Anesthesia & Analgesia:
Technology, Computing, and Simulation: Special Article
Decision Support for Hemodynamic Management: From Graphical Displays to Closed Loop Systems
Michard, Frederic MD, PhD
From the Department of Critical Care, Edwards Lifesciences, Irvine, California.
Accepted for publication November 02, 2012.
Published ahead of print February 28, 2013
Conflict of Interest: See Disclosures at the end of the article.
Reprints will not be available from the author.
Address correspondence to Frederic Michard, MD, PhD, Critical Care, Edwards Lifesciences, 1 Edwards Way, Irvine, CA. Address e-mail to firstname.lastname@example.org.
The way hemodynamic therapies are delivered today in anesthesia and critical care is suboptimal. Hemodynamic variables are not always understood correctly and used properly. The adoption of hemodynamic goal-directed strategies, known to be clinically useful, is poor. Ensuring therapies are delivered effectively is the goal of decision support tools and closed loop systems. Graphical displays (metaphor screens) may help clinicians to better capture and integrate the multivariable hemodynamic information. This may result in faster and more accurate diagnosis and therapeutic decisions. Graphical displays (target screens) have the potential to increase adherence to goal-directed strategies and ultimately improve patients’ outcomes, but this remains to be confirmed by prospective studies. Closed loop systems are the ultimate solution to ensure therapies are delivered. However, most therapeutic decisions cannot be based on a limited number of output variables. Therefore, one should focus on the development of systems designed to relieve clinicians from very simple and repetitive tasks. Whether intraoperative goal-directed fluid therapy may be one of these tasks remains to be evaluated.
“The fundamental problem with the quality of American medicine is that we’ve failed to view delivery of health care as a science. The tasks of medical science fall into three buckets. One is understanding disease biology. One is finding effective therapies. And one is ensuring those therapies are delivered effectively. That third bucket has been almost totally ignored by research funders, government, and academia. It’s viewed as the art of medicine. That’s a mistake, a huge mistake. And from a taxpayer’s perspective it’s outrageous.”
—Peter Pronovost, Johns Hopkins Hospital, Baltimore, MD1
Ensuring therapies are delivered effectively, the third task of medical science according to Dr. Pronovost,1 is the ultimate goal of decision support and closed loop systems. With regard to hemodynamic management, cardiocirculatory variables are not always understood and used properly by clinicians. Several studies have demonstrated clinicians’ insufficient knowledge of right-heart catheterization at the bedside.2–4 These observations have been made despite the fact that the pulmonary artery catheter (PAC) has been used for decades, and cardiac physiology is still often described and taught to medical students using PAC variables (e.g., hypovolemic shock = low filling pressures + low cardiac output + high systemic vascular resistance, etc.). Many new hemodynamic monitoring tools have been introduced over the last decades.5 Some of them provide new variables such as the corrected flow time for Doppler, the thoracic fluid content for bioreactance, the global end-diastolic volume for transpulmonary thermodilution, etc. Too often the advantages and limitations of these new variables are poorly understood, when not simply ignored by clinicians who may have to take care of patients monitored with these new techniques (for example during a night shift). As a result, ensuring hemodynamic therapies are delivered effectively may start by helping clinicians to understand the meaning of the variables they measure.
There is also now a consensus regarding the fact that no hemodynamic monitoring technique can improve outcome by itself.6 Clinical and economical benefits can only be expected when the information provided by hemodynamic tools is used to feed treatment protocols, with well-defined hemodynamic goals (goal-directed therapy). For instance, in medium- to high-risk surgical patients, there is a significant body of evidence demonstrating that perioperative goal-directed therapy decreases renal, gastrointestinal, and infectious postoperative complications.7–12 This beneficial effect on morbidity is often associated with a decrease in hospital length of stay11 and cost savings. Despite this level of evidence, the adoption of perioperative goal-directed therapy is poor.13 A survey14 published in 2011 showed that perioperative goal-directed therapy is used by only 5.4% of US anesthesiologists taking care of high-risk surgical patients.
In the present article, I will discuss what could be done, and more precisely, what kind of decision support tools could be used to ensure therapies are delivered effectively in anesthesia and critical care, focusing on perioperative hemodynamic management.
Graphical Displays for Data Integration
Over a couple of decades graphics and icons have become key components of the interaction between humans and machines such as computers, cell phones, tablets, or global positioning systems. The anesthesiologist work station has often been compared with an airplane cockpit, the cover of the August 2010 issue of Anesthesia & Analgesia being one of the best and most recent examples. What is striking when looking at the cockpit of recent commercial airplanes, such as the Dreamliner 787 from Boeing (Chicago, IL), is that key information is delivered to the pilots by graphical displays (Fig. 1). This is the illustration that the right picture is worth a thousand numbers, and what is true for pilots is probably also true for anesthesiologists and intensivists.15 Studies have shown that the human brain has difficulties processing more than 5 to 7 variables simultaneously,16 and this is precisely what clinicians’ brains have to do when taking care of patients in whom the cardiorespiratory status is closely monitored. Graphical displays have the advantage integrating multiple variables into 1 specific shape or pattern corresponding to a specific diagnosis and therapeutic decision. When patients’ variables are abnormal, the deviations from normal are quickly noticed, because the normal shapes are preattentively processed.17 That is, the abnormally shaped objects clearly emerge from their surroundings.18 As a result, many studies have demonstrated the superiority of graphical displays over classical numerical displays. One of the first examples is the article published in 1995 by Gurushanthaiah et al.19 Using a computer simulator in a laboratory setting, they compared the behavior of anesthesia residents using a classical numerical display of vital variables (blood pressure, heart rate, SpO2, FIO2, end-tidal CO2, tidal volume, and maximal airway pressure) or a graphical display depicting either a histogram or a polygon. They showed a faster detection time and an increased accuracy to detect changes and direction of change with the histogram or polygon displays. Response latency was lower (2.2 ± 0.1 and 2.3 ± 0.1 seconds vs 2.7 ± 0.1 seconds) and percentage accuracy was higher (74% and 72% vs 60%) with the histogram and the polygon than with the numeric screen, respectively. In another study, Blike et al.20 investigated whether a graphical display may affect the ability of anesthesiologists to recognize and differentiate rapidly and correctly 5 etiologies of shock: anaphylaxis, bradycardia, myocardial ischemia, hypovolemia, and pulmonary embolism. Datasets consisted of heart rate, blood pressure, pulmonary artery pressure, central venous pressure, and cardiac output. Information was presented either in an alpha-numeric format or in a graphical format. Anesthesiologists using the graphical display committed significantly fewer diagnostic errors (1.4% vs 4.1%, P < 0.001) when interpreting physiologic data. In addition, both the recognition of no shock and the diagnosis of shock etiology were completed more rapidly (1.0 ± 8.3 and 1.3 ± 9.7 seconds earlier, respectively) when the graphical display was used. A systematic review published in 2008 by Görges and Staggers21 reported 18 studies showing that graphical displays allow a faster detection of changes in physiologic variables, 13 studies a more accurate diagnosis, and 3 studies a decreased mental workload.
Toward Intuitive and Metaphor Displays
Yet, we know little about which graphical displays are optimal and why particular design work. However, today the trend is clearly toward the development of intuitive and metaphor displays for at least 2 main reasons: clinicians (like any other people) do not regularly read user manuals and they do not spend a lot of time looking at their monitors. This latter concept of “at a glance” monitoring has been well documented by Ford et al.22 who have video recorded anesthesiologists during surgical procedures. Three segments of video, each 10 minutes long, were selected for analysis (at the beginning, the middle, and the end of surgery). Anesthesiologists spent little time observing the monitoring display (32.3 ± 4.5 seconds per 10-minute segment), and this did not change across the 3 segments. Monitor observation occurred in frequent, brief glances. These glances were on average between 1.5 and 2.1 seconds in duration and occurred between 15 and 20 times during each 10-minute segment. These data strongly suggest that displays should be developed to optimize the information obtained from brief glances at the monitor. The Center for the Representation of Multi-Dimensional Information (http://www.cromdi.utah.edu) has developed very innovative, abstract, and futuristic graphical displays (Fig. 2). Several metaphor or anatomical displays, easy to recognize and understand at a glance, have also been developed and tested. A graphical cardiovascular display18,23 has been designed to help clinicians capture quickly and accurately changes in hemodynamic variables measured by PACs (Fig. 3). This graphical display was able to improve clinicians’ ability to detect, diagnose, manage, and treat critical events in a simulated intraoperative hemorrhage-induced myocardial infarction.18 Anesthesiologists who used the graphic display detected myocardial ischemia 2 minutes sooner than those who did not use the display and treatment was initiated sooner (2.5 vs 4.9 minutes).18 Systolic blood pressure and central venous pressure deviated less from baseline, and arterial oxygen saturation was higher at the end of the case when the graphic display was used.18 Several anatomical graphical displays have become commercially available over the last few years. Hamilton Medical (Bonaduz, Switzerland) has developed a screen for mechanical ventilators allowing clinicians to visualize the cuff pressure, the compliance of the respiratory system, the airway resistance, the diaphragmatic contraction, and an index of preload dependency (Fig. 4). Albert et al.24 have investigated the value of this screen by comparing practitioner (nurses, physicians, and respiratory therapists) performance to identify airway resistance and compliance of the respiratory system using a traditional screen (digital values, waveforms, and spirometry plots) or the graphical metaphor screen. Participants had faster response times for resistance (6.22 vs 11.22 seconds, P < 0.05) and for compliance estimations (3.99 vs 7.91 seconds, P < 0.05). Accuracy was also improved both for the determination of resistance (87.80% vs 58.33%, P < 0.05) and compliance (95.83% vs 70.83%, P < 0.05). Mental workload was significantly decreased with the graphical display.24 Edwards Lifesciences (Irvine, CA) has implemented a screen on the new hemodynamic monitoring platform EV1000 allowing the visualization of intravascular volume status, fluid responsiveness (i.e., the position on the Frank-Starling curve), cardiac output, vasomotor tone, and lung edema at a glance (Fig. 4). Studies are required to confirm that this new heart-lung metaphor screen can help clinicians to better understand the hemodynamic situation and make faster and more accurate therapeutic decisions.
Graphical Displays for Goal-Directed Therapy
Graphical displays may be even more valuable when they are designed to guide clinicians along the paths of treatment protocols or goal-directed strategies. Vallée et al.25 designed a visual tool called hemodynamic target to help clinicians interested in adopting the surviving sepsis guidelines. Their target displayed on the same graph central venous oxygen saturation, lactates, mean arterial blood pressure, SaO2, and cardiac index. They compared the way septic patients were managed before and after the implementation of their hemodynamic target. The use of the target was associated with a better compliance to the surviving sepsis guidelines, illustrated by more frequent measurements of ScvO2 (5.5 vs 0.2 measurements/patient, P < 0.001) and lactates (5.5 vs 2.9 measurements/patients, P = 0.001). This study suggests that target screens can increase the compliance to guidelines or goal-directed strategies. The above-mentioned systematic review by Görges and Staggers21 reported 4 other studies showing that the use of graphical displays is associated with an increased percentage of time spent in target. In line with these findings, several companies have recently developed customizable target screens allowing clinicians to visualize if their patient is in target, and if this is not the case, to understand at a glance what should be done to reach the therapeutic target. Several examples are shown in Figure 5. Studies published so far were not designed to show any outcome benefit.21,25 However, one can reasonably assume that a better compliance to guidelines may ultimately result in a better outcome. One can envision future outcome studies where 3 arms would be compared: a control arm with standard care, a goal-directed therapy arm with oral and/or written guidelines, and a goal-directed therapy arm with a target screen helping (not to say forcing) clinicians to reach the predefined hemodynamic goals. Assuming goal-directed therapy is useful, one can reasonably imagine a better outcome in the third group, not because of the target screen itself, but because of a better compliance to goal-directed therapy related to the use of the target screen.
CLOSED LOOP SYSTEMS
With the rapid development of computers in the 1990s, it became technically doable to ask “smart” and/or “learning” systems to make diagnosis proposals. Several computerized diagnostic tools have been developed, tested, and used in anesthesia and intensive care.26–28 With regard to hemodynamic management, Squara et al.27,28 have developed a computer program named Hemodyn for interpreting pulmonary artery catheterization data. Two studies27,28 have shown a fairly good agreement between Hemodyn and hemodynamic experts but a poor agreement between Hemodyn and residents, supporting the notion that Hemodyn may be useful for nonexperts.
Closed loops systems are a step ahead of computerized diagnostic tools: they are not only able to make therapeutic decisions, but they are also able to deliver treatments automatically. A large body of literature regarding closed loop systems is related to the type of controller, from simple so-called proportional integral derivative controllers to more complex neural network, fuzzy logic, and Bayesian probability systems.29 This is clearly beyond the scope of this article, therefore this technical aspect will not be discussed. In fact, engineers have been able to build smart systems administering treatments automatically.29 For example, closed loop systems have been used successfully to titrate anesthesia, using the Bispectral Index as the output variable triggering either an increase or a decrease in hypnotic drugs.30–33 Several studies have suggested that such systems are actually able to outperform manual control.32,33
Limitations and Risks
One of the main challenges with computerized diagnostic tools and closed loop systems is the amount of data or information that must be accessed to enable the process, limiting de facto their ability to provide accurate diagnosis and therapy. Therapeutic decisions, even very simple at first glance like the decision to administer fluid, are actually based on an impressive number of variables and amount of information. For instance, the way fluid is administered to a patient (type, volume, and speed) depends on many factors: clinical knowledge (training and experience, institutional or society guidelines and recommendations), comorbidities (left ventricular diastolic dysfunction, cor pulmonale), clinical examination (clinical signs of hypoperfusion, blood pressure, heart rate), biological variables (acidosis, lactate, arterial oxygenation, hemoglobin) and hemodynamic variables (cardiac filling pressures, dynamic predictors of fluid responsiveness), not to mention the amount and type of fluid the patient already received. Therefore, because computerized and closed loop systems are usually unable to integrate all the necessary determinants of therapeutic decisions, it would be naïve, not to say dangerous, to believe they could be used to make optimal therapeutic decisions in all circumstances.
For instance, several closed loop systems have demonstrated their ability to maintain blood pressure in a predefined target zone by automatically controlling the administration of fluid,34 or of vasopressors,35 or even of analgesic drugs.36 However, hypotension, a very common event in patients undergoing surgery, may be related to a decrease in flow or to a decrease in vascular tone. The former should be treated by increasing preload (with fluid or red blood cells) or by increasing contractility (with inotropes), and the later by giving vasopressors or simply by decreasing anesthesia and/or analgesia-induced vasodilation. If we can count on closed loop systems to administer a therapy, it becomes more challenging to ensure the right therapy is always delivered. Maintaining blood pressure artificially by the use of vasopressors is indeed technically doable, but is not the right therapeutic answer if the cause of hypotension is hypovolemia or left ventricular failure.
Another limiting factor for the development of automatic or closed loop systems in anesthesia and critical care medicine is the lack of sensor/signal redundancy. In aviation, autopilot systems receive the same information (e.g., airspeed) from several sensors (e.g., pitot tubes) simultaneously. It allows the distinction between signal and noise/artifacts and guarantees the quality of the information processed by the controller. We could imagine doubling or tripling the number of sensors used in our patients (e.g., we have enough fingers to welcome several pulse oximeter sensors) but the cost of such strategy would be prohibitive.
Do We Need Closed Loop Systems?
Assuming the above limitations and risks could be overcome, the value of automatic and closed loop systems will likely depend on manpower and then should be more visible when clinicians are in short supply, pressed for time, and overwhelmed with patients. A good illustration is given by the SmartCare closed loop system developed and commercialized by Dräger (Lubeck, Germany) for the automatic weaning from mechanical ventilation. This system is designed to automatically drive the level of pressure support, to automatically perform spontaneous breathing trials, and to display an incentive message for extubation when the trial is successfully passed. Two clinical evaluations37,38 have been published, with apparent conflicting results. The first study37 showed that the use of the computer-driven system was useful for decreasing the duration of mechanical ventilation (from 12 to 7.5 days, P = 0.003) and intensive care unit stay (from 15.5 to 12 days, P = 0.02). The second study38 failed to confirm these findings. Interestingly, in the first study the authors explained their findings as follows: “The computer-driven weaning protocol does not depend on the willingness or availability of the staff, and full compliance with the weaning protocol is therefore ensured. A permanent evaluation and adjustment of ventilatory support cannot be continuously performed by caregivers.” In the second study, experienced critical care specialty nurses were in charge of the manual weaning process and the nurse to patient ratio was 1:1. As a result, the computer-driven system was not needed to ensure optimal weaning from the ventilator. These studies suggest that closed loop systems may be helpful when manpower and/or workload are a limitation to quality of care. In other circumstances, and particularly in the operating room where there is always someone physically present to monitor the patient, their potential usefulness is questionable.
For all reasons discussed above, one may have to envision automatic and/or closed loop systems more as tools designed to relieve clinicians from very simple and repetitive tasks, when the therapeutic response is not disputable, such as adapting pressure support to tidal volume and respiratory rate, or adapting FIO2 to SaO2, or adapting insulin rate to blood glucose values.39,40 The operating room is a very busy environment where, for anesthesiologists, priority is logically given to anesthesia and analgesia. This may partly explain the poor compliance to hemodynamic goal-directed strategies in patients undergoing high-risk surgery.41 Another limitation to the use of hemodynamic treatment protocols is the repetitive measurements and actions required to follow these protocols.41 The fluid management protocol today recommended in the United Kingdom,42 and implemented with success in several hospitals,43 is based on stroke volume measurements and adaptation of fluid therapy every 5 to 10 minutes. This is fairly simple but very repetitive and somewhat time consuming. Ideally, this task may be automated in the future. This would not only free-up easily distracted clinicians but also, and maybe more importantly, ensure a perfect compliance to the intraoperative fluid management protocol.
Graphical displays (metaphor screens) may help clinicians to better capture and integrate the multivariable hemodynamic information. This may result in faster and more accurate diagnosis and therapeutic decisions. Graphical displays (target screens) have the potential to increase adherence to goal-directed strategies and ultimately to improve patients’ outcomes, but this remains to be confirmed by prospective studies. Closed loop systems are the ultimate solution to ensure therapies are delivered. However, most therapeutic decisions cannot be based on a limited number of output variables. Therefore, one should focus on the development of systems designed to relieve clinicians from very simple and repetitive tasks. Whether intraoperative goal-directed fluid therapy may be one of these tasks remains to be evaluated.
Name: Frederic Michard, MD, PhD.
Contribution: This author wrote the manuscript.
Conflicts of Interest: Dr. Michard has been a consultant for several companies who have developed decision support tools and graphical displays: Pulsion Medical Systems (Germany), Hamilton Medical (Switzerland), UPMED (Germany), and Dixtal (Brazil). He is coinventor on a graphical display patent and on a closed loop system patent. He is now Vice President, Global Medical Strategy, at Edwards Lifesciences (Irvine, CA), where he is involved in the development of decision support tools. He is also a visiting doctor at the University Hospital of Lausanne, Switzerland.
This manuscript was handled by: Dwayne R. Westenskow, PhD.
1. Gawande A The Checklist Manifesto. How to Get Things Right. 2010 London Profile Books Ltd
2. Iberti TJ, Fischer EP, Leibowitz AB, Panacek EA, Silverstein JH, Albertson TE. A multicenter study of physicians’ knowledge of the pulmonary artery catheter. Pulmonary Artery Catheter Study Group. JAMA. 1990;264:2928–32
3. Gnaegi A, Feihl F, Perret C. Intensive care physicians’ insufficient knowledge of right-heart catheterization at the bedside: time to act? Crit Care Med. 1997;25:213–20
4. Squara P, Bennett D, Perret C. Pulmonary artery catheter: does the problem lie in the users? Chest. 2002;121:2009–15
5. Geisen M, Rhodes A, Cecconi M. Less-invasive approaches to perioperative haemodynamic optimization. Curr Opin Crit Care. 2012;18:377–84
6. Vincent JL, Rhodes A, Perel A, Martin GS, Della Rocca G, Vallet B, Pinsky MR, Hofer CK, Teboul JL, de Boode WP, Scolletta S, Vieillard-Baron A, De Backer D, Walley KR, Maggiorini M, Singer M. Clinical review: Update on hemodynamic monitoring–a consensus of 16. Crit Care. 2011;15:229
7. Hamilton MA, Cecconi M, Rhodes A. A systematic review and meta-analysis on the use of preemptive hemodynamic intervention to improve postoperative outcomes in moderate and high-risk surgical patients. Anesth Analg. 2011;112:1392–402
8. Brienza N, Giglio MT, Marucci M, Fiore T. Does perioperative hemodynamic optimization protect renal function in surgical patients? A meta-analytic study. Crit Care Med. 2009;37:2079–90
9. Giglio MT, Marucci M, Testini M, Brienza N. Goal-directed haemodynamic therapy and gastrointestinal complications in major surgery: a meta-analysis of randomized controlled trials. Br J Anaesth. 2009;103:637–46
10. Dalfino L, Giglio MT, Puntillo F, Marucci M, Brienza N. Haemodynamic goal-directed therapy and postoperative infections: earlier is better. A systematic review and meta-analysis. Crit Care. 2011;15:R154
11. Corcoran T, Rhodes JE, Clarke S, Myles PS, Ho KM. Perioperative fluid management strategies in major surgery: a stratified meta-analysis. Anesth Analg. 2012;114:640–51
12. Michard F. The burden of high-risk surgery and the potential benefit of goal-directed strategies. Crit Care. 2011;15:447
13. Miller TE, Roche AM, Gan TJ. Poor adoption of hemodynamic optimization during major surgery: are we practicing substandard care? Anesth Analg. 2011;112:1274–6
14. Cannesson M, Pestel G, Ricks C, Hoeft A, Perel A. Hemodynamic monitoring and management in patients undergoing high risk surgery: a survey among North American and European anesthesiologists. Crit Care. 2011;15:R197
15. Drews FA, Westenskow DR. The right picture is worth a thousand numbers: data displays in anesthesia. Hum Factors. 2006;48:59–71
16. Miller GA. The magical number seven, plus or minus two: some limits on our capacity for processing information. 1956. Psychol Rev. 1994;101:343–52
17. Triesman A. Preattentive processing in vision. Comput Vis Graphics Image Proc. 1985;31:156–77
18. Agutter J, Drews F, Syroid N, Westneskow D, Albert R, Strayer D, Bermudez J, Weinger MB. Evaluation of graphic cardiovascular display in a high-fidelity simulator. Anesth Analg. 2003;97:1403–13
19. Gurushanthaiah K, Weinger MB, Englund CE. Visual display format affects the ability of anesthesiologists to detect acute physiologic changes. A laboratory study employing a clinical display simulator. Anesthesiology. 1995;83:1184–93
20. Blike GT, Surgenor SD, Whalen K. A graphical object display improves anesthesiologists’ performance on a simulated diagnostic task. J Clin Monit Comput. 1999;15:37–44
21. Görges M, Staggers N. Evaluations of physiological monitoring displays: a systematic review. J Clin Monit Comput. 2008;22:45–66
22. Ford S, Birmingham E, King A, Lim J, Ansermino JM. At-a-glance monitoring: covert observations of anesthesiologists in the operating room. Anesth Analg. 2010;111:653–8
23. Albert RW, Agutter JA, Syroid ND, Johnson KB, Loeb RG, Westenskow DR. A simulation-based evaluation of a graphic cardiovascular display. Anesth Analg. 2007;105:1303–11 table of contents
24. Albert R, Syroid N, Agutter J, Westenskow D. Improving patient state identification with integrated graphic data presentation. Anesth Analg. 2008;106:S30
25. Vallée F, Fourcade O, Marty P, Sanchez P, Samii K, Genestal M. The hemodynamic “target”: a visual tool of goal-directed therapy for septic patients. Clinics (Sao Paulo). 2007;62:447–54
26. Morris AH. Developing and implementing computerized protocols for standardization of clinical decisions. Ann Intern Med. 2000;132:373–83
27. Squara P, Dhainaut JF, Lamy M, Perret C, Schremmer B, Poli S, Larbuisson R, de Gournay JM, Armaganidis A, Gourgon R, Bleichner G. Computer assistance for hemodynamic evaluation. J Crit Care. 1989;4:273–82
28. Squara P, Fourquet E, Jacquet L, Broccard A, Uhlig T, Rhodes A, Bakker J, Perret C. A computer program for interpreting pulmonary artery catheterization data: results of the European HEMODYN Resident Study. Intensive Care Med. 2003;29:735–41
29. Fields AM, Fields KM, Cannon JW. Closed-loop systems for drug delivery. Curr Opin Anaesthesiol. 2008;21:446–51
30. Hemmerling TM. Automated anesthesia. Curr Opin Anaesthesiol. 2009;22:757–63
31. Absalom AR, De Keyser R, Struys MM. Closed loop anesthesia: are we getting close to finding the holy grail? Anesth Analg. 2011;112:516–8
32. Locher S, Stadler KS, Boehlen T, Bouillon T, Leibundgut D, Schumacher PM, Wymann R, Zbinden AM. A new closed-loop control system for isoflurane using bispectral index outperforms manual control. Anesthesiology. 2004;101:591–602
33. Liu N, Chazot T, Genty A, Landais A, Restoux A, McGee K, Laloë PA, Trillat B, Barvais L, Fischler M. Titration of propofol for anesthetic induction and maintenance guided by the bispectral index: closed-loop versus manual control: a prospective, randomized, multicenter study. Anesthesiology. 2006;104:686–95
34. Kramer GC, Kinsky MP, Prough DS, Salinas J, Sondeen JL, Hazel-Scerbo ML, Mitchell CE. Closed-loop control of fluid therapy for treatment of hypovolemia. J Trauma. 2008;64:S333–41
35. Ngan Kee WD, Tam YH, Khaw KS, Ng FF, Critchley LA, Karmakar MK. Closed-loop feedback computer-controlled infusion of phenylephrine for maintaining blood pressure during spinal anaesthesia for caesarean section: a preliminary descriptive study. Anaesthesia. 2007;62:1251–6
36. Luginbühl M, Bieniok C, Leibundgut D, Wymann R, Gentilini A, Schnider TW. Closed-loop control of mean arterial blood pressure during surgery with alfentanil: clinical evaluation of a novel model-based predictive controller. Anesthesiology. 2006;105:462–70
37. Lellouche F, Mancebo J, Jolliet P, Roeseler J, Schortgen F, Dojat M, Cabello B, Bouadma L, Rodriguez P, Maggiore S, Reynaert M, Mersmann S, Brochard L. A multicenter randomized trial of computer-driven protocolized weaning from mechanical ventilation. Am J Respir Crit Care Med. 2006;174:894–900
38. Rose L, Presneill JJ, Johnston L, Cade JF. A randomised, controlled trial of conventional versus automated weaning from mechanical ventilation using SmartCare/PS. Intensive Care Med. 2008;34:1788–95
39. Wysocki M, Brunner JX. Closed-loop ventilation: an emerging standard of care? Crit Care Clin. 2007;23:223–40, ix
40. Kumareswaran K, Evans ML, Hovorka R. Closed-loop insulin delivery: towards improved diabetes care. Discov Med. 2012;13:159–70
41. Michard F, Biais M. Rational fluid management: dissecting facts from fiction. Br J Anaesth. 2012;108:369–71
42. Mythen MG, Swart M, Acheson N, Crawford R, Jones K, Kuper M, McGrath JS, Horgan A. Perioperative fluid management: consensus statement from the enhanced recovery partnership. Perioperative Medicine. 2012;1:2
43. Kuper M, Gold SJ, Callow C, Quraishi T, King S, Mulreany A, Bianchi M, Conway DH. Intraoperative fluid management guided by oesophageal Doppler monitoring. BMJ. 2011;342:d3016
© 2013 International Anesthesia Research Society