University of Pittsburgh School of Medicine Department of Anesthesiology Department of Anesthesiology, VA Pittsburgh Healthcare System VAMC 3N237 University Drive C Pittsburgh, PA 15240 Phone: 412-360-6770 FAX: 412-360-6906 Dear Editorial Board Members: Manuscript Title: GlideScope Use Improves Intubation Success Rates: An Observational Study using Propensity Score Matching. Corresponding Author: Michael P. Mangione e-mail: mangionemp@anes.upmc.edu We have revised our manuscript and would like to thank the reviewers for their time and effort. Below you will find our point-by-point reply, and we have submitted a revised manuscript for consideration for publication in BMC Anesthesiology. As stated with our initial submission, we believe that this article will be of interest to the scientific community because it demonstrates, in a large-scale, real-world application, that the GlideScope is associated with greater first pass intubation success than direct laryngoscopy even though the GlideScope was used with more frequency in patients with predictors of difficult intubation. Because this was a prospective, observational study, we have used a propensity score matching algorithm in an attempt to accurately characterize the difference between the techniques. This full manuscript has not been published previously, either in whole or in part, and is not under consideration for publication elsewhere. A portion of the material within was previously presented at the American Society of Anesthesiologists Annual Meeting. Oct, 2011. Chicago, IL, and was the subject of an article in Anesthesiology News. All authors attest to the originality of the text, and the originality of any/all supporting tables, images, and supplementary electronic materials as related to this document. We also hereby affirm that ethical approval for this work was obtained as appropriate to this work and that all authors have made material contributions to this manuscript according to the rules of authorship. As detailed with the manuscript, all authors deny any competing interests. All authors deny issue with the editorial policies detailed in the instructions for authors. Finally, all authors accept that the corresponding author will be the sole author of further communication with the editorial office related to this manuscript, including any and all revisions, and he will have the authority to communicate on behalf of all authors in regards to further correspondence with the Editor and other necessary parties. Yours sincerely, Michael P. Mangione Reviewer #1: Minor Essential Revisions Important unacknowledged limitation of the study is the categorical description of risk factors such as Mallampati, cervical range of motion, mouth opening, dentition and past intubation history. In reality, these are not categorical but continuous. The decision to make them categorical may have been necessitated by propensity scoring but it differs from reality and is a limitation of the study that should be acknowledged. We agree that several of these factors have been historically described in continuous terms, especially CROM and mouth opening, and we have added the use of categorical descriptors for these to the discussion of our limitations. Mallampati has been used in the literature as either a continuous or a categorical element; but dentition is typically judged to be either present or absent. Likewise for past intubation history (difficult or not), thus we believe these should be strictly categorical. We believe that the majority of anesthesiologists do not truly measure CROM or mouth opening; instead we contend that most look at the airway and decide if the element is concerning. We initially tried to highlight this belief in the discussion when we noted that “… regarding these factors because we feel the method we employed represents routine anesthesiology practice, in which the overall impression is used more than specific numeric measurements of elements of the airway exam”. Our revised limitations paragraph in the Discussion now contains: “Additionally, our study used categorical descriptions for airway exam elements that were initially described in continuous terms to predict difficult visualization and/or intubation with direct, line-of-sight laryngoscopic approaches. Examples include CROM, mouth opening, and Mallampati score. The results might vary if continuous numerical parameters were used.” Please state if the protocol limited the number of permissible laryngoscopic attempts (with either device) before declaring failure. Was the number of attempts documented? If they were, they should be stated (i.e. number requiring 2 or > 3 attempts rather than the total number of attempts) To clarify our response to this concern, we defined failures as “Removal of the blade or giving the laryngoscope handle to another practitioner was considered an intubation failure”. For this reviewer’s concern, we are assuming that by failure he is referring to an overall failure to intubate. Our study did not limit the number of attempts with any airway device, nor did it specify what should be done after a failure with any particular device or with the overall intubation process. It was an observational study where the events of each intubation were simply recorded. We clarified this in the Methods section: There were no limitations on the number of laryngoscopic attempts with any device, and actions to be taken in the event of intubation failure were not specified. The number of attempts was documented, but not presented in the initial manuscript for brevity. Analysis of this data would indeed be complex: sub-groupings of patients would be needed for 2, 3 and 4 attempts, and the low number in each would make any conclusions purely speculative. Thus, first-pass success was our defined major outcome, and we prefer to limit our manuscript to the discussion of this focused endpoint (in line with many other airway device studies). We have added a summary for intubations that required more than 3 attempts to the Results section with the following text: There were only 13 patients that required more than 3 attempts. A single patient was not successfully intubated on the fourth attempt (three failed attempts with direct laryngoscopy, then one failed attempt with a GlideScope). This patient had a severe coagulopathy and the decision was made to use a flexible bronchoscope in order to avoid further trauma and potential bleeding from the continued use of a rigid device. After the 4th National Audit Project (16), “major airway complications” are thought of as brain injury, brain death, emergent surgical airway or airway-related ICU admission. Although Aziz et al. (5) referred to “more serious [airway] complications” (0.3%) in truth, these consisted of “one vocal cord trauma, one tracheal injury, one trauma to the hypopharynx, one tonsillar perforation and two dental injuries”. It would be more appropriate to describe these more completely or refer to them as relatively minor injuries. We would like to thank the reviewer for highlighting the minor nature of these injuries. Because of their minor nature, and because we did not further record details or follow the patients post-op, we are unable to describe them more completely. Thus, we have added the word “minor” to our discussion of complications in the Discussion section. This reviewer is not convinced of the value of including simulation studies (6, 7) performed by untrained individuals (6, 8). If the authors elect to retain these references, the limitations of such studies should be mentioned. There is much more useful reference material to draw from. Reference 9 involves the use of the GlideScope in a manner that disregards the manufacturers recommendations. We agree with the above points that a reliance on simulation in untrained individuals or use of nonrecommended equipment limits the applicability of these studies to everyday operating room practice. But we feel these studies still add to the points we are making in this study, thus we have kept them in but added wording to point out these facts to the Introduction section: “… although the applicability of these to real-world practice could be questioned. The overall experience of the operator may be less important than his/her prior experience using the GlideScope. We agree with the reviewer’s statement, as we have seen medical students with GlideScope experience perform intubations easily with the GlideScope and watched as experienced largyngoscopists struggle in their first attempts. We also believe that the varied experience of the participants in our study leads to results that reflect the expected outcomes in real world, as we had stated in our initial submission. Finally, the reviewer’s sentiment is why we have not focused on laryngoscopist level in the manuscript, as GlideScope experience is likely more important but was not well tracked through the study. “At least two GlideScope devices were available every day” provides insufficient information regarding availability. How many operating rooms are there in the facility? Are these single-use or reusable devices (sterilization may remove them from service for several hours)? We have clarified our claim that GlideScope availability was not a concern during our study with the following addition to the Methods section: Our facility consists of ten sites where general anesthesia may be provided; at least two GlideScope devices were available every day to the practitioners during this study. Additionally, our supply of cleaned, reusable blades was maintained at a level that ensures GlideScope blades are available for each intubation over the course of a clinical day. There were no recorded cases where the GlideScope could not be used due to a lack of availability. Please provide a citation for Samsoon and Young (Methods/OutcomeMeasures). Added: Samsoon GLT, Young JRB. Difficult tracheal intubation: a retrospective study. Anaesthesia 1987; 42:487-490. Were any guidelines provided regarding the classification of CROM, mouth opening, what constitutes the presence or absence of teeth, normal or abnormal TMD or a history of difficult intubation? Because we were focused on observing “real world” practices, we did not dictate that these criteria were to be objectively measured. For example, we did not ask practitioners to measure mouth opening and classify someone with less than 4 cm as being reduced. If an anesthesiologist choose to measure, however, they were not discouraged from doing so. We did, as stated in the Methods, discuss the importance of thorough completion of the data from. We have added some text specifically stating such: “The categorization of the above was left to the anesthesiologist; i.e., a numerical cut-off was not provided for these measures.” Please justify the rationale behind not requiring a stylet when using the GlideScope when the manufacturer recommends this. Have you recorded how often intubation was easily accomplished on a subsequent attempt if a stylet was used? Might this have further increased your first-pass success rate with the GlideScope? The reviewer’s comment here is an excellent point. We did not require a stylet to be used because this was an observational study, and requiring specific practices such as these would have violated our IRB protocol. The number of GlideScope intubations performed without the GlideScoperecommended stylet was quite small, but not zero, and may have altered the success rate. We have added text to the Discussion addressing this point: We should note, however, that ten of the GlideScope intubations mentioned here were performed without the manufacturer-recommend stylet, and that this may have altered the number of “inability to pass the tube” complications. RESULTS: Please clarify whether 3,384 represent the total number of patients undergoing anesthesia and 3,139 the number intubated by either DL or GS? That is correct; we have changed/added that text in the Results to now read: There were a total of 3384 patients in the entire study, with 3139 of these in either the DL or GS groups. Devices not included in this study were used for the other intubations. Please explain the apparent discrepancy between the statement in METHODS that 313 patients were in the GlideScope first-pass group and the second sentence in RESULTS that the GlideScope was used as the first-pass device in 643 [patients].” The apparent discrepancy results from the fact that only a subset of the GS patients were able to be matched to similar patients in the DL group for the propensity score matching analysis. Propensity score matching used the parameters deemed significant (Mallampati class, CROM, normal mouth opening, presence or absence of teeth, and weight, for this study) to identify similar subpopulations of larger groups so that comparisons are “apples-to-apples”. We have altered the text in the Methods to help make this clear: 313 of the 643 patients in the GlideScope first-pass group were able to be matched via a nearest neighbor algorithm (with one-to-one matching and no caliper definition) to 313 patients in the DL group, and their results were compared. Please clarify whether a history of difficult airway excluded patients with prior difficulty using a video laryngoscope. Quite the contrary! We found that our practitioners preferred to use the GlideScope with patients who had a history of a difficult airway. This is evident in the Table 1 results. Patients with history of a difficult airway were excluded only from the propensity score matching as we were unable to find matching patients with difficult airway history for the two groups. Please confirm that TABLE 3 is properly represented. Is it possible some or all of the GS and DL values are displayed in the wrong columns? More complications were noted with DL than GS laryngoscopy though, the p values do not reflect statistical significance with the exception of mucosal injury. Dental injury, esophageal intubation, hypoxemia and “others” are substantially higher in the DL group yet the p value is insignificant. Please confirm these calculations or the accuracy of the Table. We can confirm that Table 3 is indeed correct. As stated in the table legend, the numbers within the table are absolute counts. The significant differences are present because of the difference in the number of attempts with each device: 3.59% of the GlideScope intubations results in “Able to visualize but unable to pass”, while that was the case for only 1.57% of the DL intubations. We have clarified this by adding the percent of total attempts to the table for those with significant Pvalues and added a note of this to the legend. DISCUSSION: “Laryngoscopist with varying levels of expertise are included in the study’s outcomes…” It would be helpful to know that the operators in the DL and GlideScope limbs were similar with respect to their “levels of expertise”. We have been provided with no information about who attempted the intubations in either group. We do not believe that laryngoscopist level significantly affected the results. It was initially left out of the manuscript for concerns about the results being over-interpreted: we did not want a subgrouping of first-pass success rate as a function of level to become a key point when it was underpowered to reveal true differences in ability. We attempted to address this with our statement in the Discussion that “Future subgroup studies focusing on expertise level would necessitate an increased number of subjects.” However, this comment clearly shows that mention of it being approximately equal across the GS and DL groups for the propensity score analysis is needed. Therefore, we have added it in to the Results and the Discussion with the flowing text: Results: Practitioner level was not found to significantly affect the success rate for the 626 patients in the propensity score analysis as the majority of intubation attempts were done by certified registered nurse anesthetist in both groups (51.4% for the GS group and 53.4% for the DL group). Discussion: Moreover, our study’s data reflect the performance of all providers performing laryngoscopy – including medical and nurse anesthesia students, respiratory therapists, all levels of anesthesiology residents, Certified Registered Nurse Anesthetists, and Attending Anesthesiologists. This variable was not felt to affect the results as the groups were well matched on this variable for the propensity score analysis. Future subgroup studies focusing on expertise level would necessitate an increased number of subjects. Paragraph 2: “…the initial hypothesis was confirmed: the first pass success rate utilizing video laryngoscopy was higher”. I recommend that you restrict your observations to the specific device that was used rather than video laryngoscopy in general. No other device was evaluated, so it remains to be determined whether this statement applies to other video laryngoscopes. Thank you for this well-reasoned comment. We initially planned to compare the C-MAC, the GlideScope, and direct laryngoscopy approaches, however our C-MAC experiences during the study period prevented this; C-MAC usage was mostly related to teaching. We have revised the text to reflect the fact that only the GlideScope was used. Furthermore, we also tempered our statement about the mucosal injuries by adding the word “minor” as discussed in a prior bullet. A greater first-attempt success rate was found when using the GlideScope versus direct laryngoscopy. In addition, the GlideScope was found to be 99% successful for intubation after initial failure of direct laryngoscopy. In our study, however, this was found to be at the expense of a higher rate of minor mucosal injury. Paragraph 3: “Recently Aziz et al reported similar improvement in first pass success with a video laryngoscope…” Here the authors are introducing another video laryngoscope into the discussion. It would be more appropriate to comment on the GlideScope publication by Aziz (5) describing its use in 2,004 patients as a primary or rescue device. This paper does not describe video laryngoscopy in general but rather the GlideScope in particular. It is important for them to carefully consider what new information this study provides beyond that of Aziz et al. (5). If they wish to broaden the discussion by speculating on the role of video laryngoscopy in general, this should be done only after they have adequately discussed the GlideScope in particular. The same applies to the Jungbauer publication (14). As we stated above, we initially planned to discuss both the GlideScope and the C-MAC, however our C-MAC experiences during the study period did not allow a true comparison. We agree with the reviewer’s that this section deviates from the focus of the paper. As a result, we have reorganized the first three paragraphs of the Discussion extensively, including the de-emphasis of the Aziz (13) and Jungbauer (14) publications. The value of including medical and nurse anesthesia students, respiratory therapists and others is limited by the absence of information pertaining to their representation and whether this was evenly reflected in the two groups. We believe this was covered with the text additions and reorganization addressed above. The study had a very substantial male imbalance (93 and 94%; is this what is meant by a veteran population?). This requires clarification. The population in the VA system is predominately male. We have attempted to further clarify this by changing the text in the Results to: Given that the typical patient at a Veterans Administration hospital is that of older male patients, the patient characteristics for the DL and GS were consistent with expectations and are given with p-values for the comparisons in Table 1. It may be true that a consensus does not exist regarding the predictors of a difficult GlideScope visualization however they should refer to Aziz (5) and Tremblay M-H et al. (Poor Visualization During Direct Laryngoscopy and High Upper Lip Bite Test Score Are Predictors of Difficult Intubation with the GlideScope® Video laryngoscope. Anesth & Analg 2008; 106: 1495-500). We would like to thank the reviewer for this suggestion, and have changed our Discussion text to: Finally, a consensus on the parameters defining “difficult GlideScope visualization” has not yet been reached, although the literature suggests that a high Cormack and Lehane grade, high upper lip bite, and short sternothyroid distance[21] and altered neck anatomy[5] are potential indicators. Minor Discretional Revision INTRODUCTION: second paragraph—I suggest that “over 71,000” and “more than 2000 GlideScope uses” be replaced by 71,570 and 2,004 respectively. Done. Reviewer 2: Major Compulsory Revisions Abstract: The second part of the hypothesis refers to differences between complication rates. However, Neither in the results nor in the conclusions complication rates are mentioned. We regret that, upon a close review prompted by this reviewer’s comments, we also found the confirmation/rejection of our complication hypothesis not clear throughout our manuscript. We have revised the Abstract to remove the reference to complication rate, as this was a secondary aim. The Introduction was also altered to highlight the secondary nature of complication rate. The Discussion sections to clarify our hypothesis and findings. Abstract: We hypothesized that the first pass success rate with one particular video laryngoscope, the GlideScope, would be higher than the success rate with direct laryngoscopy. Introduction: We hypothesized that the first pass success rate with the GlideScope would be higher than the success rate with direct laryngoscopy regardless of the experience of the operator. Our primary aim was to collect data to confirm the aforementioned hypotheses, using first pass success as a measurement of intubation ease. Secondary aims included the identification of the patient morphologic factors that influenced anesthesiologists to select the GlideScope because of the potential for a “difficult airway” and the determination of complication rates, as we further hypothesized that the complication rates between the two would be no different. Results: Table 3 details the complications recorded during the study. Comparison between the GS and DL groups showed that the patient complications within the GS group were significantly greater only in regards to mucosal injury (1.5% vs 0.5%) and the ability to visualize the cords but an inability to pass the endotracheal tube. While the percentage of esophageal intubations was higher in the DL group (0.9% vs 0.3%), it did not reach statistical significance. Discussion: Complications associated with intubation (or with the inability to intubate) can cause significant patient harm.[16,17] Our study found that while most of the complication rates were equivalent, there was an statistical increase in mucosal injuries in the GS group. Injuries of this type were minor and included abrasions within the oropharynx and lip lacerations. These are well described in existing literature. p5, l20: Video laryngoscopic intubation time differs with regard to the blade type. Therefore, the authors should compare the effects of indirect curved vs. indirect macintosh blades. The Macintosh is a curved blade, therefore in the comment we are assuming that the reviewer is suggesting that the C-MAC (an indirect Macintosh blade) be compared with the GlideScope (which is curved, but with a different profile than the Macintosh blade). The effect of intubation time with respect to blade type, although interesting, was not a focus of this study. Time was not tracked; therefore we would not be able to comment on the effects of intubation device on time. Furthermore, as discussed in the paper, C-MAC to GlideScope comparison was not possible due to the lack of appropriately-matched cases. p13, l2: The low incidence of emergent and difficult airway cases suggest a primary use of the GlideScope in these cases. The reviewer is correct in that a majority of the emergent and suspected difficult airway intubations were performed with the GlideScope, and this is what led to the exclusion criteria cited for the propensity score matching as we were unable to find matching patients with difficult airway history for the two groups. p14, l2: Mucosal injuries are a well known problem of curved indirect blades. Agreed, which is why mucosal injuries were specifically tracked as a complication, and why the Discussion stated that “This, too, is often cited with the GlideScope, and has been substantially addressed within the literature.” p14, l5: Why are there any esophageal intubations in the video laryngoscopy group (intubation under vision)? A post-intubation debriefing did not take place; therefore we are unable to state why there were GS group esophageal intubations with certainty. We can suppose, however, that this likely occurred if the laryngoscopist ceased to look at the screen when removing the stylet from the tube, with the tip of tube then falling out of the glottis opening. Advancing the tube at that point could have caused an esophageal intubation. p15, l4: The varying level of expertise within laryngoscopists is interesting; however, it limits the conclusions of this study We agree that expertise may affect the intubation success rate. From our responses to Reviewer 1, we have seen medical students with GlideScope experience perform intubations easily with the GlideScope and watched as experienced largyngoscopists struggle in their first attempts. We also believe that the varied experience of the participants in our study leads to results that reflect the expected outcomes in real world, as we had stated in our initial submission. Finally, we have not focused on laryngoscopist level in the manuscript, as GlideScope experience is likely more important but was not well tracked through the study. But, we do not believe that laryngoscopist level significantly affected the results. It was initially left out of the manuscript for concerns about the results being over-interpreted: we did not want a subgrouping of first-pass success rate as a function of level to become a key point when it was underpowered to reveal true differences in ability. We attempted to address this with our statement in the Discussion that “Future subgroup studies focusing on expertise level would necessitate an increased number of subjects.” We have added the following to the Results and the Discussion: Results: Practitioner level was not found to significantly affect the success rate for the 626 patients in the propensity score analysis as the majority of intubation attempts were done by certified registered nurse anesthetist in both groups (51.4% for the GS group and 53.4% for the DL group). Discussion: Laryngoscopists with varying levels of expertise are included in the study’s outcomes as is normally found in an academic teaching institution. This variable was not felt to affect the results as the groups were well matched on this variable for the propensity score analysis. Future subgroup studies focusing on expertise level would necessitate an increased number of subjects. p15, l13: But the second part of the hypothesis was that there would be no difference in complication rates. Clarified, as described in the reply to this reviewer’s first comment. p15, l21: Since in the C-MAC device it is possible to interchange the blade types (Macintosh, DBlade, Miller), it is important to mention, which blade type has been used in the reported findings. We only had the Macintosh blade available to us; this has been added to the Methods section. p16, l20: This is the major shortcoming of this investigation: It again compares one video laryngoscope with direct laryngoscopy. What we need are studies that compare different video laryngoscopy devices. We agree and had initially planned to compare the C-MAC, the GlideScope, and direct laryngoscopy approaches. Our database included information regarding all C-MAC intubations, however our CMAC experiences during the study period prevented an appropriate comparison as the primary reason for the use of the C-MAC was training. We have revised the text to reflect the fact that only the GlideScope was used. We agree with the need for video laryngoscopy comparisions, but still believe that our study adds to the literature with its size, its prospective observational design, and its focus on the everyday use of rigid direct and rigid video laryngoscopy in combination. p17, l6: What was the escape strategy if GlideScope intubations failed? This has to be mentioned. A specific escape strategy was not specified in the event of GlideScope failure, similar to the fact that GlideScope use was not mandated if direct laryngoscopy failed. We follow the ASA’s difficult airway algorithm for each intubation, thus if the GlideScope failed the Anesthesiologist would have moved to the “Can’t Intubate” portion of the algorithm and would have proceeded from there. The following was added to the Methods as a result of this and Reviewer 1’s comment: There were no limitations on the number of laryngoscopic attempts with any device, and actions to be taken in the event of intubation failure were not specified. p27: Why are mucosal injuries statistically different (15 vs. 12), while the numbers of esophageal intubations are not (24 vs. 2)?? As we detailed in a response to Reviewer 1, and as stated in the table legend, the numbers within the table are absolute counts. The significant differences are present because of the difference in the number of attempts with each device: 3.59% of the GlideScope intubations results in “Able to visualize but unable to pass”, while that was the case for only 1.57% of the DL intubations. We have clarified this by adding the percent of total attempts to the table for those with significant Pvalues and added a note of this to the legend. With regards to the initial hypothesis, the authors seem to lurch between confirmation and rejection of their hypothesis. Clarified, as described in the reply to this reviewer’s first comment. Minor Essential Revisions Literature 11 should be in the same style. Corrected, thank you.