About

George L Anesi, MD, MSCE, FCCM, ATSF, is an Assistant Professor of Medicine in the Division of Pulmonary, Allergy, and Critical Care at the Hospital of the University of Pennsylvania. His clinical expertise includes critical care, sepsis, acute respiratory failure, influenza, COVID-19, respiratory viruses, and emerging/high-consequence pathogens. He also has expertise in hospital operations, preparedness and response, hospital capacity strain, global health, and clinical epidemiology. His research focuses on hospital preparedness and the evaluation of critical care and acute care resources during times of system strain, including situations of dynamic strain such as demand variation, seasonal trends, epidemics, and disasters, as well as fixed strain in resource-limited settings domestically and globally.

Research topics

• Medicine
• Internal medicine
• Virology
• Gastroenterology
• Emergency medicine
• Cardiology
• Demography

Selected publications

• Society of Critical Care Medicine Guidelines for the Allocation of Critical Care Resources to Adults During Crisis-Level Shortages

  Critical Care Medicine · 2026-03-01 · 2 citations

  article

  RATIONALE: Efficient distribution of scarce critical care resources is essential to save the most lives in times of crisis. Evidence-based practices and processes enhance clinical decision-making. OBJECTIVES: The objective of these guidelines was to develop evidence-based, rather than expert-based, recommendations for triaging critically ill patients eligible for ICU admission during times of crisis-level shortages in ICU capacity. DESIGN: The American College of Critical Care Medicine Board convened a 21-member multidisciplinary panel, comprising doctors in medicine, nursing, and law; advanced practice providers; respiratory therapists; ethicists; and patient/family representatives. The panel included two expert methodologists specialized in developing evidence-based recommendations in alignment with the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) methodology. Conflict-of-interest policies were strictly followed during all phases of guidelines development including task force selection and voting. METHODS: The panel members identified and formulated five fundamental Patient, Intervention, Comparator, and Outcomes questions. The panel conducted a systematic review for each question to identify the best available evidence, analyzed the evidence, and assessed the certainty of the evidence using the GRADE methodology. The GRADE evidence-to-decision framework was used to formulate the recommendations. Good practice statements were included to provide additional guidance. RESULTS: The panel generated one conditional recommendation and five no recommendation statements. CONCLUSIONS: Crisis-level shortages significantly disrupt patient care. Despite the role of triage in minimizing adverse outcomes, there is a lack of evidence, as opposed to expert opinion, to guide practice recommendations in the critical clinical scenarios considered by the panel.

  PublisherDOI

• SARS-CoV-2 in Pregnancy—More Hints of Rare Complications

  JAMA Network Open · 2026-05-07

  articleOpen access1st authorCorresponding
  PublisherDOI

• Executive Summary: Society of Critical Care Medicine Guidelines for the Allocation of Critical Care Resources to Adults During Crisis-Level Shortages

  Critical Care Medicine · 2026-03-01 · 1 citations

  article

  Triage is the complex process of prioritizing the allocation of critical resources, and it is essential during situations involving crisis-level shortages in the ICU (1). Crisis-level shortages refer to severe deficits of essential resources and services including products (e.g., blood, ventilators), personnel (e.g., respiratory therapists, nurses), and facilities (e.g., ICU beds) that drastically affect the delivery of care to critically ill patients. ICU capacity is the maximum number of patients that an ICU can accommodate while maintaining a high level of specialized care and monitoring. The capacity of an ICU depends on multiple factors, such as physical space, availability of medical equipment, staffing levels, and the expertise of healthcare professionals. Prioritizing patients also requires managing these resources (e.g., patient-flow and other system inefficiencies) through a process called flow-sizing, or matching capacity and demand to ensure that all patients receive appropriate care (2). The Board of the American College of Critical Care Medicine convened a multidisciplinary panel to develop focused, evidence-based recommendations for triaging critically ill patients eligible for ICU admission during times of crisis-level shortages. The panel conducted a systematic review of the published scientific literature, focusing on patient-oriented, clinically relevant outcomes to answer Patient, Intervention, Comparator, and Outcomes (PICO) questions regarding the triage of critically ill adults in the ICU. The clinical practice recommendations were developed according to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) process (3). These clinical practice guidelines reflect the state of knowledge at the time of publication. The full guidelines may be accessed via (4). RECOMMENDATIONS The panel made no recommendation statements for four of the five PICO questions addressed in these guidelines due to insufficient evidence (Table 1). Future research is needed on several topic areas covered by these guidelines. A summary of the types of palliative care services provided in the studies included is presented in Table 2. A summary of research priorities for each topic is presented in Table 3 of the full guidelines (4). The panel made one conditional recommendation as presented below. A “Conditional” recommendation reflects a lower degree of certainty in the appropriateness of the patient care strategy for all patients. It requires that the clinician use clinical knowledge and expertise and strongly considers the individual patient’s values and preferences to determine the best course of action. The ultimate judgment regarding any specific care must be made by the treating clinician and the patient, taking into consideration the individual circumstances of the patient, available treatment options, and resources. TABLE 1. - ICU Triage Patient, Intervention, Comparator, and Outcomes Questions Question 1. In situations of limited ICU bed availability for critically ill adult patients, should clinician judgment for prioritization of admission be used over time-based admission to ICU? Population Intervention Comparator Outcomes Adult critically ill patients (age ≥ 18 yr old defined by an ICU admission request or order as per local hospital processes Clinician judgment (defined as use of physician assessment, patient factor based, or severity of illness) system or strategy for prioritization of admissions to the ICU Time-based admission (first-come, first-served) to ICU (patients that “meet ICU criteria” are admitted in the order of arrival or referral to ICU) See Supplemental Material and Table S2 in (4) Setting: Limited ICU bed availability as defined by included studies Subgroups of interest 1) Normal operational surge conditions (> 90% of ICU beds are occupied) 2) Short-lived disasters (hurricanes, etc) 3) Longer-term disasters (pandemics, etc) Question 2. In situations of limited ICU bed availability for critically ill adult patients, should a formal triage tool be used vs. no triage tool? Population Intervention Comparator Outcomes Adult critically ill patients (age ≥ 18 yr old defined by an ICU admission request or order as per local hospital processes Objective triage tool (e.g., Sequential Organ Failure Assessment) No objective triage tool See Supplemental Material and Table S2 in (4) Setting: Limited ICU bed availability as defined by included studies Intervention subgroups of interest (subject to data availability): Subgroups of interest 1) Triage system using machine learning/artificial intelligence 1) Normal operational surge conditions (> 90% of ICU beds are occupied) 2) Diagnosis-based 2) Short-lived disasters (hurricanes, etc) 3) Other specific triage tools 3) Longer-term disasters (pandemics, etc) Question 3A. In situations of limited ICU bed availability for critically ill adult patients, should patients awaiting an ICU bed be transferred to another facility? Question 3B. For patients not transferred to another facility, should hospitals with ICUs have a designated non-ICU area that is prepared to board ICU patients during surge conditions vs. no designated non-ICU area? Population Intervention Comparator Outcomes Adult critically ill patients (age ≥ 18 yr old defined by an ICU admission request or order as per local hospital processes) Interfacility transfer of patients No interfacility transfer See Supplemental Material and Table S2 in (4) Setting: Limited ICU bed availability as defined by included studies Subgroups of interest 1) Normal operational surge conditions 2) Short-lived disasters (hurricanes, etc) 3) Longer-term disasters (pandemics, etc) Hospitals with ICUs Having a designated non-ICU area that is prepared in advance of surges to board ICU patients during surge conditions Standard care (no designated non-ICU area) See Supplemental Material and Table S2 in (4) Setting: Limited ICU bed availability as defined by included studies Potential subgroups: Subgroups of interest 1) ED (e.g., ED-ICU) 1) Normal operational surge conditions 2) Designated overflow areas (e.g., post-anesthesia care unit, operating room, intermediate care unit, etc) 2) Short-lived disasters (hurricanes, etc) 3) Specialty ICUs where admitting diagnosis is discordant to the specialty (e.g., chronic obstructive pulmonary disease exacerbation admitted to surgical ICU), if available. 3) Longer-term disasters (pandemics, etc) Question 4. In situations of limited ICU bed availability for critically ill adult patients, should patients boarded in non-ICU care areas be managed by ICU trained practitioners or by the usual practitioners for that area (i.e., emergency medicine physician, anesthesiologist, hospitalist)? Population Intervention Comparator Outcomes Adult critically ill patients (age ≥ 18 yr old defined by an ICU admission request or order as per local hospital processes) Managed by usual ICU practitioners Managed by non-ICU practitioners (e.g., ED, anesthesiologist, hospitalists, etc) See Supplemental Material and Table S2 in (4) Setting: Patients boarded outside ICU Non-ICU practitioners supervised by ICU practitioners Question 5. During situations of limited ICU bed availability, in critically ill patients at high risk of dying in the ICU, should palliative care services be involved early vs. at the usual clinician discretion? Population Intervention Comparator Outcomes Critically ill patients at high risk of dying in the ICU during situations of limited ICU capacity Early involvement of palliative care service Routine or no involvement of palliative care service See Supplemental Material and Table S2 in (4) Setting: Limited ICU bed availability as defined by included studies ED = emergency department. TABLE 2. - Summary of Palliative Care Services Provided in the Studies Study PC Involvement Ahrens et al (5), 2003; Mosenthal et al (6), 2008 Structured communication with families of seriously ill patients by communication team Campbell et al (7), 2003; Campbell et al (8), 2004 Identification of patient’s advance directives or preferences about end-of-life care. Assistance with discussion of the prognosis and treatment options. Implementation of PC strategies when treatment goals change to a focus on comfort measures Curtis et al (9), 2008; Curtis et al (10), 2011 Clinician education, local champions, academic detailing, feedback to clinicians, and system support. Not targeted to patients or family members Daly et al (11), 2010; Lilly et al (12), 2003 Intensive communication system for family decision-makers of long-stay ICU patients. Each meeting addressed medical update, values and preferences of the patient, goals of care, treatment plan, and milestones Hsu-Kim et al (13), 2014 PC consultation with patient Norton et al (14), 2015 Basic PC consultation consisting of: 1) chart review, 2) history of present illness, 3) discussions with medical ICU team, 4) review of PC recommendations with attending physician, and 5) completion of assessment form with PC recommendations. Complete PC consultation consisting of: 1) basic PC consultation, 2) regular involvement by the PC team with patient’s family members, 3) full involvement of PC physician, 4) regular involvement of PC team in the patient’s treatment, and 5) availability of PC team for additional support for the patient and family as needed Ma et al (15), 2019 PC consultation comprising of regular visits by PC team. PC consultation included: 1) chart review, 2) meeting with patient, 3) identifying physical and emotional needs of patient and family, 4) PC plan, 5) communication with all parties regarding goals, values, and treatment decisions, and 6) follow-up with patient until discharge PC = palliative care. TABLE 3. - ICU Triage Research Priorities Topic Research Priorities Clinician judgment vs. time-based admission Explore mathematical models’ applications in large databases (e.g., operational modeling and queueing theory) Study the use of different types of artificial intelligence in real-time decision-making across diverse clinical contexts (e.g., crisis-level resource shortages) Use of objective triage tool Evaluate the impact of the use of objective triage tools in resource-limited situations on overall (system and population level) rates of survival, as well as factors such as consistency of decision-making across providers, equity, and fairness Boarding of patients when no ICU space available High-quality studies regarding the transfer of ICU patients to another facility in the setting of limited ICU bed capacity Effects on healthcare systems, patient-centered outcomes, and cost efficiency Broader availability and access to data from existing hospital systems or insurance companies for research Designated non-ICU areas during surge conditions High-quality observational studies of current practices. Randomized studies would be preferred but are unlikely Evaluation of timing of care delivery and quality outcomes, including to adherence to guidelines as compared with traditional settings Evaluation of cost and satisfaction (patient and staff) Who should care for ICU patients in non-ICU areas? High-quality outcome studies of patients in non-ICU areas when cared for by ICU providers (all team members) vs. non-ICU team members For physicians, a comparison of board certification in critical care vs. not. For other providers/team members like advanced practice providers, registered nurses, etc. An exploration of core competencies Involvement of palliative care services Studies better defining resource requirements for the intervention and evaluating cost-effectiveness Randomized trials, cost-effectiveness analyses, and strategies to integrate palliative care seamlessly into ICU workflows to determine the best ways to support critically ill patients and their families during times of limited capacity In critical care patients at high risk of dying, we suggest involvement of palliative care services be involved early as compared with no palliative care services or usual care (conditional recommendation, very low certainty evidence). Remarks: Defining and aligning goals of care with patients’ wishes and values is important across all patients admitted to ICU who are at risk of dying, and particularly relevant in conditions of limited ICU and hospital capacity. Early palliative care involvement during a patient’s ICU stay has been proposed to address patient and family goals of care during times of critical illness, yet evidence on its clinical and systemic impacts remains inconclusive. A comprehensive literature search was completed, and 163 studies were screened that pertained to early palliative care services. Out of these articles, 11 studies (5–15) were included in consultation with the panel. The GRADE evidence synthesis highlights very low certainty across all evaluated outcomes. Mortality outcomes showed no significant difference (risk ratio, 1.02; 95% CI, 0.77–1.35; very low certainty), while measures like hospital length of stay (mean difference [MD], 2.24 d lower; 95% CI, 4.11 lower to 0.37 lower; very low certainty) and ICU length of stay (1.9 d lower; 95% CI, 2.32 lower to 1.48 lower; very low certainty) suggest potential but limited benefits. Family satisfaction did not demonstrate significant clinical improvement with early palliative care involvement (MD, 0.85 higher; 95% CI, 1.99 lower to 3.7 higher; very low certainty). Additionally, there was no difference in the quality of death and dying (MD, 0.21 higher; 95% CI, 3.51 lower to 3.94 higher; very low certainty). We make a conditional recommendation in favor of early provision of palliative care services. While evidence from included studies was inconsistent and retrospective in design, aligning a patient’s care with their values and preferences is vital to a critical illness-related hospitalization. The desirable effects were weighed against the absence of observed harm and the overarching priority of ensuring dignified care at the end of life. The absence of robust cost-effectiveness data further underscores the need for resource-specific research. Cultural acceptability and equity concerns were also noted, with disparities in access and skepticism among marginalized groups requiring targeted institutional approaches to mitigate biases. The recommendation reflects the panel’s view that defining and aligning goals of care with patient and family values, as early as possible, remains critical, even without high-quality evidence. Future research should focus on randomized trials, cost-effectiveness analyses, and strategies to integrate palliative care seamlessly into ICU workflows to determine the best ways to support critically ill patients and their families during times of limited capacity.

  PublisherDOI

• The Saga of Equitable Oxygen Saturation Measurement Continues: The Role of Arterial Blood Gases

  American Journal of Respiratory and Critical Care Medicine · 2025-03-25

  letterOpen accessSenior author
  PublisherOA PDFDOI

• The Aftermath of Acute Surge Events: Quantifying the “Bystander Effect” During the COVID-19 Pandemic

  Critical Care Medicine · 2025-08-06

  article1st authorCorresponding
  PublisherDOI

• Proteomic Analysis Identifies Novel Immune Mediators and Soluble Receptors That Are Robustly Associated With Mortality in COVID-19 Critical Illness

  American Journal of Respiratory and Critical Care Medicine · 2025-05-01

  article

  Abstract Rationale: Determining circulating proteins robustly associated with mortality is an essential step in identifying novel therapeutic targets to test in critically ill COVID-19 patients. Methods: We identified and validated circulating protein levels associated with mortality in critically ill COVID-19 patients using an aptamer-based proteomics platform (Somalogic). We analyzed two prospective cohorts: (1) single-site COVID-19 Host Response and Clinical Outcomes (CHROme) (discovery cohort, N=198) and (2) multi-site Severe Acute Respiratory Infection – Preparedness (SARI-PREP) (validation cohort, N=100). Associations between 4,718 plasma protein levels collected within 48 hours of ICU admission and 14-day mortality were assessed using Cox proportional hazards models, adjusted for age, sex, and BMI (FDR <0.01). We performed over-representation analysis to annotate top protein associations. We cross-referenced our validated protein set with single-nucleotide polymorphisms (SNPs) in cis-associated genes associated with mortality in the COVID-19 Host Genetics Initiative (HGI) GWAS (PMID: 34237774). To further assess whether the validated proteins might be causal intermediates, we analyzed pooled genotype and proteomic data from CHROme and SARI-PREP to determine if identified SNPs were cis-protein quantitative trait loci (pQTLs). Results: In the discovery cohort, 1,529 proteins were associated with increased risk, while 71 proteins were associated with reduced risk of 14-day mortality. Validation confirmed that 520 of these proteins were associated with increased risk, and three—BCHE, SERPINC1, and F2—were associated with reduced risk. The top 100 proteins included pro-inflammatory biomarkers (ADM, CCL19, CD46), circulating immune receptors (CD300C, TNFRSF1B), interferon-associated biomarkers (IFNAR1, INHBA/INHBB, IFNGR1), and novel biomarkers related to neuronal development (NLGN2, NLGN1, NLGN4X). Over-representation analysis (reference Hallmark Pathways) revealed enrichment in immune pathways (inflammatory response, IL6-STAT2 signaling and TNF-α via NF-κB signaling) and epithelial-mesenchymal transition. Five proteins in the validated set have a cis-SNP that was significantly associated with mortality in the COVID-19 HGI GWAS: CD300C (rs4351086), EPHB4 (rs314296), EPHA7 (rs76462658), PDGFRA (rs1553659), and B3GALT1 (rs78286703). Exploratory causal inference analyses in genotyped subjects from both cohorts (n = 249), did not identify cis pQTLs between these SNPs and protein levels or associations with mortality that met our FDR threshold. However, rs4351086 (CD300C – co-inhibitory receptor on antigen-presenting cells) and rs314296 (EPHB4 –receptor that plays a key role in vascular development) have been previously associated with gene expression (eQTLgen.org). Conclusions: Our analysis identified and validated previously unreported immune mediators and soluble receptors that are associated with mortality in severe COVID-19. Further investigation is required to determine whether these proteins are causal intermediates.

  PublisherDOI

• 322: OPERATIONALIZING THE NEW GLOBAL DEFINITION OF ARDS: AN EPIDEMIOLOGICAL REPORT FROM SOUTH AFRICA

  Critical Care Medicine · 2025-01-01

  article1st authorCorresponding
  PublisherDOI

• 317: AN EXPLANATORY MIXED-METHODS STUDY OF “ICU NET BENEFIT” FOR SEPSIS AND ACUTE RESPIRATORY FAILURE

  Critical Care Medicine · 2025-01-01

  article1st authorCorresponding
  PublisherDOI

• An Explanatory Mixed-Methods Study of Intensive Care Unit Net Benefit: Triage and Trajectory for Sepsis and Acute Respiratory Failure

  Annals of the American Thoracic Society · 2025-01-07 · 2 citations

  articleOpen access1st authorCorresponding

  Abstract Rationale Patients with sepsis and/or acute respiratory failure are at high risk for death or long hospital stays, yet limited evidence exists to guide triage to intensive care units (ICUs) or general medical wards for the majority of these patients who do not initially require life support. Objectives To identify factors that influence how hospitals triage patients with capacity-sensitive conditions and those factors that may account for observed ICU relative to ward, or ward relative to ICU, benefits for such patients. Methods We conducted an explanatory sequential mixed-methods study. As part of a 27-hospital, two–health system retrospective cohort study, we calculated hospital-specific measurements of ICU net benefit for patients with sepsis and/or acute respiratory failure. Hospitals among the highest ICU net benefit and lowest ICU net benefit (or highest ward net benefit) from each study health system were selected for in-depth qualitative study. At each hospital, interviews were conducted with emergency department, ward, and ICU clinicians and administrators. Interview transcripts were analyzed using flexible coding and the framework method. Results Interviews were conducted with 118 respondents (46 physicians, 43 nurses, 5 advanced practice providers, and 24 administrators) from four hospitals. Respondents across hospitals agreed that the prediction of patient trajectory is central to triage decisions, but there was variation in opinion across work locations about optimal pretriage emergency department interventions in terms of intensity, repetition, clinical reassessment, and observation duration. The main difference observed between high and low ICU net benefit hospitals related to the way respondents working in the ICU and ward described their responses to patients who experience rapid clinical deviations from triage-expected trajectories, including sustained lack of critical care needs after admission to the ICU and acute critical care needs after admission to the ward. Hospitals with low ICU net benefit (or high ward net benefit) had particularly robust and proactive rapid response and clinical decompensation surveillance practices for ward-admitted patients. Conclusions Particularly proactive rapid response programs that deliver on-location critical care may quantitatively increase ward net benefit by bringing ICU benefits without ICU-associated harms to ward patients who become critically ill.

  PublisherOA PDFDOI

• Conserved Plasma Proteomics Signatures of Early Respiratory Outcomes Across COVID-19 and Non-COVID-19 Acute Respiratory Failure

  American Journal of Respiratory and Critical Care Medicine · 2025-05-01

  article

  Abstract Rationale: There is a pressing need to identify noninvasive biomarkers that delineate early outcomes in acute respiratory failure (ARF) and guide care. Methods: Our objective was to discover and validate novel plasma biomarkers for ARF outcomes, using a high-throughput, aptamer-based proteomics platform (SomaLogic). We analyzed patients on supplemental oxygen from three multicenter cohorts: 1) Derivation – COVID-19 patients from the Severe Acute Respiratory Infection-Preparedness (SARI-PREP) consortium (N=371); 2) Validation – COVID-19 patients from a publicly available Netherlands trial dataset of imatinib (N=277); 3) Validation – non-COVID-19 patients from SARI-PREP (N=98) spanning ARF of diverse infectious and noninfectious causes. Our primary outcome was new or persistent respiratory failure, defined by receipt of high-flow oxygen or mechanical ventilation, or death, at one week after cohort enrollment. We examined differential abundance of 4718 proteins in plasma collected at enrollment (within 48h of ARF or ICU admission). We contextualized findings by evaluating a) pathways (KEGG) enriched in poor respiratory outcomes; b) added value of proteins to clinical variables in prediction; and c) imatinib treatment effect heterogeneity by validated proteins. Results: The cohorts exhibited marked differences in ARF severity and outcomes, with 63% of the COVID-19 derivation cohort experiencing the outcome compared to 15% of the COVID-19 validation cohort (Figure 1A). Controlling for age, sex, and baseline respiratory severity, we identified several differentially abundant proteins in all cohorts, with greatest overlap between COVID-19 derivation and non-COVID-19 validation (1B). Fifty-seven proteins validated across all three cohorts (1C). Findings confirmed previously described immune response proteins (IL6, TNF superfamily, S100A), and identified new biomarkers including structural (keratins, COL6A2, CLTA) and RNA-binding (TAF15, SRSF6, RBM17) proteins. Extracellular matrix-receptor interaction, cytosolic DNA-sensing, and many immune response pathways were consistently enriched in patients with poor outcomes. Applying LASSO to all 4718 proteins, we fit a logistic regression model with 51 proteins and clinical variables (age, sex, obesity, baseline respiratory severity) in the derivation cohort. This model outperformed clinical variables alone in COVID-19 (AUROC 0.70, 95%CI 0.64-0.77 vs. 0.58, 95%CI 0.52-0.64) and non-COVID-19 (0.76, 95%CI 0.67, 0.85 vs. 0.61, 95%CI 0.50-0.71) validation cohorts. In an exploratory causal forests model, two proteins were associated with treatment effect heterogeneity (IL6, CHP1—known regulators of imatinib response pathways). Conclusions: We validated 57 proteins distinguishing respiratory outcomes across cohorts varying in ARF severity and cause. Protein signatures aligned by ARF severity rather than COVID-19 status. Further study of proteins may yield strategies to identify high-risk patients for treatment.

  PublisherDOI

Recent grants

• Hospital adaptation and resiliency for infected and uninfected patients during respiratory viral surge events: from seasonal influenza to COVID-19

  NIH · $837k · 2021–2026

Frequent coauthors

• Scott D. Halpern

  University of Pennsylvania

  60 shared

• Robert Wise

  University of Oxford

  40 shared

• Rachel Kohn

  University of Pennsylvania

  39 shared

• Gary E. Weissman

  California University of Pennsylvania

  35 shared

• Michelle Smith

  Grey's Hospital

  26 shared

• M. Kit Delgado

  University of Pennsylvania

  24 shared

• Stella Savarimuthu

  University of Rochester Medical Center

  21 shared

• Arisha Ramkillawan

  University of Rochester Medical Center

  21 shared

Awards & honors

• Senior Fellow, Leonard Davis Institute of Health Economics (…
• Faculty Fellow, Center for Clinical Epidemiology and Biostat…
• Senior Fellow, Center for Public Health Initiatives (CPHI)