this exclusion. Second, all patients transferred to or from another acute-care facility were excluded. Transfers to tertiary care centers, particularly in the setting of pediatric injuries, is common practice. No clarification was provided to suggest patients triaged at 1 center and immediately transferred to a tertiary facility were included and the assumption is that they were excluded. If this is accurate, significant limitations in generalizability may exist. The number excluded as a result of these criteria was not provided. Although alternatives to International Classification of Diseases Ninth Revision codes to determine diagnosis and procedural coding are limited, inherent inaccuracies may introduce significant errors. This is most obvious regarding craniectomy vs craniotomy coding. This is particularly problematic because decompressive craniectomy as a treatment strategy for intracranial hypertension and craniotomy for evacuation of a mass lesion are 2 very different etiologies and are a consequence of very different neurosurgical reasoning. This is not trivial, as more than 65% of the study population was reported to have intracranial hemorrhage. Unfortunately, this is typically an inherent risk when using retrospective registries.
ADEQUACY OF EXPERIMENTAL TECHNIQUES As previously noted, retrospective registry studies (a type of cohort trial) are at substantial risk of confounding. The authors correctly use multivariate analysis to evaluate each statistically significant association detected in their univariate analyses, but this procedure only protects the findings from known confounders. Unknown confounders may still seriously distort the findings. In the setting of important confounders, the large numbers may produce tight confidence intervals (ie, greater precision) for some of the authors’ associations, but these associations may simply be more precisely wrong.
192 | VOLUME 75 | NUMBER 2 | AUGUST 2014
Registry studies also commonly perform large numbers of statistical comparisons, resulting in a substantial risk of type I error (detecting a difference when, in fact, a difference does not exist). This will result in a familywise error, which multivariate analysis does not control for and must be acknowledged. For example, in the authors’ Table 2, 23 comparisons are presented, resulting in a 69.3% risk of at least 1 type I error, and a 38.5% risk of 2 such errors. In Table 3, 13 comparisons are presented, resulting in a 48.7% risk of at least 1 type I error. Statistical methods exist to help control for this type of error (for example, the Bonferroni correction), but in this study and in most registry trials, these are rarely used. In addition, because there is no “clearinghouse” for registry trials equivalent to clinicaltrials.gov for randomized, controlled trials, readers have no way of knowing whether dozens, scores, or even hundreds of additional comparisons were made but not reported.6 The authors also acknowledge large numbers of patients with missing data in many of their variable categories, accounting for 10% to 15% of patients in these categories. The impact of this degree of missing data cannot be unequivocally evaluated. If the data loss is truly random, it will reduce the authors’ ability to show differences that are really present (type II error). Even more ominously, if the loss is nonrandom (and we have no way of knowing this), bias will be introduced into the study, and a type I error may result.
SOUNDNESS OF CONCLUSIONS AND INTERPRETATION The authors’ conclusion is concise and simply states pertinent findings. They are commended for avoiding overstating the data and recognizing that such a study does not justify recommendations regarding intervention or practice changes. The authors recognize the studies limitations and the need for more reliable and detailed prospective trials.
RELEVANCE OF DISCUSSION The authors’ discussion appropriately accounts for the largest segment of their paper. As the authors clearly state, theirs is a hypothesis-generating observational study. Their results cannot prove causality but do suggest many interesting potential associations worthy of subsequent examination. A clear recitation of potential sources of error is given. Many of the limitations noted in this critique are openly discussed, and the authors do not make far-reaching claims unsupported by the data. The hypotheses offered to explain the findings are plausible and suggest potential targets of future studies. One important assumption made by authors is not addressed in their discussion. They assume the existence of guidelines that are widely accepted by those treating pediatric TBI. In fact, no such compelling guidelines exist. As mentioned earlier, only poorquality evidence, producing “treatment options” rather than established treatment recommendations, exists. This lack of
www.neurosurgery-online.com
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
JOURNAL CLUB
consensus on published guidelines likely underlies the appreciable practice variation documented by the authors and, from a public health perspective, is perhaps the most important lesson to be drawn from this study.
appropriate to the study design and include the requisite measures of association (relative risk) and statistical significance and precision (confidence intervals), and each is reported in an easyto-read way.
CLARITY OF WRITING, STRENGTH, AND ORGANIZATION OF PAPER
FUTURE/NEXT STEPS
The paper is well organized and simple to read and understand. The data source and analyses are transparent and in some cases are easy for the reader to recapitulate. The inclusion criteria are clearly defined, and the results and tables are easy to read. The 2 neurosurgical interventions of interest are clearly stated in the introduction. The discussion and limitations sections provide an honest assessment and valuable insight into the authors’ recognition of the need for higher quality analysis. Conjecture is clearly separated from data analysis. The invited Comment also makes 2 very important points.
This study underscores, with its demonstration of persistent, substantial practice variability, the continuing lack of consensus regarding the neurosurgical management of pediatric TBI and the need for further high-quality research to refine the current evidence-impoverished practice guidelines. The specific associations detected in this study provide several interesting hypotheses that should form the basis of focused prospective, randomized trials or prospective registry studies, which, in turn, may provide more compelling evidence of causation. The authors should be congratulated for their effort and their recognition of the long road ahead.
ECONOMY OF WORDS
Disclosure
Both data and discussion are presented succinctly. The conclusion is not overly long.
RELEVANCE, ACCURACY, AND COMPLETENESS OF BIBLIOGRAPHY An extensive and thorough bibliography is provided. The cited papers are appropriate to the content of the paper. The predominant standard for management of severe pediatric TBI has been published by the Brain Trauma Foundation. Although the initial report released in 2003 is cited, the most recent revision from 2012 is omitted. Although no changes in recommendations regarding intracranial monitoring and craniectomy/craniotomy exist, it represents a thorough and critical assessment of all published data through 2012 providing the most up-to-date findings.
NUMBER AND QUALITY OF FIGURES, TABLES, AND ILLUSTRATIONS Three tables are provided that clearly present the characteristics and univariate and multivariate analysis findings. The tables are
NEUROSURGERY
The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.
REFERENCES 1. Carr BG, Nance ML. Access to pediatric trauma care: alignment of providers and health systems. Curr Opin Pediatr. 2010;22(3):326-331. 2. Van Cleve W, Kernic MA, Ellenbogen RG, et al. National variability in intracranial pressure monitoring and craniotomy for children with moderate to severe traumatic brain injury. Neurosurgery. 2013;73(5):746-752. 3. Kochanek PM, Carney N, Adelson PD, et al. Guidelines for the acute medical management of severe traumatic brain injury in infants, children, and adolescentssecond edition. Pediatr Crit Care Med. 2012;13(suppl 1):S1-S82. 4. Schonfeld D, Bressan S, Da Dalt L, Henien MN, Winnett JA, Nigrovic LE. Pediatric emergency care applied research network head injury clinical prediction rules are reliable in practice. Arch Dis Child. 2014;99(5):427-431. 5. Zehtabchi S, Nishijima DK, McKay MP, Mann NC. Trauma registries: history, logistics, limitations, and contributions to emergency medicine research. Acad Emerg Med. 2011;18(6):637-643. 6. Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet. 2002;359(9302):248-252.
Acknowledgments The authors are grateful for the support from faculty members who provided guidance for the Journal Club’s critical appraisal. Particular thanks to Drs Michael J. Glantz, Robert Harbaugh, Christopher Zacko, Phillip Villanueva, Mark Dias, Elias Rizk, and Scott Simon.
VOLUME 75 | NUMBER 2 | AUGUST 2014 | 193
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
Journal Club: National Variability in Intracranial Pressure Monitoring and Craniotomy for Children With Moderate to Severe Traumatic Brian Injury Brian L. Anderson, MD Einar Bogason, MD Nicholas Brandmeir, MD, MS Ephraim Church, MD Jonathon Cooke, MD* Justin Davanzo, MD Gareth Davies, MD Namath Hussain, MD, MBA Akshal Patel, MD Russell Payne, MD Pratik Rohatgi, MD, MSE Emily Sieg, MD Omar Zalatimo, MD Department of Neurosurgery, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania *This author is a military service member. This work was prepared as part of his official duties. Title 17, USC, §105 provides that, Copyright protection under this title is not available for any work of the US Government. Title 17, USC, §101 defines a US Government work as a work prepared by a military service member or employee of the US Government as part of that person’s official duties. The views expressed in this presentation are those of the author and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or the United States government. Correspondence: Brian L. Anderson, MD, 30 Hope Drive, P.O. Box 859 Hershey, PA 17033. E-mail: [email protected] Copyright © 2014 by the Congress of Neurological Surgeons.
NEUROSURGERY
SIGNIFICANCE AND IMPORTANCE OF STUDY
T
raumatic injury is the number one cause of morbidity and mortality in the pediatric population. Of the 14 million injuries occurring each year in the United States to children younger than 15 years of age, head injury is the leading cause of death and permanent disability.1 Although significant progress in the evidence-based evaluation and treatment of traumatic brain injury (TBI) has been made, important knowledge gaps remain, and current practices are frequently supported by low-quality evidence derived from retrospective studies. This is particularly true in the setting of pediatric TBI, where current practice is largely based on extrapolation from the adult literature, despite important differences in circumstance and pathophysiology of injury and the clinical evaluation of TBI in infants, toddlers, and adolescents. In this context, the study by Van Cleve et al2 provides a “snapshot” of national practices regarding the use of intracranial pressure monitoring and craniectomy/craniotomy in the treatment of significant pediatric TBI. Theirs is a large retrospective cohort study that highlights dramatic variations in the management of head trauma in the pediatric population despite widely publicized evidence-based practice parameters published in 2003. Substantial practice variation is generally seen in situations in which compelling evidence-based guidelines are lacking or are not accepted by the majority of relevant practitioners. This is not surprising because the evidence-based guidelines published by the Brain Trauma Foundation in 2003 are almost exclusively based on Level III evidence and consequently can offer only treatment options rather than formal recommendations. These guidelines were revised in 2012, emphasizing even more clearly the low quality of evidence that exists.3 Similarly, only weak evidence exists for the use of decompressive craniectomy in
patients with evidence of herniation.3 With such a lack of evidence-based guidelines, substantial practice variability is not surprising. The findings by Van Cleve et al provide a quantitative measure of this variability, a stark and urgent reminder of how desperately high-quality prospective research is needed and a clear ethical foundation for such research. As such, this is a critically important study.
ORIGINALITY OF THE WORK The use of large retrospective registry studies has become increasingly common in the medical literature, as technology has allowed ready access to and easy analysis of large-volume databases at very low cost. The popularity of this approach to the neurosurgical community is illustrated by the observation that this type of study has been most frequently selected for this Journal Club competition. Although previous studies have addressed issues of initially evaluation and management, with a focus on imaging decisions and predictors of poor outcome,4 no previous study has addressed the issues of neurosurgical intervention. This approach also opens the door to similar investigations of other aspects of pediatric neurotrauma.
APPROPRIATENESS OF STUDY DESIGN Patients younger than 18 years of age who experience significant TBI as determined by the Centers for Disease Control and Prevention criteria, an Abbreviated Injury Score of 3 or higher, and a Glasgow Coma Scale motor score of 3 or lower are defined as the population of interest. Data were analyzed from the American College of Surgeons National Trauma Databank from 2002 through 2008. Although such a database supplies a large volume of data, the lack of quality-control standards and inherent sources of error compromises their integrity with regard to
VOLUME 75 | NUMBER 2 | AUGUST 2014 | 191
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
ANDERSON ET AL
developing reliable and consistent findings necessary in evidencebased research.5 Data entry errors, missing data, and no effort to standardize data collection are among the most common deficiencies encountered. This is evident in the large number of missing data points reported in this trial, which accounted for 10% to 15% of each comparison group. Retrospective registries are limited to the data points collected which are often coarse, leaving many questions of interest unanswerable, and many potential confounders unaccounted for. In 2009, the National Trauma Databank included data collection from more than 550 voluntary centers across the United States. The database includes patient demographic information, International Classification of Diseases Ninth Revision procedural and discharge codes, as well as hospital trauma designation, region, and case volumes. The voluntary nature of participation in this database results in considerable risk of lack of generalizability (external validity). Two other exclusion criteria further increase the risks to external validity. First, all centers with fewer than 10 moderate/severe TBI cases were excluded. No information was provided to give insight into the magnitude of this exclusion. Second, all patients transferred to or from another acute-care facility were excluded. Transfers to tertiary care centers, particularly in the setting of pediatric injuries, is common practice. No clarification was provided to suggest patients triaged at 1 center and immediately transferred to a tertiary facility were included and the assumption is that they were excluded. If this is accurate, significant limitations in generalizability may exist. The number excluded as a result of these criteria was not provided. Although alternatives to International Classification of Diseases Ninth Revision codes to determine diagnosis and procedural coding are limited, inherent inaccuracies may introduce significant errors. This is most obvious regarding craniectomy vs craniotomy coding. This is particularly problematic because decompressive craniectomy as a treatment strategy for intracranial hypertension and craniotomy for evacuation of a mass lesion are 2 very different etiologies and are a consequence of very different neurosurgical reasoning. This is not trivial, as more than 65% of the study population was reported to have intracranial hemorrhage. Unfortunately, this is typically an inherent risk when using retrospective registries.
ADEQUACY OF EXPERIMENTAL TECHNIQUES As previously noted, retrospective registry studies (a type of cohort trial) are at substantial risk of confounding. The authors correctly use multivariate analysis to evaluate each statistically significant association detected in their univariate analyses, but this procedure only protects the findings from known confounders. Unknown confounders may still seriously distort the findings. In the setting of important confounders, the large numbers may produce tight confidence intervals (ie, greater precision) for some of the authors’ associations, but these associations may simply be more precisely wrong.
192 | VOLUME 75 | NUMBER 2 | AUGUST 2014
Registry studies also commonly perform large numbers of statistical comparisons, resulting in a substantial risk of type I error (detecting a difference when, in fact, a difference does not exist). This will result in a familywise error, which multivariate analysis does not control for and must be acknowledged. For example, in the authors’ Table 2, 23 comparisons are presented, resulting in a 69.3% risk of at least 1 type I error, and a 38.5% risk of 2 such errors. In Table 3, 13 comparisons are presented, resulting in a 48.7% risk of at least 1 type I error. Statistical methods exist to help control for this type of error (for example, the Bonferroni correction), but in this study and in most registry trials, these are rarely used. In addition, because there is no “clearinghouse” for registry trials equivalent to clinicaltrials.gov for randomized, controlled trials, readers have no way of knowing whether dozens, scores, or even hundreds of additional comparisons were made but not reported.6 The authors also acknowledge large numbers of patients with missing data in many of their variable categories, accounting for 10% to 15% of patients in these categories. The impact of this degree of missing data cannot be unequivocally evaluated. If the data loss is truly random, it will reduce the authors’ ability to show differences that are really present (type II error). Even more ominously, if the loss is nonrandom (and we have no way of knowing this), bias will be introduced into the study, and a type I error may result.
SOUNDNESS OF CONCLUSIONS AND INTERPRETATION The authors’ conclusion is concise and simply states pertinent findings. They are commended for avoiding overstating the data and recognizing that such a study does not justify recommendations regarding intervention or practice changes. The authors recognize the studies limitations and the need for more reliable and detailed prospective trials.
RELEVANCE OF DISCUSSION The authors’ discussion appropriately accounts for the largest segment of their paper. As the authors clearly state, theirs is a hypothesis-generating observational study. Their results cannot prove causality but do suggest many interesting potential associations worthy of subsequent examination. A clear recitation of potential sources of error is given. Many of the limitations noted in this critique are openly discussed, and the authors do not make far-reaching claims unsupported by the data. The hypotheses offered to explain the findings are plausible and suggest potential targets of future studies. One important assumption made by authors is not addressed in their discussion. They assume the existence of guidelines that are widely accepted by those treating pediatric TBI. In fact, no such compelling guidelines exist. As mentioned earlier, only poorquality evidence, producing “treatment options” rather than established treatment recommendations, exists. This lack of
www.neurosurgery-online.com
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
JOURNAL CLUB
consensus on published guidelines likely underlies the appreciable practice variation documented by the authors and, from a public health perspective, is perhaps the most important lesson to be drawn from this study.
appropriate to the study design and include the requisite measures of association (relative risk) and statistical significance and precision (confidence intervals), and each is reported in an easyto-read way.
CLARITY OF WRITING, STRENGTH, AND ORGANIZATION OF PAPER
FUTURE/NEXT STEPS
The paper is well organized and simple to read and understand. The data source and analyses are transparent and in some cases are easy for the reader to recapitulate. The inclusion criteria are clearly defined, and the results and tables are easy to read. The 2 neurosurgical interventions of interest are clearly stated in the introduction. The discussion and limitations sections provide an honest assessment and valuable insight into the authors’ recognition of the need for higher quality analysis. Conjecture is clearly separated from data analysis. The invited Comment also makes 2 very important points.
This study underscores, with its demonstration of persistent, substantial practice variability, the continuing lack of consensus regarding the neurosurgical management of pediatric TBI and the need for further high-quality research to refine the current evidence-impoverished practice guidelines. The specific associations detected in this study provide several interesting hypotheses that should form the basis of focused prospective, randomized trials or prospective registry studies, which, in turn, may provide more compelling evidence of causation. The authors should be congratulated for their effort and their recognition of the long road ahead.
ECONOMY OF WORDS
Disclosure
Both data and discussion are presented succinctly. The conclusion is not overly long.
RELEVANCE, ACCURACY, AND COMPLETENESS OF BIBLIOGRAPHY An extensive and thorough bibliography is provided. The cited papers are appropriate to the content of the paper. The predominant standard for management of severe pediatric TBI has been published by the Brain Trauma Foundation. Although the initial report released in 2003 is cited, the most recent revision from 2012 is omitted. Although no changes in recommendations regarding intracranial monitoring and craniectomy/craniotomy exist, it represents a thorough and critical assessment of all published data through 2012 providing the most up-to-date findings.
NUMBER AND QUALITY OF FIGURES, TABLES, AND ILLUSTRATIONS Three tables are provided that clearly present the characteristics and univariate and multivariate analysis findings. The tables are
NEUROSURGERY
The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.
REFERENCES 1. Carr BG, Nance ML. Access to pediatric trauma care: alignment of providers and health systems. Curr Opin Pediatr. 2010;22(3):326-331. 2. Van Cleve W, Kernic MA, Ellenbogen RG, et al. National variability in intracranial pressure monitoring and craniotomy for children with moderate to severe traumatic brain injury. Neurosurgery. 2013;73(5):746-752. 3. Kochanek PM, Carney N, Adelson PD, et al. Guidelines for the acute medical management of severe traumatic brain injury in infants, children, and adolescentssecond edition. Pediatr Crit Care Med. 2012;13(suppl 1):S1-S82. 4. Schonfeld D, Bressan S, Da Dalt L, Henien MN, Winnett JA, Nigrovic LE. Pediatric emergency care applied research network head injury clinical prediction rules are reliable in practice. Arch Dis Child. 2014;99(5):427-431. 5. Zehtabchi S, Nishijima DK, McKay MP, Mann NC. Trauma registries: history, logistics, limitations, and contributions to emergency medicine research. Acad Emerg Med. 2011;18(6):637-643. 6. Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet. 2002;359(9302):248-252.
Acknowledgments The authors are grateful for the support from faculty members who provided guidance for the Journal Club’s critical appraisal. Particular thanks to Drs Michael J. Glantz, Robert Harbaugh, Christopher Zacko, Phillip Villanueva, Mark Dias, Elias Rizk, and Scott Simon.
VOLUME 75 | NUMBER 2 | AUGUST 2014 | 193
Copyright © Congress of Neurological Surgeons. Unauthorized reproduction of this article is prohibited.
