About

Clifford S. Deutschman, M.S., M.D., F.C.C.M., is an Emeritus Professor of Anesthesiology and Critical Care at the University of Pennsylvania's Perelman School of Medicine. He serves as an Attending Physician in the Surgical Critical Care Service and is involved in anesthesiology at the Hospital of the University of Pennsylvania, including the Lamont Operating Rooms and the Bariatric Surgery Program. His educational background includes studies at Fieldston School, Trinity College, Northwestern University, and New York Medical College. His professional work encompasses critical care research and clinical practice, with a focus on anesthesiology and critical care medicine. Deutschman has contributed to the field through research on critical illness, hepatocellular regeneration in sepsis, and lung transplant outcomes, among other topics. His publications include studies on critical care funding, the burden of critical illness, and the pathophysiology of graft dysfunction after lung transplantation. His work is characterized by a focus on improving understanding and treatment of critical conditions in surgical and intensive care settings.

Research topics

• Medicine
• Intensive care medicine
• Internal medicine
• Immunology
• Biology

Selected publications

• T cell memory affects lung responses to cecal ligation and puncture.

  Shock · 2026-03-16

  article

  Murine sepsis models are limited by an inability to recapitulate several common features of human sepsis. One possible explanation is that laboratory mice lack the robust preexisting memory T cell repertoire that is a key feature of the human immune system. We therefore investigated how inducing T cell memory by treating C57BL6 mice with anti-CD3ε activating antibody ("Immune-Educated" mice) affected the pulmonary immune response to the cecal ligation and puncture (CLP)1 model of sepsis. Twenty-four hours after CLP, Immune-Educated mice had higher alveolar inflammatory cytokine and chemokine concentrations and more pulmonary interstitial macrophages than what was observed in untreated ("Uneducated") animals. After 72 hours, there were more alveolar macrophages in the lungs of Educated mice. In a separate experiment, we performed adoptive transfer of memory CD4 and CD8 T cells from immunized C57Bl/6J to B6.SJL mice. Interstitial macrophage recruitment 24 hours post-CLP was more pronounced in mice undergoing adoptive transfer of memory T cells compared to mice that did not undergo adoptive transfer. Finally, to evaluate whether observed differences in CLP-induced lung inflammation between Educated and Uneducated mice are driven by IFN we subjected Educated and Uneducated mice to IFN blockade at the time of CLP. IFN blockade resulted in higher absolute numbers of T cells, memory T cells, and innate cells in the lungs of Educated mice 24 hours post-CLP suggesting that IFN acts as an immune-regulator and curbs an overactive immune response in these mice. In conclusion, the presence of memory T cells affects the course of the lung immune response to CLP.

  PublisherDOI

• IL17F+ naïve and IFNγ+ memory CD8 T cells drive hepatic dysfunction in the cecal ligation and puncture model of sepsis

  Molecular Medicine · 2026-01-08

  articleOpen accessSenior author

  T cell memory significantly alters the immune response and organ dysfunction induced by the murine cecal ligation and puncture (CLP) model of sepsis. Enhanced T cell memory activation promotes hepatic neutrophilic responses and induces hepatic dysfunction, which is a common complication of human sepsis associated with poor outcomes. We used a novel Immune Educated CLP sepsis mouse model to examine the role of memory T cell cytokine responses in driving innate immunity and organ dysfunction. Through this approach, we found that induced T cell memory prior to CLP led to higher serum levels of IL12, TNF, IL17, and IL1β – all dependent on memory CD8 T cell derived IFNγ following CLP. IFNγ induced activated hepatic IL12+ monocyte-derived dendritic cells. Increased neutrophilic responses occurred in Educated CLP which was found to depend on TNF, and were suppressed by IFNγ. Hepatic dysfunction in response to CLP was worsened by CD4 and CD8 T cell immune memory and prevented by IFNγ and IL17F blockade. These findings were recapitulated in naturalized outbred pet shop mice with natural immune memory to provide translational relevance to our Immune Educated CLP sepsis model. IL17F or IFNγ blockade may represent potential targets for treatment in sepsis with hepatic dysfunction.

  PublisherDOI

• T cell memory alters pulmonary inflammatory responses to cecal ligation and puncture

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-05-19

  preprintOpen access

  To date, murine sepsis models have failed to recapitulate human acute respiratory distress syndrome, one of the leading complications of human sepsis. We set out to determine if preexisting T cell memory, which is common in human adults and lacking in laboratory mice, could contribute to lung inflammation in the cecal ligation and puncture (CLP) model of sepsis. After administering an anti-CD3ε activating antibody to C57Bl/6 mice to induce a T cell memory repertoire, we compared the pulmonary immune response to CLP in these "Immune-Educated mice" to responses observed in Uneducated control animals. Compared to Uneducated mice, 24 hours after CLP, Immune-Educated mice had higher alveolar inflammatory cytokine and chemokine concentrations and more pulmonary interstitial macrophages. After 48 hours, the proportion of effector CD4 T cells that produced interferon-gamma was greater in Immune-Educated mice. After 72 hours, there were more alveolar macrophages in the lungs of Educated mice. Separately, we performed adoptive transfer of memory CD4 and CD8 T cells from immunized C57Bl/6J to B6.SJL mice and IFNγ blockade at the time of CLP. Interstitial macrophage recruitment 24 hours post-CLP was more pronounced in mice undergoing adoptive transfer of memory T cells compared to mice that did not undergo adoptive transfer. IFNγ blockade resulted in higher absolute numbers of T cells, memory T cells, and innate cells in the lungs of Educated mice 24 hours post-CLP suggesting that IFNγ is necessary for curbing an overactive immune response in these mice. In conclusion, the presence of memory T cells affects the course of CLP-induced lung inflammation and may provide a model that more closely resembles sepsis-associated lung injury.

  PublisherOA PDFDOI

• Knowledge Transfer in the 21st Century: The Continuing Evolution of Critical Care Medicine

  Critical Care Medicine · 2025-01-01 · 1 citations

  article
  PublisherDOI

• Optimization of Preclinical Rodent Research Models of Human Shock: Part One Intra-Abdominal Sepsis

  Shock · 2025-11-03 · 2 citations

  article

  Preclinical models using animals are crucial for medical advancements despite their limitations and criticisms. Critical illnesses like sepsis, trauma, and burns remain huge causes of morbidity and mortality despite medical advances, and human studies may not always be feasible. In this part one of two reviews about animal models for critical illness, we discuss sepsis and the considerations one should take to optimize the rodent sepsis model. There are multiple models of sepsis used, each with advantages and disadvantages, and they can be modified to reflect how patients are treated in the hospital, including intensive care unit care. Patient factors such as age, sex, and comorbidities are important considerations given the different responses to sepsis. Aspects of sepsis that our patients encounter, including muscle and neurocognitive dysfunction, can be modeled to try and improve those aspects of outcomes. Choosing the right models for the question one is asking and optimizing that model is key to recapitulate the human condition to make animal models more translatable to humans. In other words, we suggest that, in lieu of abandoning animal models of sepsis, we seek to enhance translatability to the human condition.

  PublisherDOI

• Value of animal sepsis research in navigating the translational labyrinth

  Frontiers in Immunology · 2025-04-15 · 10 citations

  articleOpen access

  Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection (1), as manifested by early activation of both pro-and anti-inflammatory responses (2), along with major alterations in non-immunologic pathways such as cardiovascular, neuronal, autonomic, hormonal, bioenergetic, metabolic, and coagulation (3). It accounts for almost 20% of total deaths worldwide (4), and annually costs more than $60 billion in the U.S. alone. The onset of the disease and the intricate interplay of various immune cells, inflammatory mediators, signaling pathways, and organ systems makes studying sepsis in humans ethically and logistically challenging. This necessitates the use of animal models to systematically dissect its intricate pathophysiology and evaluate potential therapies in a controlled setting.Animal models allow researchers to manipulate key variables such as infection type and severity, intervention timing, and the genetic background (e.g., gene knockout or knock in strategy) of experimental animals (5)(6)(7)(8). This level of control enables researchers to isolate the effects of specific interventions and identify potential therapeutic targets, such as tumor necrosis factor (TNF) (9), high mobility group box 1 (HMGB1) (10), cold-inducible RNA-binding protein (CIRP) (11), sequestosome-1 (SQSTM1) (12), and procathepsin L (pCTS-L) (13). Moreover, these models allow for tracking the temporal progression of sepsis from initial insult to subsequent organ dysfunctions (14) and eventual outcomes (7,8), offering invaluable insight into the complex interplay of multiple pathophysiological processes (15,16), including hyperinflammation (17), immunocoagulation (18,19), pyroptosis-mediated immune cell death (20), and immunosuppression (21,22). Given the influence of comorbidities and other factors on disease progression and treatment response in human sepsis, it is essential to incorporate comorbidities (e.g., diabetes, hypertension, or coronary artery disease) and pre-existing injuries (e.g., smoke inhalation occurring in burn patients) into animal modeling, thereby improving the translatability of experimental findings into future clinical therapies (6)(7)(8)23).Despite advancements in understanding sepsis pathophysiology, translating preclinical findings into effective human therapies remains challenging, as exemplified by the failure of anti-TNF antibodies in clinical trials (3,24). However, attributing this translational gap solely to the limitations of animal models is an oversimplification (6), because the inherent complexity and heterogeneity of human sepsis, coupled with challenges in clinical trial design, also contribute to this difficulty.Animal models typically use a single, standardized insult in genetically homogeneous animals.However, this genetic and environmental homogeneity of laboratory animals contrasts sharply with the genetic and environmental diversity of human populations, as well as the variety of infections in clinical sepsis (8,25). The inherent heterogeneity in septic patients is further compounded by other factors such as age, sex, underlying health conditions (comorbidities), environmental exposure/history, and time to treatment initiation (2). Because patient variability often creates a broad spectrum of pathophysiological endotypes, it is important to develop animal models to recapitulate some human sepsis endotypes. While comprehensive immune profiling (cytokine/chemokine levels, immune cell function, gene expression) can potentially characterize "endotypes" in animal models, their accuracy in reflecting human sepsis endotypes (such as hyper-or hypo-inflammatory states) remains unclear (6)(7)(8), presenting challenges for translational research. Thus, the failure of identifying and recruiting homogenous patient subgroups in previous clinical trials might have diluted treatment effects due to potential outcome variations (8,23,26).In addition, potential differences in immune responses between animals and humans may pose another significant challenge (27). Although genomic comparisons between mouse models and human sepsis have revealed significant similarities (28,29), they also highlight some noticeable differences (30), underscoring the complexity and difficulty of extrapolating experimental findings across species (Table 1). Therefore, developing more diverse and sophisticated animal models that incorporate polymicrobial infections, comorbidities, and genetic variability is crucial to improving translatability (7,8). For example, a refined murine sepsis model that adhered the Minimum Quality Threshold in Pre-Clinical Sepsis Studies (MQTiPSS) guidelines (5,31) by incorporating daily chronic stress closely recapitulated the genomic and phenotypic responses observed in human surgical sepsis (32). Similarly, humanized mice that express human genes or possess a humanized immune system may similarly offer a more promising approach to enhance translatability (7,8,33). Conversely, stratifying patients based on specific biomarkers (34,35) indicative of the unique pathobiology one is hoping to modify, along with relevant clinical parameters beyond overall mortality, should similarly enhance the precision and power of future clinical trials (3,(35)(36)(37).The rationale for targeting TNF in sepsis stemmed from its early and prominent role in initiating the inflammatory response (9,38). However, sepsis is a highly heterogeneous syndrome characterized by variable timing for the release and pathogenic actions of various cytokines (17).While TNF may be critical in the initial hyperinflammatory phase, its importance diminishes over time. Therefore, blocking it at a wrong time could even be detrimental to the host by impairing essential immunities needed for pathogen clearance (39)(40)(41)(42). Accordingly, shifting the focus to some later-acting mediators such as HMGB1 (10) and pCTS-L (13), which have relatively wider therapeutic windows (43), may present a more promising avenue for future sepsis trials. Given the multifaceted nature of sepsis, combinatorial therapies targeting multiple mediators may be more effective than targeting a single cytokine. Such combination therapies, if tailored to the specific cytokine profile and disease stage of individual patients, may offer a more personalized and effective approach to sepsis treatment (8,35,44), unlocking the therapeutic potential of cytokine-targeting therapies for this devastating condition.The dynamic nature of sepsis requires timely interventions and optimal dosing regimens (45,46), which are also difficult to translate from animal models to human clinical trials. Simplified dosing regimens used in animal models (32) often struggle to capture the complex pharmacokinetics and pharmacodynamics observed in humans (3). Beyond the heterogeneity of sepsis patients, interspecies differences in these kinetic and dynamic parameters further complicate translation, necessitating personalized dosing algorithms based on individual patient characteristics such as age, sex, weight, comorbidities, and disease severity.Although directly translating animal sepsis research to human sepsis therapies remains challenging, the knowledge gained from animal models has advanced treatments for other inflammatory diseases, such as rheumatoid arthritis (RA) (47), Crohn's diseases (48), and ulcerative colitis (49).RA, a chronic autoimmune disease affecting 0.5-1% of the global population (50), shares unexpected commonalities with sepsis in its inflammatory pathways, particularly the involvement of TNF (51,52) and other cytokines. The identification of TNF as an early mediator of sepsis, largely through animal models (9,38), paved the way for the development of anti-TNF biologics like infliximab, etanercept, and adalimumab (53,54) as cornerstone therapies for RA (53, 55), Crohn's diseases (48), and ulcerative colitis (49). These success stories have exemplified the broader impact of sepsis research using animal models, extending beyond sepsis itself to benefit patients with other inflammatory conditions. It highlights the value of fundamental research in revealing unexpected connections between seemingly disparate fields and potentially driving significant therapeutic advances for many inflammation disorders.Animal models remain indispensable in sepsis research, providing a controlled setting to unravel complex pathophysiological mechanisms, identify therapeutic targets, and evaluate novel interventions. While acknowledging their limitations and actively refining these models is crucial, abandoning animal research would be a short-sighted setback hindering scientific progress (6).Therefore, the future of sepsis research still hinges on developing more sophisticated and clinically relevant animal models that incorporate age, sex, polymicrobial infections, comorbidities (e.g., diabetes, hypertension, or coronary artery disease), pre-existing injuries (e.g., smoke inhalation in burn patients) (56), and genetic diversity, thereby better reflecting the complexity and heterogeneity of human sepsis. Animal models of sepsis can be developed in a vast array of different animal species (Table 1), but each species possesses unique strengths and weaknesses that influence the translatability of research findings. Therefore, model selection depends on the specific research question and a balanced assessment of species-specific strengths and weaknesses.For instance, a multi-species approach, strategically incorporating larger animal models like pigs or non-human primates when necessary, might be needed to bridge this species gap.Traditional animal sepsis studies rely on limited number of physiological and biochemical markers, and thus lack the granularity to capture the complex molecular landscape of clinical sepsis. It is thus critical to refine animal sepsis models (57) by integrating transcriptomic, epigenomic, metabolomic, and proteomic analyses across various body compartments at single-cell level (58).Comparing these multi-omics profiles between animal models and human sepsis patients will allow for rigorous model validation, ensuring an accurate reflection of dysregulated molecular pathways of human sepsis (32). This improved approach may also help identify key pathways and biomarkers conserved across species, guiding relevant therapeutic target selection to improve the predictive power of preclinical studies.To fully replicate organ dysfunctions characteristic of human sepsis, it is paramount to integrate organ support technologies (e.g., mechanical ventilation, fluid resuscitation, and vasopressor support) into larger animal models by creating "ICU-like" experimental conditions that allow realtime monitoring and modulation of immunological and physiological parameters that mirror clinical sepsis management. Large animal models, like pigs and sheep (56,59), offer a unique platform for sepsis research due to their larger size, as well as immunological and physiological similarities to humans (e.g., heart rate, blood pressure, and lung mechanics), enabling the use of advanced organ support technologies for real-time monitoring of immune and physiological dysregulations during sepsis (Table 1). However, these models are also limited by higher cost, longer experimental duration, species-specific differences, restricted genetic manipulation, and heightened ethical concerns (Table 1).To overcome the translational limitations of animal models in sepsis research, several human ex vivo models have also been developed, including i) whole blood assays (60); ii) precision-cut tissue slices (lung and others) (61); iii) human blood-perfused organ models (62,63); and iv) miniaturized "organ-on-a-chip" systems (64,65). These models can replicate the complex interplay between immune cells, endothelial cells, and pathogens within a controlled microenvironment, offering a more human-relevant platform for studying sepsis (Table 1).Another key strength is their capacity to dissect patient-specific responses, using samples from diverse cohorts (varying ages, comorbidities, genetic backgrounds) to identify individualized responses and to potentially tailor personalized therapies. Despite these advantages, human ex vivo models suffer from many limitations (Table 1), such as the inability to fully replicate in vivo complexity, restricted experimental duration due to ex vivo tissue/cell viability, and challenges in obtaining/utilizing human samples, including ethical approvals, logistical coordination, and cost/expertise limitations.Refining animal models of sepsis requires a multifaceted approach encompassing: 1) the development of more sophisticated and clinically relevant animal models that incorporate age, sex, polymicrobial infections, and comorbidities; 2) the integration of organ support technologies and multi-omics calibration; 3) the strategic use of multiple species; and 4) the selection of multiple more feasible therapeutic targets including HMGB1, CIRP, and pCTS-L. By implementing these refinements, future animal research can more accurately reflect the complexity of human sepsis, thereby enhancing the predictive validity of preclinical studies and accelerating the development of effective therapies for this devastating condition. In addition to refining animal models, improving clinical trial design through better patient stratification based on biomarkers and cytokine profiles may be equally important. The continued pursuit of knowledge through welldesigned animal models, coupled with rigorous clinical research, holds the key to unlocking effective therapies for sepsis and other inflammatory diseases. The unexpected success of anti-TNF therapies in RA, born from sepsis research, serves as a powerful testament to the value of animal research.

  PublisherOA PDFDOI

• Loss of M1 Acetylcholine Receptor-mediated Orexinergic Activity Contributes to Immune Dysfunction in Experimental Sepsis

  Research Square · 2025-08-25 · 1 citations

  preprintOpen accessSenior author
  PublisherOA PDFDOI

• Age influences the circulating immune profile in pediatric sepsis

  Frontiers in Immunology · 2025-01-28 · 7 citations

  articleOpen access

  Background The immune response changes as patients age, yet studies on the immune dysregulation of sepsis often do not consider age as a key variable. Objective We hypothesized that age would influence the immune response in septic children and that there would be a distinct variation in the immune profile in healthy children and children with either sepsis, uncomplicated infection, or acute organ dysfunction without infection. We characterized the circulating immune profile of children presenting to our tertiary care children’s hospital. Methods This investigation was a prospective, observational cohort study that enrolled patients from July 2020 – September 2022. Patients were included if they were < 21 years, admitted to the PICU, and received fluid resuscitation and antibiotics. Peripheral blood mononuclear cells were isolated from samples collected on PICU day 1. Results Eighty patients were enrolled. Children with sepsis had more regulatory CD4 + T cells and memory CD4 + T cells and less CD4 + IL-10 + and CD8 + T-bet + T cells than healthy children. After ex vivo stimulation, sepsis samples had less of a reduction in CD4 + T cells producing IL-10 than healthy controls. Memory CD4 + T cells and regulatory CD4 + T cells were positively associated with age in sepsis alone. Conclusion A regulatory T cell failure may contribute to pediatric sepsis pathogenesis. Age is an important variable affecting sepsis-associated immune dysregulation and memory T cells in peripheral circulation correlate with age in sepsis alone.

  PublisherOA PDFDOI

• Single-cell RNA sequencing reveals Immune Education promotes T cell survival in mice subjected to the cecal ligation and puncture sepsis model

  Frontiers in Immunology · 2024-03-18 · 2 citations

  articleOpen access

  Background Individual T cell responses vary significantly based on the microenvironment present at the time of immune response and on prior induced T cell memory. While the cecal ligation and puncture (CLP) model is the most commonly used murine sepsis model, the contribution of diverse T cell responses has not been explored. We defined T cell subset responses to CLP using single-cell RNA sequencing and examined the effects of prior induced T cell memory (Immune Education) on these responses. We hypothesized that Immune Education prior to CLP would alter T cell responses at the single cell level at a single, early post-CLP time point. Methods Splenic T cells were isolated from C57BL/6 mice. Four cohorts were studied: Control, Immune-Educated, CLP, and Immune-Educated CLP. At age 8 weeks, Immune-Educated and Immune-Educated CLP mice received anti-CD3ϵ antibody; Control and CLP mice were administered an isotype control. CLP (two punctures with a 22-gauge needle) was performed at 12-13 weeks of life. Mice were sacrificed at baseline or 24-hours post-CLP. Unsupervised clustering of the transcriptome library identified six distinct T cell subsets: quiescent naïve CD4 + , primed naïve CD4 + , memory CD4 + , naïve CD8 + , activated CD8 + , and CD8 + cytotoxic T cell subsets. T cell subset specific gene set enrichment analysis and Hurdle analysis for differentially expressed genes (DEGs) were performed. Results T cell responses to CLP were not uniform – subsets of activated and suppressed T cells were identified. Immune Education augmented specific T cell subsets and led to genomic signatures favoring T cell survival in unoperated and CLP mice. Additionally, the combination of Immune Education and CLP effected the expression of genes related to T cell activity in ways that differed from CLP alone. Validating our finding that IL7R pathway markers were upregulated in Immune-Educated CLP mice, we found that Immune Education increased T cell surface IL7R expression in post-CLP mice. Conclusion Immune Education enhanced the expression of genes associated with T cell survival in unoperated and CLP mice. Induction of memory T cell compartments via Immune Education combined with CLP may increase the model’s concordance to human sepsis.

  PublisherOA PDFDOI

• Video Laryngoscopy in Critically Ill Adults: Nascent, Evolving, or Established?*

  Critical Care Medicine · 2024-10-15

  articleSenior author

  Tracheal intubation via laryngoscopy is among the most commonly performed procedures in critically ill patients. Dramatic improvements in the safety of tracheal intubation over time can be attributed to a variety of factors. Perhaps the earliest refinement directed at the risk of failed intubation was institution of immediate post-intubation capnometry to aid in the early detection of esophageal intubation (1). Additional efforts have included educational programs to facilitate skill development without placing patients at risk, data collection on success rates, national quality improvement efforts, and algorithmic guidance (2,3). In parallel, the instruments used to perform intubation and to more directly evaluate the process have evolved. Large-scale studies and meta-analyses have identified cognitive and technical practices designed to increase the efficacy and safety of tracheal intubation (2–4). The impact of these refinements has been studied in operating room (OR), emergency department (ED), and ICU settings (4). Multicenter prospective observational studies have demonstrated that technical difficulties with tracheal intubation in critically ill adults occur less frequently than physiologic disturbances, including cardiovascular instability and hypoxemia (5,6). These same studies have also confirmed that repeated attempts at tracheal intubation are associated with increasingly frequent physiologic complications. In this issue of Critical Care Medicine, McDougall et al (7) offer a methodologically robust systematic review and meta-analysis of randomized controlled trials (RCTs) comparing video laryngoscopy (VL) with direct laryngoscopy (DL) in critically ill adults in either an ED or ICU setting. The authors analyzed 20 RCTs representing 4569 patients, including data from the Direct Versus Video Laryngoscope (DEVICE) trial and others not previously included in similar recent publications (8–10). The analysis identified an overall rate of first pass success of 81.7% with VL and 71.7% with DL. The meta-analysis indicated that VL was associated with a 10.6% increase in first pass success (95% CI, 4.9–17.2%). The study was unable to identify any subgroup (e.g., physical location, elective vs. emergent intubation, device design) where the efficacy of VL and DL differed significantly. The data did suggest that esophageal intubations, aspiration, and dental injuries were less likely with VL, although significance was not achieved. The analysis was unable to demonstrate a difference between VL and DL with respect to failed intubation, peri-intubation physiologic complications, or mortality at any time point studied. Importantly, the authors noted that limitations in outcome reporting increased uncertainty in their analyses. Meta-analysis is only as robust as the examined literature, and the authors stressed that imprecision and uncertainty limited their assessment. Indeed, the definitive results in the study by McDougall et al (7) are limited to demonstration of the superiority of VL with respect to first pass success and esophageal intubation, and these findings are based on evidence of only moderate certainty. Although the data were collected from fewer patients than the ideal number suggested by trial sequential analysis (n = 5861), key findings were consistent across studies with varying levels of bias. This consistency across studies of different quality strengthens the credibility of the findings. As is often the case, the evidence itself documents a level of uncertainty that differs from the impression of many clinicians that VL improves first pass success. Clinicians, especially those who have experience predating both VL and the institution of other practices that have improved airway management, often assert the belief that VL reduces the intubator’s and, where applicable, supervisor’s anxiety during tracheal intubation. Such subjective observations are supported by the meta-analysis’ findings that VL was comparably more successful in the hands of less experienced operators. These findings are similar to those reported by the DEVICE investigators, who found that operator experience with tracheal intubation, and VL specifically, moderated the effect of VL on first pass success (8). The transition from DL to VL has also likely afforded other subjective but tangible improvements, such as facilitating a shared view of the airway, therefore improving communication and likely reducing the cognitive load when both instructing and learning a complex, multistep process. Although large prospective studies suggest that the use of VL outside of the OR is less common than DL, these patterns may reflect limited availability or selection of DL over otherwise available VL based on clinical judgment or familiarity (5,6). In a post hoc analysis, data from the International Observational Study to Understand the Impact and Best Practices of Airway Management In Critically Ill Patients (INTUBE) study suggest that VL was more often employed for patients in whom technically difficult airway management was subjectively anticipated, and first pass success with VL exceeded that of DL with no significant difference in adverse event rates (11). Such results should be interpreted and applied cautiously, however, as only two of the trials identified and analyzed by McDougall et al (7) reported outcomes based on anticipated airway difficulty (8,12). As such, additional data are needed to better inform clinical practice when selecting an approach to tracheal intubation in critically ill adults with predictors of a difficult airway. In the long run, the material question for clinicians is how VL might be used most effectively in critical care settings. Given ongoing, persistent, and sometimes severe workforce shortages, particularly in rural settings, techniques that enable and democratize safer advanced airway management are of paramount importance. While VL may ultimately supplant DL as the standard of care for tracheal intubation, experience suggests that such a change will be preceded by diverse opinions and strong emotional responses. But beyond the inevitable vitriol, certain factors must be considered and adjudicated. Both VL and DL require extensive education and practice, but the curriculum and the procedures are different. Appropriate learning paradigms and an optimal educational approach are already under consideration (13,14). VL has well-recognized shortcomings; these issues are substantial enough to ensure that DL will remain as a fundamental component of airway management in specific circumstances. As such, work to raise awareness of VL’s limitations, technological approaches to address these issues, and the means by which to facilitate training in both approaches will likely need to be further developed (14). Most importantly, VL, like DL, is not a benign procedure, and it carries substantial risk for patient harm. Therefore, a more complete demonstration of the superiority of VL over DL should be sought. That the study by McDougall et al (7) could not identify more substantial differences between VL and DL is proof positive of the need for additional data. Perhaps explorations currently underway in perioperative settings will prove sufficient (13). However, additional investigation, including perhaps RCTs targeting the specific subpopulations where data were insufficient for McDougall et al (7) to draw conclusions, may be necessary. Importantly, in constructing additional studies, we might reappraise our thoughts about what constitutes a meaningful outcome variable. McDougall et al (7) point out that first pass success is a process-oriented outcome and may affect patient-centered or patient-important outcomes only indirectly. The ability to link direct, acute interventions to global outcomes is an increasingly important conundrum, especially in the critically ill. The complexity of care and the multiplicity of interventions make it difficult, if not impossible, to establish cause and effect. For example, that survival of an elderly individual with sepsis and several underlying conditions who spent a month in the ICU was determined by a failure to intubate on the first pass lacks credibility. Perhaps focusing on more tangible, short-term, “process-directed” variables is inevitable. That approach might be especially valuable in the construction and validation of protocols to broadly guide management, programs to evaluate safety and efficacy, and algorithms to improve practice (15). Additionally, process-oriented outcomes are of particular interest to novice practitioners of VL and DL, as well as their instructors. While guidelines for airway management in critically ill adults may not be specifically intended to address all technical aspects of airway management, including tracheal intubation, future iterations should aim to at least broadly address the seemingly beneficial impacts of VL and other techniques on first pass success as a likely mediator of physiologic difficulties during tracheal intubation (16). Finally, it is important to remind ourselves that “newer,” “advanced,” and “technology-based” are not synonyms for “better.” Critical care clinicians are arguably technophilic, and previous technologic innovations that enhanced practice while improving patient safety met with approval and eventual adoption. A comparison of pulse oximetry and ultrasound for vascular access is instructive. Pulse oximetry, which was logically useful, relatively straightforward from educational and implementational standpoints, and conferred a low risk of harm, was adopted quickly. In contrast, the emergence of ultrasound as a standard of care for vascular access required systematic evaluation owing to its costs and potential risks absent appropriate education and training; the process was thus more deliberate. Ultimately, tools are developed to improve the ability to perform tasks, and improvement requires comparison using a metric. The study by McDougall et al (7) suggests that VL has advantages over DL in some regards. But of equal importance is the paucity of data available to compare the two approaches in aspects important to both clinicians and patients. VL is “newer” and relies on more “advanced technology.” Although clinicians may be voting with their hands in reaching for VL more often, a definitive demonstration that VL is “better” than DL requires a more complete comparison.

  PublisherDOI

Recent grants

• NIH Grant T32GM007612

  NIH · $3.3M · 2013

• Orexinergic Modulation of Experimental Sepsis

  NIH · $1.6M · 2017–2023

• NIH Grant R01GM059930

  NIH · $2.5M · 2011

• NIH Grant R21AI070929

  NIH · $429k · 2012

• NIH Grant K08DK002179

  NIH · $465k · 1998

Frequent coauthors

• Richard J. Levy

  Columbia University Irving Medical Center

  482 shared

• Ernest E. Moore

  University of Colorado Denver

  416 shared

• Denis D. Bensard

  Denver Health Medical Center

  416 shared

• Philip F. Stahel

  East Carolina University

  416 shared

• Christopher H. Mody

  University of Calgary

  408 shared

• Berton R. Moed

  Orthopaedic Trauma Association

  404 shared

• Julie P. Chou

  400 shared

• Tom Lim

  392 shared

Education

• MD

  New York Medical College

  1980

• MS, Chemistry

  Northwestern University

  1976

• BS

  Trinity College

  1975