Anesth Analg 2003;97:1403-1413
© 2003 International Anesthesia Research Society
TECHNOLOGY, COMPUTING, AND SIMULATION
Evaluation of Graphic Cardiovascular Display in a High-Fidelity Simulator
James Agutter, M.Arch*,
Frank Drews, PhD
,
Noah Syroid, MS
,
Dwayne Westneskow, PhD,
Rob Albert, MS,
David Strayer, PhD,
Julio Bermudez, PhD*, and
Matthew B. Weinger, MD
*Graduate School of Architecture,
Department of Anesthesiology, and
Department of Psychology, University of Utah, Salt Lake City, Utah; and
Department of Anesthesiology, University of California, San Diego, and San Diego Center for Patient Safety, Veterans Affairs San Diego Medical Center, San Diego, California
Address correspondence to James Agutter, M.Arch, Graduate School of Architecture, University of Utah, 375 S. 1530 E. Rm. 235, Salt Lake City, UT 84112. Address e-mail to agutterja{at}arch.utah.edu
 |
Abstract
|
|---|
"Human error" in anesthesia can be attributed to misleading information from patient monitors or to the physician’s failure to recognize a pattern. A graphic representation of monitored data may provide better support for detection, diagnosis, and treatment. We designed a graphic display to show hemodynamic variables. Twenty anesthesiologists were asked to assume care of a simulated patient. Half the participants used the graphic cardiovascular display; the other half used a Datex As/3 monitor. One scenario was a total hip replacement with a transfusion reaction to mismatched blood. The second scenario was a radical prostatectomy with 1.5 L of blood loss and myocardial ischemia. Subjects who used the graphic display detected myocardial ischemia 2 min sooner than those who did not use the display. Treatment was initiated sooner (2.5 versus 4.9 min). There were no significant differences between groups in the hip replacement scenario. Systolic blood pressure deviated less from baseline, central venous pressure was closer to its baseline, and arterial oxygen saturation was higher at the end of the case when the graphic display was used. The study lends some support for the hypothesis that providing clinical information graphically in a display designed with emergent features and functional relationships can improve clinicians’ ability to detect, diagnose, manage, and treat critical cardiovascular events in a simulated environment.
IMPLICATIONS: A graphic representation of monitored data may provide better support for detection, diagnosis, and treatment. A user-centered design process led to a novel object-oriented graphic display of hemodynamic variables containing emergent features and functional relationships. In a simulated environment, this display appeared to support clinicians’ ability to diagnose, manage, and treat a critical cardiovascular event in a simulated environment. We designed a graphic display to show hemodynamic variables. The study provides some support for the hypothesis that providing clinical information graphically in a display designed with emergent features and functional relationships can improve clinicians’ ability to detect, diagnosis, mange, and treat critical cardiovascular events in a simulated environment.
|
Introduction
|
|---|
"Human error" is possibly associated with >80% of anesthesia-related critical incidents and >50% of anesthesia-related deaths (1). Critical incident studies report that adverse outcomes are often described as a catastrophic "evolving chain" of subtle events, each of which alone might not have led to a disaster (2,3). Many of these events can be directly attributed to erroneous or misleading information from patient monitors or to the physician’s failure to recognize a pattern in the patient’s clinical information that would have led to a correct diagnosis of the problem (4,5). We sought to test the hypothesis that providing clinical information graphically in a display designed with emergent features and functional relationships can improve the anesthesiologist’s ability to detect, diagnose, manage, and treat a cardiovascular (CV) crisis in a simulated environment.
Most currently available anesthesia display systems use a "single-sensor-single-indicator" display paradigm (6). That is, a single display element is provided for each sensor used. As a result, clinicians must observe and integrate multiple data elements generated by the independent sensors. This process of sequential, piecemeal data gathering may be an impediment to a coherent understanding of the patient’s underlying physiologic processes, particularly during large workload or crisis situations (3). An attractive enhancement to traditional physiologic displays is to provide a higher level of integration of the data with a graphic representation that is consistent with the clinician’s cognitive representation (mental model) of patient physiology. Such integrated graphic displays may provide better support for diagnosis and treatment of problems involving alterations of multiple physiologic variables.
The more complex and critical the information, the more imperative it is to communicate that information effectively (7). When the visual information is presented in a manner consistent with how the user must process that information, the resulting performance is often more rapid, accurate, and consistent. Therefore, it is imperative that the visual presentation minimizes the mental transformations required to effectively process the information. Graphic representation has been shown to more effectively support the assimilation, interpretation, manipulation, and use of information (8–12). In this study, we used an iterative user-centered design to create a novel integrated graphic CV display and tested its potential clinical value in a high-fidelity whole-body patient simulator.
|
Methods: Graphic Display Design Process
|
|---|
A graphic display was designed and implemented to show functional relationships of CV physiology by integrating related hemodynamic variables. The design process began by examining other CV monitors and generating a list of the CV variables that provide critical clinical information about the patient’s CV state (Table 1).
To conform to the clinician’s mental model of the CV system, the variables from Table 1 were spatially arranged to diagrammatically mimic physiologic blood flow through the circulatory system (Fig. 1). On the left, venous blood (returning from the systemic capillaries) flows into the vena cava (a). The right heart (b) pumps the deoxygenated blood (blue color) through the pulmonary arteries (c) to the lungs (d); oxygenated (red) blood in the pulmonary veins (e) flows to the left heart (f) where it is pumped into the aorta (g). This arrangement places relevant variables together, depicts more clearly the relationships between variables, and generates display patterns consistent with changing CV physiology. We hypothesized that this approach would facilitate more rapid understanding and diagnosis of simulated CV events by highlighting dynamic and interrelated changes in preload, afterload, and cardiac output (CO).
View larger version (43K):
[in this window]
[in a new window]
|
Figure 1. Cardiovascular display. CVP = central venous, PAP = pulmonary artery pressure, LAP = left arterial pressure, PVR = pulmonary vascular resistance, HR = heart rate, SV = stroke volume, CO = cardiac output, MAP = mean arterial blood pressure, SVR = systemic vascular resistance, SaO2 = arterial oxygen saturation.
|
|
The display conveys the look of a three-dimensional pipe and is a geometric graphic metaphor for a blood vessel. Vertical movement of each part of the pipe represents a change in blood pressure. As the blood pressure increases for a portion of the CV system, the corresponding object becomes larger by increasing in diameter in the vertical direction (Fig. 2). Thus, the objects’ movement provides a dual notion of the patient’s pressure and volume status. In addition, the display was designed using normally shaped and uniformly spaced elements to create a smooth balanced design (13). Thus, when variables are normal, they all occur within a uniform tubular reference frame. The design was intended to promote rapid detection of change just as artificial horizon and polygon displays improve performance in detecting events in aviation and anesthesia, respectively (14,15). When patient variables are abnormal, the deviations from normal are quickly noticed, because the normal shapes are preattentively processed (16). That is, the abnormally shaped objects clearly emerge from their surroundings. (Fig. 2)
View larger version (56K):
[in this window]
[in a new window]
|
Figure 2. Cardiovascular display showing how its shapes change during the development of a myocardial ischemia. CVP = central venous pressure, PAP = pulmonary artery pressure, LAP = left arterial pressure, PVR = pulmonary vascular resistance, HR = heart rate, SV = stroke volume, CO = cardiac output, SVR = systemic vascular resistance, MAP = mean arterial blood pressure, SaO2 = arterial oxygen saturation.
|
|
The CV display evolved in response to extensive clinician feedback and usability testing (17,18). Each design iteration was evaluated using a four-step usability testing protocol. This shows how design changes impact the intuitiveness (ability to recognize the general organizational aspects, the ability to easily map specific data to the general organization), and usability (ability to recognize appropriate patterns). By maximizing the ease of use and integrating end users in the design process, we anticipated that acceptance of the display would be enhanced. The questions were intended to answer questions regarding:
1) Intuitiveness
- Are the visual objects in the designed metaphor easily recognizable?
- Do the data clearly associate with the information to be presented by the designed elements in the metaphor?
2) Usability
- Is the metaphor effective for visualizing the pertinent patterns of information?
- Changes in design are made by using the evaluation protocol for each design iteration. The results direct which portions of the design need refinement. The protocol is presented as static images on paper or on a computer. To assess how design changes affect intuitiveness and the confusion matrices, we compared the original and the new designs. An improvement in intuitiveness, for example, would be reflected in a reliable increase in correct judgments about the anatomical features and the physiologic structures presented in display images.
Simulation Study Methods
After approval from the IRB at the University of Utah Health Sciences Center and written informed consent, 20 anesthesiologists (average experience 6.5 yr, range of experience 1–25 yr) participated in the study. Each subject was studied only once (between subjects experimental design). Subjects received $50 compensation. Study sessions lasted approximately 1 h.
Training on Display Use.
All subjects, regardless of subsequent experimental group, underwent training on the use of the novel CV display. The self-guided computer-based training program began by explaining the anatomical organization of the display. Subjects were then told how the measurement values were mapped on the display. Subjects manipulated slider bars on a graphical user interface to modify each physiologic variable to understand how the corresponding shapes changed. Then, they were asked to match the appropriate state of CV shock or to identify the expected CV response to a specific drug from a list that corresponded to an image of the metaphor. By design, the training did not show images of the events that were later used for testing (myocardial ischemia [MI], hypovolemia, or anaphylaxis [AP]). Finally, subjects were informed about the testing procedure. Training lasted about 15 min for each subject.
Scenarios.
Participants were asked to assume care of a simulated patient (METI, Sarasota, FL) because the preceding anesthesiologist had become ill and had to leave the operating room (OR). Subjects were randomly assigned to one of two conditions in which half the participants used a traditional Datex AS/3 monitor (Datex-Ohmeda, Andover, MA) along with the novel graphic CV display showing values for CO, systemic vascular resistance (SVR), pulmonary vascular resistance (PVR), left arterial pressure (LAP), and central venous pressure (CVP). The other half used a traditional Datex AS/3 monitor, and a secondary numeric monitor showing real-time values for CO, SVR, PVR, LAP, and CVP (the digital values of the same variables that appeared on the graphic display).
One scenario was a total hip replacement in which the anesthesiologist had inadvertently started a blood transfusion with mismatched blood 2 min before the subject’s arrival. An anaphylactic reaction ensued and worsened during the next 10 min (Appendix 1). The second scenario was a radical prostatectomy with a 1.5-L blood loss over the first 3 min after the subject’s assumption of care. After 5 min, the simulated patient experienced MI with associated ST segment changes that worsen over the ensuing 5 min (Appendix 2).
Procedure.
When the subjects arrived for the study, they completed a questionnaire asking about their clinical experience, the length of time that they worked before the study, caffeine consumption, sleep history, color-vision, and whether they required vision correction. Each subject then participated in a 15-min computer-based training video session that explained the features of the graphic display.
Subjects wore a wireless microphone and were videotaped as they participated in the study. The order of the two patient scenarios and display condition (with or without the graphic display) were completely balanced such that five subjects were randomly allocated to each of four patient/display experimental groups. The subjects were asked to think aloud during the case and to explain what they saw and what they were doing as they cared for the patient. The physical environment was similar to a real OR. Throughout the study, the Datex display was placed on a platform above the anesthesia machine with the supplemental display placed at the same level just to the left.
Subjects were asked to care for a simulated patient in a "normal anesthetized" state as they would during actual patient care. Subjects were instructed to detect any changes in the variables, diagnose any evolving clinical event as quickly as possible, and to treat it appropriately by administering the correct drugs. The simulator was programmed to respond appropriately to the subjects’ clinical interventions. Regardless of the treatment administered, the scenario lasted 10 min.
Four to five investigators were in the simulation room at all times. One or two investigators played the role of surgeons who followed a scripted dialog and made efforts to increase the realism of the simulation. One investigator played the role of an anesthesia assistant while another filmed the subjects and recorded detection times. A final investigator controlled the METI simulator according to a script, collected vital sign data from the simulated patient, and entered information about the drugs and fluids that the subject administered. After finishing each scenario, subjects completed the NASA-TLX workload questionnaire (Appendix 3). At the end of the experimental session, subjects answered a usability and realism questionnaire about the cardiac display (Appendix 4).
Simulation Setup.
The scenarios were conducted in a dedicated high-fidelity simulation center. The simulation room was configured to resemble an OR and the METI HPS Simulator was controlled from a station within the same room. The simulator scripts were created, reviewed, modified, and approved (Appendices 1 and 2) iteratively by three expert anesthesiologists. After implementation in the METI scripting language pilot runs confirmed validity and clinical realism.
Performance was assessed by measuring the time to detect an adverse event, the time to formulate the correct diagnosis, the time to provide supportive care, and deviations of vital signs from baseline. Time points were obtained from the time-code on videotapes of the scenarios. Naive research student assistants reviewed the raw video, recording the amount of time from the beginning of the scenario until the subject verbalized that a change was observed, verbalized a diagnosis, and initiated a treatment. The time of diagnosis was the time that the subject took to correctly identify and classify the event. Treatment time was based on the time of initiation of appropriate therapy for the critical event. The data were analyzed using one-tailed t-tests.
The following vital signs recorded from the patient simulator were analyzed for differences between subjects who used the graphic display versus those who did not:
- The amount of time the systolic blood pressure was 15 mm Hg less than baseline value
- The amount of time the patients experienced tachycardia (heart rate >110 bpm)
- The CVP at the end of the scenario
- The arterial oxygen saturation (SaO2) at the end of the scenario
Baseline was determined to be the steady-state systolic blood pressure, CV, and SaO2 before the occurrence of the critical event in the scenario (i.e., volume loss in the MI scenario and changes in SVR and CO in the AP scenario). A t-test (one tailed, assuming equal variance) was used to determine statistical significance.
|
Results
|
|---|
The time of detection, diagnosis, and treatment for both scenarios under the two display conditions are presented in Table 2. There was a significant improvement in detection times using the graphic CV display in the hypovolemia/MI (ISCH) scenario (P < 0.05, Figs. 3 and 4), but no significant difference was seen between groups in the AP scenario (P > 0.05). Time to diagnosis was not statistically different with the novel display in either the ISCH scenario (P > 0.05) or in the AP scenario (P > 0.05). However, treatment was instituted more quickly in the ISCH scenario (P < 0.05) when subjects used the novel display. There was no difference in time to treatment in the AP scenario (P > 0.05).
View this table:
[in this window]
[in a new window]
|
Table 2. Time in Seconds to Detect, Diagnose, and Treat Ischemia and Anaphylaxis With and Without the Graphic Display
|
|
View larger version (60K):
[in this window]
[in a new window]
|
Figure 3. Hypovolemia and myocardial ischemia. Histogram bars in each band show the time each subject took to detect, diagnose, or treat. The top bars show times for the display group and the bottom bars show control. The lower graph shows the evolution of the scenario.
|
|
View larger version (61K):
[in this window]
[in a new window]
|
Figure 4. Anaphylaxis. Histogram bars in each band show the time each subject took to detect, diagnose, or treat. The top bars show times for the display group and the bottom bars show control. The lower graph shows the evolution of the scenario. SVR = systemic vascular resistance, CVP = central venous pressure.
|
|
Table 3 shows the results of the vital signs analysis for the MI scenario. In the MI scenario, systolic blood pressure deviated less from baseline when the graphic display was used than when it was not (P < 0.05). When clinicians used the graphic display, the CVP at the end of the case was 5 mm Hg closer to its pre-event baseline (P < 0.05). Similarly, the SaO2 was higher at the end of the case when the graphic display was used (P < 0.05). There were no statistically significant differences in vital signs between groups in the AP scenario.
View this table:
[in this window]
[in a new window]
|
Table 3. The Vital Sign Results During the Myocardial Ischemia Scenario for Subjects Who Used the Graphic Display and for Those Who Did Not
|
|
Figure 3 shows time points of all subjects in the MI scenario in addition to visual information regarding scenario development. Figure 4 shows the same information for the AP scenario. These results are summarized in Table 2.
The graphic CV display was rated by the subjects as significantly more useful in the group that used the display (P < 0.01, 8.2 ± 1.1) than the group that did not (7.3 ± 1.3). Both groups rated the realism of the simulations with no statistical significance (overall mean rating 7 of 10 with range of 3–8) with no significant difference between. The two groups were also similar in their postscenario workload ratings (NASA-TLX). There were no significant differences in the subjects’ postsimulation ratings of the value of adding the new display to standard OR equipment. Table 4 shows information collected about subject demographics.
|
Discussion
|
|---|
The most significant finding of this study was that the graphic CV display aided the clinician in managing the complex (physically, mentally, and temporally demanding) simulated scenario of worsening MI brought on by rapid blood loss. The majority of participants (80% with display versus 40% without display) administered the optimal treatment for this adverse CV event. Several features of the integrated graphic display may have aided the subjects in detecting MI. However, we speculate that a major factor was the emergent graphic icon that indicated the presence of ST segment depression (Fig. 2). The simple shape and texture changes associated with this display object may have provided the clinician with a more salient indication that something was wrong with the heart. This new object was presented as a disfigured sphere that changed from mild to moderate to severe "crinkling," based on predefined thresholds of ST segment depression. In contrast, in the control condition, as in current clinical practice, ST segment depression was reflected by subtle changes in the electrocardiogram waveform and as a numerical value on the Datex AS/3 monitor. Both of these display attributes may be difficult to detect, especially when the clinician’s task load is large. It remains to be determined whether a more traditional display that depicted the ST segment changes with greater salience would produce similarly enhanced detection and treatment of MI.
In addition, subjects using the graphic display were more likely to administer earlier optimal treatment in the MI scenario compared with those using traditional display technology. The optimal treatment for this scenario was to administer nitroglycerin (oral or IV), but only after hypotension was treated (systolic blood pressure >80 mm Hg). Thus, recognition of hypovolemia and aggressive fluid administration were necessary to appropriately manage this event. In fact, the simulated patient spent less time in a state of hypotension when clinicians were using the graphic display. This finding suggests that the clinicians likely used other elements of the graphic CV display besides the "crinkled heart."
CVP was significantly closer to the patient’s initial value (CVP 0 mm Hg) at the end of the ISCH case when using the graphic display (0.6 mm Hg) compared with the CVP at the end of the case in the control group (5.6 mm Hg). The desired outcome of these results was to be as close to initial value to indicate that the patient was stable and very similar to the state that they were in before anesthetic intervention. Because of the fluid model used by the simulator, changes in CVP were indicative of fluid administration. These results suggest that fluid administration was used more often as a primary treatment in the control group whereas a combination of fluid and nitroglycerin was more common in the display condition. The SaO2 was slightly but significantly higher at the end of the case in the graphic display group suggesting better overall tissue perfusion with the therapeutic strategy chosen by these clinicians. Although the physiologic responses were driven by the simulator and validated by three experts, it remains a possibility that the simulator did not behave exactly as would a real patient in the same clinical situation.
In the AP scenario, no effects were evident from the use of the CV display. Changes in SVR and PVR could have helped in making this diagnosis but the results suggest that changes in these display elements were apparently not seen. These findings point to the need to redesign these elements of the new display to improve their salience. In addition, the AP scenario involved pulmonary variables that were not included in the CV graphic display. Thus, key diagnostic variables such as peak inspiratory pressure were presented identically in the two display conditions possibly reducing any display effect. This scenario was difficult and required the clinicians to monitor multiple CV and pulmonary variables to make a diagnosis. Thus, a single display element was not effectively communicated to the clinician by way of a clear emergent feature. Finally, the diagnosis in this scenario may have depended on examination of breath sounds in the simulator. Technical limitations associated with this physical simulation (e.g., sometimes difficult to hear) may have confounded any possible differences between experimental groups.
Although a number of investigators have evaluated anesthesia graphic displays, the present study is unique in its attempt to assess the utility of displays using high-fidelity simulation of clinically relevant tasks. The simulator was perceived to be realistic by the majority of subjects, suggesting that the simulation environment may be a good test bed for display evaluation, potentially increasing our knowledge about their value in a range of clinical applications and situations.
Because of the nature of the setup of the simulated OR and the location of the displays and video camera, it was impossible to blind the reviewers as to display condition. However, the naive reviewers were unaware of the hypotheses under test, and data scoring was based on rigorous a priori criteria.
Despite a significant effort to make the experience clinically realistic, some subjects might have behaved differently during the simulation, when compared with their normal practice. Despite being familiar with the simulator before the experiment, subjects may have had a variety of expectations and preconceptions of the simulation’s behavior (e.g., the simulator’s responses to drugs). Thus, generalized conclusions about the display’s positive or negative clinical impact cannot be made based on a single study, and one or more future evaluations in a clinical environment will be needed to assess actual clinical utility.
Participants in the graphic display group rated this display as more useful than did those who were exposed to the graphic display only during training. This suggests that the clinicians perceived added benefit of the display through increased exposure and applied use. However, there was no difference between the two groups in the questions that asked whether the graphic display would be a valuable addition to the OR.
The display was evaluated in an intensely monitored patient. The clinical scenarios were crafted to include use of invasive arterial and pulmonary artery catheters. In clinical situations in which invasive monitoring is not used, the utility of the current display’s design and layout is unknown. To achieve more general clinical utility, it is necessary to account for clinical situations when not all of the data required by the integrated display are available (19). Noninvasive continuous monitoring technologies (e.g., noninvasive CO) may provide data to support the current design. CV modeling might be another option to drive the display when isolated data elements are unavailable. Models might provide appropriate estimation of pressures and resistances (such as SVR and LAP) when the potential benefits of invasive measurement cannot be justified. Alternatively, future designs should address how these missing data will affect the display’s graphic presentation.
|
Conclusions
|
|---|
The present study provides support for the hypothesis that providing clinical information graphically in a display designed with emergent features and functional relationships can improve clinicians’ ability to detect, diagnosis, manage, and treat critical events in a simulated MI scenario. However, more studies must be conducted using a broader range of scenarios before any generalizable conclusions can be drawn. In addition, the study results must be qualified because of the known limitations of high-fidelity simulation studies. Nevertheless, the knowledge obtained from these types of technology assessment studies can be used, and the refinement of simulation-based test methods that have increased clinical relevance may lead to medical technologies that improve the quality of anesthesia care.
|
Appendix 1. Scenario: Anaphylaxis
|
|---|
|
Appendix 2. Scenario: Hypovolemia + Myocardial Ischemia
|
|---|
|
Appendix 3. NASA Task Load Index
|
|---|
|
Appendix 4. Questionnaire
|
|---|
|
Acknowledgments
|
|---|
This research was supported by National Institutes of Health Grant 1 RO1 HL64590.
We are grateful to Ken Johnson for his support in developing the scenarios.
|
References
|
|---|
- Runciman WB, Sellen A, Webb RK, et al. The Australian Incident Monitoring Study: errors, incidents and accidents in anaesthetic practice. Anaesth Intensive Care 1993; 21: 506–19.[ISI][Medline]
- Gaba DM, Maxwell M, DeAnda A. Anesthetic mishaps: breaking the chain of accident evolution. Anesthesiology 1987; 66: 670–6.[ISI][Medline]
- Cooper JB, Newbower RS, Long CD. Human error in anesthesia management: the quality of care in anesthesia. Springfield, IL: Thomas Books, 1980.
- Cooper JB, Newbower RS, Long CD, McPeek B. Preventable anesthesia mishaps: a study of human factors. Anesthesiology 1978; 49: 399–406.[ISI][Medline]
- Reason J. The contribution of latent human failures to the breakdown of complex systems. Philos Trans R Soc Lond B Biol Sci 1990; 327: 475–84.[ISI][Medline]
- Goodstein LP. Discriminative display support for process operators: human detection and diagnosis of system failure. New York: Plenum, 1981.
- Boff KR, Lincoln JE. Engineering Data Compendium: human perception and performance. Dayton, OH: Harry G. Armstrong Medical Research Laboratory, 1988.
- Vicente KJ, Christoffersen K, Pereklita A. Supporting operator problem solving through ecological interface design. IEEE Trans Syst Mass Cybernet 1995; 25: 529–45.
- Jungk A, Thull B, Hoeft A, Rau G. Evaluation of two new ecological interface approaches for the anesthesia workplace. J Clin Monit Comput 2000; 16: 243–58.[ISI][Medline]
- Jungk A, Thull B, Hoeft A, Rau G. Ergonomic evaluation of an ecological interface and a profilogram display for hemodynamic monitoring. J Clin Monit Comput 1999; 15: 469–79.[ISI][Medline]
- Effken JA, Kim NG, Shaw RE. Making the constraints visible: testing the ecological approach to interface design. Ergonomics 1997; 40: 1–27.[Medline]
- Vicente KJ. Ecological interface design: progress and challenges. Hum Factors 2002; 44: 62–78.[ISI][Medline]
- Ware C. Information visualization: perception for design. San Francisco: Morgan Kaufmann, 2000.
- Roscoe SN. Airborne displays for flight and navigation. Hum Factors 1968; 10: 321–32.[ISI]
- 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.[Medline]
- Triesman A. Preattentive processing in vision. Comput Vis Graphics Image Proc 1985; 31: 156–77.
- Agutter J, Syroid ND, Drews FA, et al. Case study: graphic data display for cardiovascular system. IEEE Symp Inform Vis 2001: 163–8.
- Wachter SB, Agutter J, Syroid N, et al. The employment of an iterative design process to develop a pulmonary graphical display. J Am Med Inform Assoc 2003; 10: 363–72.[Abstract/Free Full Text]
- Sharp T, Helmicki A. The application of the ecological interface design approach to neonatal intensive care medicine. Santa Monica, CA: HFES, 1998: 350–4.
Accepted for publication June 19, 2003.
Top
Abstract
Introduction
