Search for a command to run...
To the Editor: Best practice in the development of a clinical practice guideline demands engagement in both internal and external review before publication.1 The role of the latter is to provide a broader, independent perspective on the document in preparation and to serve as a checkpoint for integrity of the work product and potential ethical breaches. It is the responsibility of an external reviewer to provide a critical appraisal that considers transparency and inclusivity of process, adherence to rigorous methodology, and reliability and usability of the recommendations rendered. It was in this spirit that the American Association of Neurological Surgeons/Congress of Neurological Surgeons (AANS/CNS) Joint Guidelines Review Committee (JGRC) was asked to participate in review of an AO Spine/Praxis-sponsored project entitled Clinical Practice Guidelines for the Management of Acute Spinal Cord Injury.2 The content of this document was planned for distribution (and has since been published online) as a “focus issue” of Global Spine Journal, the flagship publication of AO Spine. It is important to note that the authors were not seeking endorsement by the AANS/CNS; such a request would obligate them to provide a comprehensive response to the JGRC critique (possibly requiring manuscript modifications) as a condition of potential endorsement. In fact, the JGRC was informed that the manuscript was already deep into the editorial process at the time of the review. As the submission did not meet criteria for review of evidence-based documents by the full JGRC, it was routed to the Trauma and Spine Sections for consideration. The comments generated by these bodies were collated by the JGRC and returned to the author group. Although it is the intention of the JGRC to maintain neutrality in the review process, this document elicited an unusually intense visceral reaction among individual reviewers, generating discussion at the Section level about how best to address perceived flaws in methodology, misleading statements, and concerns about potential misuse of the final product—without the ability to influence constructive changes in the document before dissemination. Indeed, some members of the AO Spine writing group who had contributed to individual chapters expressed similar misgivings on seeing the manuscript in its entirety. There was a strong sentiment that these concerns be aired in a public forum, both to provide a more balanced perspective on the issues at hand and to protect the interests of neurosurgeons and patients in the community. A complementary article—focused on an appraisal of the clinical aspects of this document—will be published in the Journal of Neurosurgery.3 The guideline author group has been afforded an opportunity to respond.4,5 The title of this focus issue, Clinical Practice Guidelines for the Management of Acute Spinal Cord Injury, suggests a wide-ranging treatment of the subject when, in fact, its scope is quite narrow. Its 13 parts are devoted to exactly three topics: timing of surgery, blood pressure management, and intraoperative spinal cord injury (SCI). The chosen topics are not intrinsically interrelated. The organizational structure of the document is convoluted and confusing. The methodologic approach taken by the author group presents a paradox. Although very deliberate in outlining the steps taken to demonstrate adherence to best practices, the authors simultaneously take liberties with the construction of the foundational questions and interpretation of evidence that sidestep these guardrails. Most of the PICO (Population, Intervention, Comparison, Outcome) questions are not formatted properly. Many of these questions may be considered “leading” as the intended outcome is embedded in the question itself. For example, Key Question 1 for the timing of surgery asks “Should we recommend early decompressive surgery (≤24 hours after injury) for adult patients with SCI regardless of injury severity and neurological level?”6 The question broadcasts the desired outcome of support for early surgery. The question would more appropriately be phrased as follows: “For patients with acute SCI, does early decompressive surgery (≤24 hours after injury) vs surgery >24 hours after injury (or no surgery or some other comparative time frame) result in improved outcomes?” One might also argue that there is a certain partiality in the tone of a manuscript that prizes the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework as the best and only methodological strategy for evaluating and reporting evidence, while making thinly veiled comments about perceived weaknesses of prior guideline efforts in this area that were constructed using alternate rating schemes.7 For example, the authors choose to discuss “best practice” in the use of bias risk assessment tools and guideline updating within the context of the section addressing hemodynamic management—a not-so-subtle critique of the 2013 AANS/CNS Guidelines for the Management of Acute Spine and Spinal Cord Injuries that they frame as the precursor to their current recommendations.8 To be fair, guideline methodology continues to evolve, and GRADE is only one of many rubrics available to support evidence grading and recommendation reporting. A document published a decade ago should not be penalized for a difference in methodological framework appropriate for the era in which it was composed. In explaining how the GRADE process was applied, the authors emphasize the place of expert opinion where evidence might be lacking, laying the groundwork for the outsized role that it would ultimately play in the recommendations provided. This sleight of hand is an intrinsic weakness of the GRADE system, exploited here to achieve the intended outcome. Furthermore, there is an alarming dependence on the work of members of the author group as the core evidence base from which recommendations are derived. An overreliance on literature composed by the author group raises concerns about potential bias in the literature selection process and data abstraction. Nowhere is this more evident than in the first section, addressing the timing of surgery, where the authors opine that the weakness of their previous (2017) recommendation in this area stemmed from a “paucity of evidence and inability to perform a meta-analysis”—which they then published in 2021 and now rely on to make the current, stronger recommendation.9,10 It is unclear whether this stronger recommendation is justified. The second section of the document is broadly about hemodynamic management in the setting of acute SCI. Here, the authors have chosen to focus on the mean arterial pressure parameters, to the exclusion of newer evidence supporting a role for spinal cord perfusion monitoring. The authors ultimately recommend targeting a mean arterial pressure with a lower level of 75 to 80 mm Hg and an upper limit of 90 to 95 mm Hg over 3 to 7 days.11 Presenting the data in such a manner is an odd way of expressing a range that effectively spans 75 to 95 mm Hg and will only serve to confuse our medical and trauma partners who may be participating in the critical care management of these patients. In addition, the authors repeatedly refer to this topic as an “update” to the 2013 AANS/CNS Guidelines for the Management of Acute Spine and Spinal Cord Injuries chapter on The Acute Cardiopulmonary Management of Patients with Cervical Spinal Cord Injuries.12 Any reference to an update would be an inaccurate characterization. The documents share no common lineage regarding authorship or methodology. The third section of the document—addressing intraoperative SCI—differs fundamentally from the other two and leaves the reader with the impression that it should be the subject of a separate undertaking. Although the first two sections consider only traumatic pathology, the population considered here is far more heterogeneous. It is much less mature in its exposition and, arguably, not yet appropriate for evidence-based medicine (EBM) treatment. The authors effectively create their own premise, define their own terms, and then present recommendations based on that terminology in one fell swoop—bypassing the typical germination of a topic in the literature that engenders equipoise or controversy and, in turn, lends itself to an eventual effort to answer specific questions. The authors assert that there is “a pressing need to standardize nomenclature.” The intent of this effort might be a good first step, but it also underscores the immaturity of the topic. The authors go on to assert that the results of various systematic reviews should be synthesized to provide a “summary” of the accuracy of intraoperative neuromonitoring and potential indications. This proposed amalgamation of systematic reviews runs counter to best practice in data management for guideline construction and further suggests this topic is more appropriate for a literature review than an EBM guideline. The reader might be asking why these seemingly very technical criticisms warrant a public airing. This manuscript has been published online in the flagship journal of its cosponsor AO Spine—a journal that will not entertain external, unsolicited commentary on its content in the name of encouraging a balanced perspective. Once this document has been disseminated and can be retrieved through an electronic search for the generic phrase “SCI guideline,” all is lost. The average reader will not venture beyond the bullet point recommendations to probe the validity and reliability of the conclusions presented. Third parties will represent the recommendations contained therein as a “benchmark,” whether to inform the creation of performance measures, determine payment on a claim, or argue that care rendered in the community did not meet some artificial standard. In this context, there is well-founded concern that the recommendation regarding the timing of surgery, in particular, might be enforced in a manner that is not practical in the community setting (let alone many academic centers) where the resources required to perform a complex spine intervention safely and efficiently simply cannot be marshalled at all hours of the night. The data do not support a significant difference in outcome, whether a procedure is performed at 18 or 28 hours postinjury. This recommendation appears to treat all SCI as equal and does not adequately account for the intricacies that influence clinical decision making, such as the severity of neurological injury, the presence or absence of ongoing neurological compression, clinical stability to tolerate early surgery, and medical comorbidities. Similarly, the recommendations regarding the use of intraoperative neuromonitoring are supported by a very low quality of evidence and are primarily based on the consensus opinion of experts. These factors should imply a limited value to clinical practice. Yet, the repercussions of a document that suggests mandatory monitoring for high-risk patients and then defines “high-risk” to include virtually anyone with a traumatic mechanism are potentially profound, particularly when such a recommendation runs counter to the community standard of practice. Words matter. It is our responsibility as guardians of evidence-based practice to point out where guidelines fall short to protect providers and patients alike.