Systems Anesthesiology: Integrating Insights From Diverse Disciplines to Improve Perioperative Care : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Featured Articles: The Open Mind

Systems Anesthesiology: Integrating Insights From Diverse Disciplines to Improve Perioperative Care

Ruscic, Katarina Jennifer MD; Hanidziar, Dusan MD; Shaw, Kendrick Matthew MD; Wiener-Kronish, Jeanine MD; Shelton, Kenneth Tierney MD

Author Information
doi: 10.1213/ANE.0000000000006166
  • Free

See Article, page 671

The twenty-first century saw the rise of “systems biology,” an interdisciplinary approach in biomedical research integrating computational modeling, cell biology, proteomics, and genomics to provide an understanding of the larger picture of a tissue or organism.1 This Open Mind article focuses on rapidly evolving areas of critical care anesthesiology, including the use of novel molecular markers for disease diagnosis and phenotyping, and the use of predictive models for perioperative risk assessment. Taken together, these emerging methods define a new frontier of "systems anesthesiology," an approach to taking care of patients that incorporates information from the diverse spectrum of molecular, cellular, and computational approaches (Figure 1). We also discuss the rise of telemedicine and remote monitoring, which the systems anesthesiologist will use in the present and emerging future to extend their reach and benefit more patients with their unique multidisciplinary expertise (Figure 2).

MOLECULAR INVESTIGATIONS TO ADVANCE PERSONALIZED CARE IN THE ICU

The promise of more personalized care in the intensive care unit (ICU) relies on identifying clinically relevant patient characteristics (eg, genetic, physiologic, and immunologic) and subsequent delivery of tailored interventions that would result in superior outcomes. An example of such tailored intervention would be titration of positive end-expiratory pressure (PEEP) based on transpulmonary pressure measurements in obese patients with acute respiratory distress syndrome (ARDS).2 However, not all the important characteristics are obvious at the patient bedside or with routine clinical measurements. Novel cellular or molecular markers are needed to identify disease processes in early stages (eg, sepsis and acute kidney injury [AKI]), to identify the highest-risk patients who could benefit from advanced therapies (eg, immunologic or metabolic modulation), or to improve inclusion in clinical trials. Laboratory investigations producing high-dimensional data sets (eg, transcriptomics, proteomics, metabolomics, lipidomics, and microbiomics) have now become an integral part of many clinical trials (ie, mechanistic clinical trials)—a growing effort supported by the National Institutes of Health to increase mechanistic understanding of human disease and differential responses of patients to therapy.3,4

F1
Figure 1.:
Personalized high-dimensional data, novel biomarkers, and physiological and imaging data will inform predictive models and will serve as key tools for the future systems anesthesiologist.
F2
Figure 2.:
The systems anesthesiologist will use tele-medicine to extend their reach on a global scale.

Sepsis and ARDS are complex syndromes commonly encountered in ICUs and are associated with substantial mortality and long-term morbidity. The pathophysiology of both conditions involves dysregulated immune cell and platelet activation, endothelial dysfunction, coagulopathy, metabolic dysregulation, and microbiome alterations, but differences exist among individuals as to the specific and dominant alterations, organ involvement, and response to treatments.5,6 Analysis of publicly available blood transcriptomic data previously revealed “inflammopathic,” “adaptive,” and “coagulopathic” phenotypes in patients with sepsis, with the adaptive phenotype (upregulation of adaptive immunity genes) being associated with superior survival.7 Whole blood transcriptomics has also led to the identification of steroid-responsive subgroups in patients with sepsis, opening the avenue for targeted corticosteroid administration.8 Single-cell ribonucleic acid (RNA) sequencing of blood mononuclear cells recently led to the discovery of a novel circulating monocyte phenotype (“MS1”) that is enriched in patients with sepsis, potentially representing a sepsis biomarker.9 The host response in sepsis is also associated with changes in the plasma metabolome and lipidome—important phospholipids such as lysophosphatidylcholine (LPC) that are depleted in sepsis were found to be increased in early responders to treatment and implicated in the regulation of inflammatory responses and as potential therapies.10

Similar to sepsis, ARDS is a syndrome where unique molecular features and disease subphenotypes are increasingly identified.6 Transcriptomic analysis of leukocytes identified distinct proinflammatory gene signatures in monocytes, natural killer (NK) cells, and neutrophils in patients with ARDS,11,12 indicating that specific and aberrant immune pathways are associated with the syndrome. Defects in T-cell immunity are a common feature in patients with severe coronavirus disease 2019 (COVID-19).13 The analysis of plasma protein biomarkers from patients with ARDS identified “hyperinflammatory” and “hypoinflammatory” subphenotypes with distinct outcomes (higher mortality in the hyperinflammatory phenotype), and distinct responses to PEEP and fluid therapy.6 These findings increased our understanding of ARDS pathogenesis and heterogeneity; however, there has been a lack of therapeutic translation thus far. In the future, the most successful strategies for translation may entail combining granular physiologic information with high-dimensional molecular measurements.14

PREDICTIVE MODELS FOR PERIOPERATIVE AND CRITICAL CARE

Predictive models will play an increasing role in the care we provide to patients in the perioperative period and in the ICU. Much like how the modern meteorologist has become an expert in interpreting and synthesizing sometimes contradictory predictions from different models, the future systems anesthesiologist will interpret and synthesize models to guide patient care, both in the operating room and in the ICU. Models are increasing in power and complexity as statistical and machine learning (ML) tools progress, but much work remains to integrate them into patient care. Complex deep learning models have better accuracy than traditional statistical models in, for example, predicting preventable acute care use and medical spending.15 However, difficulty in understanding how such models arrive at their predictions limits translation to a clinical action plan.16 Many newer ML models include interpretability as part of their design, such as models providing explainable causes of intraoperative hypoxemia17 or identifying keywords in patient clinical notes that influence the calculated risk of readmission.18 Many predictive models significantly increase in accuracy with increased data,19 thus complex full-waveform data streams, such as intraoperative electroencephalogram (EEG) signals, heart rate, blood pressure, and end-tidal Co2, have been increasingly used as model inputs.20–22 Real-time data streams also make possible closed-loop control systems to assist with patient care, such as titration of sedation or vasopressors.23 Because of the challenge of complex artificial intelligence (AI) models to perform well across different institutions and time periods, traditional statistical models still have a role and can sometimes outperform deep learning models.24 Human performance has not yet been exceeded by AI, but AI’s “ability to quickly and accurately sift through large stores of data and uncover correlations and patterns that are imperceptible to human cognition25” will make predictive models a key tool for future systems anesthesiologists (Figure 1).

Predictive models of patient perioperative risk do not yet include cognition, despite delirium and perioperative neurocognitive disorders being one of the most common complications of anesthesia and surgery, particularly in patients 65 years and older.26,27 Patients with delirium have an increased length of stay, morbidity, mortality, and costs.28 Predictions of the development of delirium, muscle wasting, and the duration of mechanical ventilation may all be improved using deep learning algorithms. Experimental data suggest an inflammatory response after surgery, and anesthesia affects immune activation and cognitive processes.29–31 As the geriatric population increases, the future requires the ability to recognize patients at risk of delirium and cognitive problems, with genetic and biomarker assessment, formal cognitive testing that is repeated, and multidisciplinary methods of prevention and treatment. Delirium screening, both with validated tools like the confusion assessment method (CAM) and with automated predictive models, should also be done preoperatively and then later repeated. Finally, as our advanced tools continue to evolve, simple nonpharmacologic interventions (mobilization, orientation, physiotherapy, and family presence) also need to be utilized more universally, including in ICUs.

EXTENDING THE REACH OF THE SYSTEMS ANESTHESIOLOGIST: TELE-ICU AND REMOTE MONITORING

Telehealth has recently skyrocketed from an underused innovation to a commonplace reality, as in-person physician time decreased at the start of the COVID-19 pandemic and many outpatient health visits became fully digital. This model of care is likely to endure, with some medical schools implementing formal telemedicine training for students32 and the US Department of Health and Human Services awarding ~$55 Million in 2022 to increase access to virtual health care in underserved populations.33 While there has been visible recent movement of ambulatory care to the virtual setting, multiple recent pandemics, including H1N1 influenza, Middle East Respiratory Syndrome, severe acute respiratory syndrome, and COVID-19, have increased the use of telemedicine in critical care (tele-ICU). In a North Carolina health system, the utilization of tele-ICU strategies during the start of the COVID-19 pandemic increased surge ICU capacity, decreased consumption of personal protective equipment, and increased workforce efficiency while maintaining a high quality of care standards.34 During the time of the coronavirus 2020 surge in New York, a health system in the Bronx created a new 12-bed ICU staffed by residents, a fellow, and a tele-ICU attending. This employment of virtual resources facilitated rapid implementation of expanded care during a public health crisis.35 Examples of successful application of tele-ICU during recent health crises beg the question of whether the use of tele-ICU should be expanded beyond pandemics. In the future, expanded use of tele-ICU can extend the reach of expert systems anesthesiologists to help deliver more acute care to local community hospitals and distant global locations (Figure 2).

While patients are highly monitored in the ICU, have favorable nursing ratios, and are frequently assessed by intensivists, a large number of inpatients receive care on the general floor wards with a much lower rate of monitoring of vital signs and laboratory values, fewer nursing visits, and less frequent chart and in-person assessments by physicians. There is increasing interest in developing early warning systems to recognize deteriorating patients in general care wards who are becoming critically ill and need a higher level of care.36 An ML algorithm developed using a retrospective cohort in a UK hospital system could detect 42% of cardiac arrests or unplanned ICU admissions up to 48 hours in advance,37 while a remote surveillance program monitoring vital sign and laboratory abnormalities in otolaryngology and ophthalmology general care ward patients generated actionable alarms at a rate unlikely to cause alarm fatigue.38 Another group used an automated warning system for early detection of severe postpartum hemorrhage in an obstetric population.39 While emerging studies will likely continue to show the benefits of remote monitoring systems, systems anesthesiologists need to position themselves as part of collaborative multidisciplinary teams receiving notifications of patients who are becoming critically ill outside of the ICU setting. In this way, the unique skillset of the anesthesia critical care doctor can be used in a targeted and efficient fashion to help teams who normally take care of patients in nonacute wards. With timely automated notification of patient decompensation, intensivists can facilitate the execution of critical interventions as soon as deterioration is recognized, even if the patient is still outside of the ICU.

CONCLUSIONS

While the current role of the anesthesia intensivist continues to evolve within the field of critical care, the future will allow physicians to grow outside of our siloed critical care units, outside of our hospitals, and outside of our own institutions.40 Our ability to understand mechanisms of pathology with personalized molecular investigations, the ability to use real-time data to predict perioperative risk, and our ability to reach patients and providers on a larger scale will transform the field of medicine. The systems anesthesiologist as the meteorologist of medicine will use available data to provide a glimpse into the future for each individual patient. This advanced knowledge will allow us to deliver complex care to far-reaching patients and change lives and the world.

DISCLOSURES

Name: Katarina Jennifer Ruscic, MD.

Contribution: This author contributed to the design and writing of the manuscript.

Name: Dusan Hanidziar, MD.

Contribution: This author contributed to the design and writing of the manuscript.

Name: Kendrick Matthew Shaw, MD.

Contribution: This author contributed to the design and writing of the manuscript.

Name: Jeanine Wiener-Kronish, MD.

Contribution: This author contributed to the design and writing of the manuscript.

Name: Kenneth Tierney Shelton, MD.

Contribution: This author contributed to the design and writing of the manuscript.

This manuscript was handled by: Jean-Francois Pittet, MD.

    REFERENCES

    1. Wanjek C. Systems biology as defined by NIH—an intellectual resource for integrative biology. The NIH Catalyst 2011. Accessed May 28, 2022. https://irp.nih.gov/catalyst/v19i6/systems-biology-as-defined-by-nih.
    2. Florio G, Ferrari M, Bittner EA, et al.; investigators of the lung rescue team. A lung rescue team improves survival in obesity with acute respiratory distress syndrome. Crit Care. 2020;24:4.
    3. Shrager JB, Wang Y, Lee M, et al. Rationale and design of a mechanistic clinical trial of JAK inhibition to prevent ventilator-induced diaphragm dysfunction. Respir Med. 2021;189:106620.
    4. Brittain EL, Niswender K, Agrawal V, et al. Mechanistic phase II clinical trial of metformin in pulmonary arterial hypertension. J Am Heart Assoc. 2020;9:e018349.
    5. Schuurman AR, Reijnders TDY, Kullberg RFJ, Butler JM, van der Poll T, Wiersinga WJ. Sepsis: deriving biological meaning and clinical applications from high-dimensional data. Intensive Care Med Exp. 2021;9:27.
    6. Alipanah N, Calfee CS. Phenotyping in acute respiratory distress syndrome: State of the art and clinical implications. Curr Opin Crit Care. 2022;28:1–8.
    7. Sweeney TE, Azad TD, Donato M, et al. Unsupervised analysis of transcriptomics in bacterial sepsis across multiple datasets reveals three robust clusters. Crit Care Med. 2018;46:915–925.
    8. Cohen J, Blumenthal A, Cuellar-Partida G, et al. The relationship between adrenocortical candidate gene expression and clinical response to hydrocortisone in patients with septic shock. Intensive Care Med. 2021;47:974–983.
    9. Reyes M, Filbin MR, Bhattacharyya RP, et al. An immune-cell signature of bacterial sepsis. Nat Med. 2020;26:333–340.
    10. Cambiaghi A, Pinto BB, Brunelli L, et al. Characterization of a metabolomic profile associated with responsiveness to therapy in the acute phase of septic shock. Sci Rep. 2017;7:9748.
    11. Juss JK, House D, Amour A, et al. Acute respiratory distress syndrome neutrophils have a distinct phenotype and are resistant to phosphoinositide 3-kinase inhibition. Am J Respir Crit Care Med. 2016;194:961–973.
    12. Jiang Y, Rosborough BR, Chen J, et al. Single cell RNA sequencing identifies an early monocyte gene signature in acute respiratory distress syndrome. JCI Insight. 2020. 5:e135678.
    13. Popescu I, Snyder ME, Iasella CJ, et al. CD4+ T-Cell dysfunction in severe COVID-19 disease is tumor necrosis factor-α/tumor necrosis factor receptor 1-dependent. Am J Respir Crit Care Med. 2022;205:1403–1418.
    14. Goligher EC, Telias I, Sahetya SK, et al. Physiology is vital to precision medicine in ARDS and sepsis. Am J Respir Crit Care Med. 2022.
    15. Lewis M, Elad G, Beladev M, et al. Comparison of deep learning with traditional models to predict preventable acute care use and spending among heart failure patients. Sci Rep. 2021;11:1164.
    16. Yoon CH, Torrance R, Scheinerman N. Machine learning in medicine: should the pursuit of enhanced interpretability be abandoned? J Med Ethics. 2021.
    17. Lundberg SM, Nair B, Vavilala MS, et al. Explainable machine-learning predictions for the prevention of hypoxaemia during surgery. Nat Biomed Eng. 2018;2:749–760.
    18. Kexin Huang JA, Rajesh R. ClinicalBERT: modeling clinical notes and predicting hospital readmission. 2019. Accessed May 29, 2022. https://arxiv.org/abs/1904.05342.
    19. Halevy A, Norvig P, Pereira F. The unreasonable effectiveness of data. IEEE Intelligent Systems. 2009;24:8–12.
    20. Wingert T, Lee C, Cannesson M. Machine learning, deep learning, and closed loop devices-anesthesia delivery. Anesthesiol Clin. 2021;39:565–581.
    21. Connor CW. Artificial intelligence and machine learning in anesthesiology. Anesthesiology. 2019;131:1346–1359.
    22. Rellum SR, Schuurmans J, van der Ven WH, et al. Machine learning methods for perioperative anesthetic management in cardiac surgery patients: a scoping review. J Thorac Dis. 2021;13:6976–6993.
    23. Desebbe O, Rinehart J, Van der Linden P, et al. Control of postoperative hypotension using a closed-loop system for norepinephrine infusion in patients after cardiac surgery: a randomized trial. Anesth Analg. 2022;134:964–973.
    24. Christodoulou E, Ma J, Collins GS, Steyerberg EW, Verbakel JY, Van Calster B. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models. J Clin Epidemiol. 2019;110:12–22.
    25. Hashimoto DA, Witkowski E, Gao L, Meireles O, Rosman G. Artificial intelligence in anesthesiology: current techniques, clinical applications, and limitations. Anesthesiology. 2020;132:379–394.
    26. Neary WD, Prytherch D, Foy C, Heather BP, Earnshaw JJ. Comparison of different methods of risk stratification in urgent and emergency surgery. Br J Surg. 2007;94:1300–1305.
    27. Kivrak S, Haller G. Scores for preoperative risk evaluation of postoperative mortality. Best Pract Res Clin Anaesthesiol. 2021;35:115–134.
    28. Paixao L, Sun H, Hogan J, et al. ICU delirium burden predicts functional neurologic outcomes. PLoS One. 2021;16:e0259840.
    29. Wang CM, Chen WC, Zhang Y, Lin S, He HF. Update on the mechanism and treatment of sevoflurane-induced postoperative cognitive dysfunction. Front Aging Neurosci. 2021;13:702231.
    30. Subramaniyan S, Terrando N. Neuroinflammation and perioperative neurocognitive disorders. Anesth Analg. 2019;128:781–788.
    31. Qiu LL, Pan W, Luo D, et al. Dysregulation of BDNF/TrkB signaling mediated by NMDAR/Ca2+/calpain might contribute to postoperative cognitive dysfunction in aging mice. J Neuroinflammation. 2020;17:23.
    32. Rahul Sharma M, Thomas H. L. Telemedicine is medicine: training medical students on virtual visits. Innovations in Care Delivery. 2021. Accessed May 29, 2022. https://catalyst.nejm.org/doi/full/10.1056/CAT.21.0325.
    33. Office HP. HHS Awards Nearly $55 Million to Increase Virtual Health Care Access and Quality Through Community Health Centers. 2022. Accessed May 28, 2022. https://www.hhs.gov/about/news/2022/02/14/hhs-awards-nearly-55-million-increase-virtual-health-care-access-quality-through-community-health-centers.html.
    34. Singh J, Green MB, Lindblom S, Reif MS, Thakkar NP, Papali A. Telecritical care clinical and operational strategies in response to COVID-19. Telemed J E Health. 2021;27:261–268.
    35. Krouss M, Allison MG, Rios S, et al. Rapid implementation of telecritical care support during a pandemic: lessons learned during the coronavirus disease 2020 surge in New York City. Crit Care Explor. 2020;2:e0271.
    36. Colquhoun DA, Davis RP, Tremper TT, et al. Design of a novel multifunction decision support/alerting system for in-patient acute care, ICU and floor (AlertWatch AC). BMC Anesthesiol. 2021;21:196.
    37. Pimentel MAF, Redfern OC, Malycha J, et al. Detecting deteriorating patients in the hospital: development and validation of a novel scoring system. Am J Respir Crit Care Med. 2021;204:44–52.
    38. Safavi KC, Deng H, Driscoll W, Nikolov M, Tolia K, Wiener-Kronish JP. A remote surveillance platform to monitor general care ward surgical patients for acute physiologic deterioration. Anesth Analg. 2021;133:933–939.
    39. Klumpner TT, Kountanis JA, Meyer SR, et al. Use of a novel electronic maternal surveillance system and the maternal early warning criteria to detect severe postpartum hemorrhage. Anesth Analg. 2020;131:857–865.
    40. Shelton KT, Wiener-Kronish JP. Evolving role of anesthesiology intensivists in cardiothoracic critical care. Anesthesiology. 2020;133:1120–1126.
    Copyright © 2022 International Anesthesia Research Society