| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *Applied Medical Visualizations LLC; Departments of Architecture and Anesthesiology, University of Utah, Salt Lake City, Utah; and Department of Anesthesiology, University of Arizona, Tucson, Arizona.
Address correspondence and reprint requests to Robert Albert, MA, Applied Medical Visualizations, LLC, 925 East 900 South, Salt Lake City, UT 84105. Address e-mail to ralbert{at}medvis.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
METHODS: Sixteen anesthesiologists and anesthesia residents were enrolled in a simulation-based evaluation of the GCD. Participants were randomly split into two groups balanced for expertise and asked to manage six simulated adverse cardiopulmonary events. The GCD was present in half of the simulations, balanced across scenarios and groups. Participants’ verbalizations and actions during each scenario were recorded and transcribed. Transcripts of treatment interventions were subsequently rated by two blinded expert anesthesiologists. Perceived workload, time to detection, and proper treatment of the adverse event were compared between groups.
RESULTS: Experts ranked anesthesiologists using the GCD as being more effective overall and individually in three of six scenarios. Use of the GCD was demonstrated to influence the time to detection and the time to treatment of some critical events. There were no workload differences between display groups.
DISCUSSION: Treatment intervention by participants using the GCD was rated superior by two blinded experts. The presence of the GCD resulted in a modest improvement in the time to detect myocardial ischemia and increased pulmonary capillary wedge pressure. The GCD may be a useful adjunct to monitor patients during adverse cardiopulmonary events.
|
Introduction |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
We have developed and evaluated a graphic cardiovascular display (GCD) (Fig. 1) designed to improve situational awareness when managing adverse cardiopulmonary events (4). The GCD provides an animated graphical presentation in real time of Spo2, vascular pressures, stroke volume, vascular resistances, and ST segment depression. The intent of this design is to 1) more efficiently convey information about cardiovascular function, 2) improve detection of adverse cardiopulmonary events, 3) improve the ability to track effectiveness of treatment interventions, and 4) minimize cognitive workload.
|
One potentially advantageous feature of the GCD is that it presents abnormal values of cardiovascular variables in a manner that may be easier to detect than conventional modes of data presentation (5–7). When cardiovascular variables are within normal limits, the GCD appears symmetrical and balanced (Fig. 1A). When variables deviate from the normal range, the GCD becomes asymmetric, orienting the user’s attention to the physiologic variable(s) that are beyond a normal range (Fig. 1B) (8,9). Graphic changes in the display have been appropriately scaled to the functional limits of the human visual perceptual system with respect to clinically meaningful changes of a given variable (7).
In a prior simulation-based study by Agutter et al. (4), volunteer clinicians managed two adverse cardiovascular events: 1) severe intraoperative myocardial ischemia and 2) life-threatening anaphylaxis in response to intravenous therapy administration of an incompatible blood type. Results from this preliminary evaluation demonstrate that the GCD reduced the time to diagnose and treat myocardial ischemia by 2 and 2.5 min respectively.
There were several limitations in the previous evaluation of the GCD: 1) The GCD was evaluated using two scenarios. 2) Investigators recording the volunteer’s actions were not blinded to the presence or absence of the GCD. 3) Treatments of the simulated conditions were assessed according to rigid a priori criteria. This may have biased the results since more than one therapeutic strategy was appropriate. 4) The GCD required use of a pulmonary artery catheter (PAC) to obtain the central venous pressure (CVP), pulmonary capillary wedge pressure (PCWP), cardiac index (CI), and systemic vascular resistance (SVR) values. A PAC is not part of routine monitoring for most anesthetics.
The purposes of this study were: 1) to address the limitations from our previous investigation in an effort to explore how the GCD will perform over a wider range of adverse cardiopulmonary events; 2) to use blinded post hoc expert evaluations of treatment interventions to allow for a variety of clinically acceptable therapeutic strategies by participants; and 3) to broaden the applicability of the GCD by presenting a full display with PAC data and an abbreviated display without PAC data (Fig. 1C).
Our hypotheses were: 1) clinicians using the GCD would have superior treatment interventions and detection times compared with clinicians using conventional monitoring equipment when managing simulated adverse cardiopulmonary events and 2) that the use of the GCD would not increase clinician’s perceived workload when managing these adverse events.
|
METHODS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
Stimuli and Apparatus
Two monitors were placed at eye level on top of the anesthesia machine. The right monitor, a Datex vital signs monitor (AS/3, Helsinki, Finland) displayed the electrocardiograph, arterial blood pressure (BP), pulse oximeter, and capnograph waveforms and discrete values for heart rate, arterial BP, oxygen saturation, end-tidal carbon dioxide, and fraction of inspired oxygen. All standard alarms were enabled and set to factory default values. On the left, a 17 in. computer monitor displayed the GCD.
A full body patient simulator (HPS version 5.55, METI, Sarasota, FL), a computer interface, and patient monitors (Fig. 2) were used to present participants with six adverse cardiopulmonary events shown in Table 1 (Appendix). Hemodynamic information including CVP, PCWP, stroke volume, hemoglobin saturation, and heart rate was obtained from the patient simulator and was used to drive the GCD. With this information, standard calculations were used to derive CI (L · min–1 · m–2), and SVR and pulmonary vascular resistance (dynes · s–1 · cm–5).
|
|
A modified GCD was developed that does not use PAC information. In this abbreviated GCD, a noninvasive measure of CI is assumed to be available and used to estimate SVR. Selected adverse event scenarios used the full (expanded) GCD and others the abbreviated GCD, as indicated in Table 1. The computer monitor presenting GCD information operated in one of two modes. One mode presented only numeric information; the other mode presented numeric information in concert with the graphical representation. Regardless of mode, all participants received the same hemodynamic information with the availability of the GCD manipulated as an independent variable.
Experimental Design
Participants were randomly assigned to one of two groups balanced with equal numbers of faculty and residents. Using a randomized, counterbalanced design, each group was presented three simulations where only numeric values were provided and three simulations where the numeric values and the GCD were provided. Participants were additionally yoked into pairs, such that the order of scenario presentation was the same for both participants in a given pair with a counterbalanced presentation of the GCD or numeric information. This control allowed each scenario to be presented with and without the graphic information within each of the eight yoked pairs.
Procedure
Participants completed a questionnaire soliciting experience level, age, and prior familiarity with the GCD. The participants were oriented to the patient simulator with a 10-min scripted overview of the simulated operating room and equipment. Participants were then shown a 15-min training video that introduced them to the GCD, explaining the individual elements of the GCD and the corresponding physiologic information that it represented.
Before each scenario, participants were able to review the simulated patient’s history and anesthetic chart for 1 min. Participants then assumed care of the simulated patient.
During each scenario, participants wore a microphone and were encouraged to verbally indicate their observations, thoughts, diagnoses, and treatments. Three trained observers independently recorded the time each participant verbalized a diagnosis or began a therapeutic intervention. At the conclusion of each scenario, participants rated their perceived workload by completing the National Aeronautical Space Administration (NASA) Task Load Index (TLX) 10 and answered a question on the simulation’s realism. During this time, observers compared their records and prepared a final transcript of events. The video tape, recorded during the scenario, was reviewed in the event of a discrepancy among the observers.
Data Analysis
P < 0.05 was considered statistically significant for all analyses.
Expert Ranking of Participant Treatment Intervention
Two expert raters, blinded to participant and use of the GCD, independently rank ordered all transcripts of treatment interventions in each scenario from best to worst in terms of the treatment’s accuracy, timeliness, and quality. Raters were board certified anesthesiologists with specialties in cardiovascular anesthesia and transesophageal echocardiography. For each scenario, transcripts of the patient’s history, anesthetic record, surgery, adverse event description, as well as the time and details of participants’ treatment interventions were compiled for review by the raters. Ties in participant rankings were assigned the same average rank order. Treatment rankings were analyzed for interrater reliability using the Pearson R correlation statistic. Treatment ranking differences due to display condition were analyzed with the Freidman test (11).
Measurement of Detection and Treatment Times
Detection time was defined as beginning when the participant entered the simulated operating room and ending when the participant correctly verbalized the adverse event. Treatment time was defined as beginning when the participant entered the simulated operating room and ending when proper therapy was initiated to adequately treat the adverse event. Therapies for which the simulated patient would respond positively were determined a priori by informal consensus of two anesthesiologist investigators who designed the scenarios. When an adverse event was not verbalized during the allotted simulation time, the maximum duration time of the scenario was used for analysis. Detection and treatment time analyses were conducted with a 2 x 2 analysis of variance using presence of the GCD, and expertise (defined as faculty versus resident) as independent variables.
Assessment of Perceived Workload and Simulator Realism
Data from the NASA-TLX workload assessment were analyzed using repeated measures analysis of variance with respect to GCD condition for each scenario. All scenarios were rated by participants for realism on a 10-point likert scale (from 0 to 9).
|
RESULTS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
Expert Ranking of Participant Performance
The Friedman rank order analysis indicated treatment therapy was significantly improved when using the GCD overall and individually for the hypertension from inadequate analgesia, myocardial ischemia, and left ventricular failure scenarios (Fig. 3 and Table 2). No significant difference was found for treatment rankings from the hemorrhagic hypovolemia or the adult respiratory distress syndrome scenarios. The performance rankings between the two expert raters were highly correlated and had a significant interrater reliability (Table 3).
|
|
|
Measurement of Detection and Treatment Times
Detection and treatment times for each scenario are listed in Table 4. Using the GCD, participants detected and treated myocardial ischemia faster than those who did not use the GCD. In the left ventricular failure scenario, high PCWPs were detected faster using the GCD. No other times were significantly different. No differences in detection or treatment times were found as a function of expertise.
|
Assessment of Workload and Simulator Realism
Table 5 lists the NASA TLX scores for each scenario. There was no significant effect of the GCD on self-assessed workload as a function of expertise or display group. Participants rated the overall realism of the 6 scenarios with a median score of 5.5 (range, 1–8). There were no differences in ratings with regard to level of expertise.
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
Treatment rankings for participants who managed these events with the abbreviated GCD were better than those without the GCD in two of three scenarios (hypertension due to inadequate analgesia and myocardial ischemia). The abbreviated display was designed to function in clinical settings where a PAC was not available but would still require central line access to measure CVP and a noninvasive measure of CI. These results suggest that the abbreviated GCD may be as effective in communicating selected abnormalities in cardiovascular function as the complete GCD.
The results of this study indicate that use of the GCD shortened the time to detect and treat myocardial ischemia with underlying hypertension. These findings support the results of the previous evaluation of the GCD by Agutter et al. (4) in part, where the time to detection and treatment of myocardial ischemia was reduced with the use of the GCD. Since the GCD provides an integrated presentation of hemodynamic information in context with a salient presentation of ST segment changes, GCD users may have had more confidence in their diagnosis and treatment plan.
In the adult respiratory distress syndrome scenario, there was no improvement in detection time, treatment ranking, or treatment time using the GCD. The null finding from this scenario may be explained in part by two observations. First, the clinicians may have focused primarily on detecting, diagnosing, and treating the compromised pulmonary state of the simulated patient instead of the GCD’s hemodynamic information. Second, not all participants verbalized detection of myocardial ischemia, resulting in a small sample size for analysis. Six of eight GCD participants reported continuing ST segment depression of the simulated patient’s electrocardiogram, compared with four of eight participants without the GCD. It is unknown for the remaining participants if and when they detected myocardial ischemia.
Detection, treatment time, and treatment ranking of hypovolemia did not improve in the presence of the GCD. The hypovolemia event simulated profound fluid depletion and the numeric values for CVP and mean arterial BP changed enough to orient participants to the presence of hemorrhagic hypovolemia with or without the GCD. If the participants are adept at detecting volume status using traditional numeric information, then graphic presentation of the information may not further enhance performance.
During the left ventricular failure scenario, use of the GCD shortened the time to detect high PCWP, and the GCD users were ranked better for treatment therapy. The GCD may have allowed participants to more readily confirm their suspicion that PCWP was increased and subsequently initiated an effective therapy. Yet, use of the display did not necessarily decrease the time required to start an appropriate treatment intervention.
In the hypertension due to inadequate analgesia scenario, detection times were not different for GCD users, but they were ranked as more effective in their treatment intervention. Inadequate analgesia and mild hypertension are common problems encountered in an operating room environment, and participants were quick to diagnose and treat this event, since hypertension is easy to recognize using conventional monitors. The GCD’s representation of mean arterial BP may have helped clinicians target a more effective treatment.
With the GCD, participants were provided with a different modality of presentation for hemodynamic information; however, there was no significant impact of workload with the GCD, as observed in the NASA-TLX workload results. This finding is of interest, since anesthesiologists are susceptible to overload, given the many data sources in the clinical environment.
This study supports prior research investigating the utility of graphically oriented hemodynamic (12) and pulmonary displays (13,14). Investigations using a visual presentation of pharmacologic information during anesthetic cases have shown improvements in clinical performance (15,16).
Participants rated the simulations as adequate representations of typical anesthetics during surgery, confirming adequate face validity for the scenarios. Two anesthesiologists, with extensive patient simulation experience from two institutions, were responsible for scenario design and content, yet a more in-depth validation of the scenarios would have been desirable. One approach to achieving better validity of each scenario would have been to use a modified Delphi process to reach consensus by multiple experts (17), which has been demonstrated in previous evaluations of simulated scenarios (18,19). One concern with using the Delphi methodology for this investigation was the large amount of resources required to establish consensus on reliability and validity for six complicated scenarios.
Limitations to Clinical Relevance
Although the results of this study indicate that an abbreviated GCD may be useful in anesthetics with less invasive monitoring (no PAC), the abbreviated presentation still requires a central venous catheter and a measure of CI. Noninvasive technologies, such as cardiac output monitoring based on partial rebreathing of CO2, may be practical to provide near real-time information to the GCD. However, there is some debate on its reliability and accuracy, depending on the type of surgery and the ventilation conditions (20–26). A second limitation is that the information provided to the GCD in the simulated scenarios was ideal and updated approximately once every second. This is critically important for data like PCWP, which are updated automatically to the GCD by the simulation script, but would only be available intermittently when measured in the clinical environment. In the clinical environment, some information may also be provided at slower intervals, which may prolong the time to diagnosis and intervention.
Study Limitations
There are some limitations associated with this simulation-based investigation. With a talk-aloud protocol in a stressful scenario, participants may neglect to state a differential diagnosis before proceeding with experimental treatments. A desired measure in this study would have been the accuracy of a differential diagnosis, as ranked by the blinded experts. An approach to eliciting a reliable diagnosis would be for the investigator to briefly pause the scenario and explicitly query the participant. This procedure was not implemented in this study in preference to maintaining some clinical reality. Second, although face validity of the simulated environment was deemed adequate, individual scenarios may have had unrealistic elements, causing doubt among participants. Third, the time course of the simulated adverse events may have been different than what might be typically observed in a clinical setting (27). Fourth, participants received only 15 min of instruction on how to use and interpret information from the GCD.
The ranking analysis of treatment intervention by the two clinical experts has limitations. Raters had a limited set of information describing the time and details of participants’ treatment interventions to determine relative rankings among the participants. Thus, raters sometimes grouped participant’s treatment rankings, resulting in many ties (Fig. 3). The ranking process was subjective, requiring raters to judge the treatment interventions based on their own clinical expertise. A more rigorous method would have raters use a validated a priori checklist of objective criteria.
|
CONCLUSION |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
APPENDIX: SCENARIO DESCRIPTIONS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
Scenario one simulated a 52-yr-old man undergoing a repair of a left distal tibia and fibula fracture sustained in a climbing accident. A central line catheter was placed to optimize fluid resuscitation. A tourniquet was placed in the lower left extremity a little over an hour before the participant assumed the case. Without any treatment, there was a small increase in arterial blood pressure and heart rate to 145/90 and 95, respectively, in response to prolonged tourniquet time. The vital signs abated with deepening anesthetic either with additional opioid or inhaled anesthetic. If left untreated, his arterial blood pressure and heart rate continued to increase. The scenario duration was 5 min.
Scenario 2: Myocardial Ischemia
Scenario two simulated a 71-yr-old man with peripheral vascular disease, hypertension, and diabetes undergoing a Fem-Pop bypass on the left lower extremity. Without any treatment intervention, his arterial blood pressure (BP) increased from 130/85 to 205/160 in the first 2.5 min. Depending on the severity of the untreated change in BP, associated ST segment changes consistent with an evolving myocardial infarction progressed in the following manner: systolic BP <140 = no ST segment change, systolic BP >140 but <160 = mild ST segment depression (approximately 1 mm), systolic BP >160 but <175 = moderate ST segment depression (approximately 2 mm), and systolic BP >175 = severe ST segment depression (approximately 3 mm). The simulated patient’s vital signs returned to the scenario’s baseline values if a moderate administration of nitroglycerin (bolus: 0.5–3 µg/kg or infusion 0.4–1.5 µg · kg–1 · min–1 or 0.2–0.39 µg/kg bolus, followed by an infusion of 0.2–1 µg · kg–1 · min–1) or sodium nitroprusside (05–2 µg · kg–1 · min–1) or substantial increase in the patient’s depth of anesthesia in combination with 100% Fio2 or ß blockers or a slight increase in the patient’s depth of anesthesia. If left untreated the patient simulated a myocardial infarction. The scenario duration was 7 min.
Scenario 3: Left Ventricular Failure
Scenario three simulated a 68-yr-old woman with history of myocardial infarction 10 yr ago with a subsequent coronary stent placement in proximal left anterior descending. She has been smoking 1/2 pack a day for 50 yr. She’s able to do daily living tasks, but is short of breath after climbing two flights of stairs. She was involved in a motor vehicle accident. She has a right femoral midshaft fracture and a left distal ulnar fracture and is to undergo an emergent open reduction internal fixation fracture repair. A Swan-Ganz catheter has been placed for intraoperative and postoperative management. The starting vital signs were: arterial blood pressure (BP) = 105/70, heart rate (HR) 95, pulmonary capillary wedge pressure (PCWP) = 14, central venous pressure (CVP) = 8, SaO2 of 94, cardiac index (CI) = 2 L · min–1 · m–2, systemic vascular resistance (SVR) = 1700. Over the course of the first 2.5 min, vital signs changed to: a decreased BP (80/40), CI (1.1) and SaO2 down to 90 with increased HR (120), PCWP (20), and CVP (16). If left untreated, the patient experienced a complete left ventricular failure. Vital signs returned to the scenario’s baseline values upon administration of one of the following: dobutamine (6–20 µg · kg–1 · min–1), dopamine (9–20 µg · kg–1 · min–1), milrinone (0.5–0.75 µg · kg–1 · min–1 or loading dose of 50–75 µg with infusion of 0.25–0.5 µg · kg–1 · min–1), or epinepherine (0.05–0.1 µg · kg–1 · min–1 or loading dose 20 µg with infusion of 0.02–0.05 µg · kg–1 · min–1). The scenario duration was 7 min.
Scenario 4: Hemorrhagic Hypovolemia
Scenario four simulated a 40-yr-old man, 5 feet 10 in., 235 lbs involved in a construction site accident when a large cement block fell and pinned him to the ground. He suffered a pelvic fracture with multiple left rib fractures and a pulmonary contusion. A computed tomography scan indicated a retrograde peritoneal hemotoma. He underwent an exploratory laparatomy. Approximately 3 L of crystalloid have been administered. The starting vital signs were: BP (110/70), HR (105), CVP (4), SaO2(96), CI (3), SVR (1200), spontaneous respiratory rate (14), temperature (35.1°C). Over the course of the first 2.5 min, his vital signs changed to BP (70/35), HR (130), SVR (2000), CI (1.5), and CVP (0). If left untreated, the patient experienced extreme severe hypotension. The simulated patient’s vital signs returned to the scenario’s baseline values upon administration of 
2 U of O– packed red blood cells and 1.5 L of crystalloid. The scenario duration was 9 min.
Scenario 5: Sepsis as a Consequence of Gut Ischemia
Scenario five simulated a 79-yr-old woman who underwent an abdominal aortic aneurysm repair 6 days ago. She had remained in the intensive care unit since then. She has a history of alcoholism, poor liver function, and a coagulopathy with a prothrombin time of 18.1. She is a brittle insulin-dependent diabetic and received 8 U of insulin that morning. Her last blood glucose reading was 350. She presented with hypotension and abdominal distention. The starting vitals were: BP (100/70), HR (120), CVP (5), PCWP (10), SaO2(96), CI (4), SVR (700), spontaneous respiratory rate (14). In the first 2.5 min, the vital signs became: BP (80/50), HR (140), SVR (350), CI (8), CVP (2), PCWP (6). If left untreated, the patient experienced sepsis as a consequence of gut ischemia. Vital signs returned to the scenario’s baseline values with 2 of the following treatments: 1 L of crystalloid, norepinephrine drip (>5 µg/min), cumulative phenylepherine boluses >200 µg. The scenario duration was 7 min.
Scenario 6: Acute Respiratory Distress Syndrome, Accompanied by Myocardial Ischemia
Scenario six simulated a 65-yr-old veteran with a 50-year, pack-a-day history of smoking and of heavy alcohol use. He takes Plavix for history of coronary artery disease. The patient fell and broke his right hip at home. He has been complaining of pain and was in significant distress before anesthetic induction. He has an obvious deformity in the right lower extremity. Radiograph results show a displaced right femoral neck fracture. He denies any other injuries, but there is a contusion on the chest wall. The starting vital signs were: BP (105/65), HR (55), CVP (5), SaO2(97), CI (4), SVR (1200), lung sounds including crackles or rales, indicated high lung compliance. In the first 90 s, the SaO2 decreased to 90. In the next 30 s, the SaO2 decreased to 86 with mild ST segment depression. By 2.5 min, his BP was down to 100/60, SaO2 down to 80, CI down to 3, with moderate ST segment depression. If left untreated, the patient experienced hypoxia that caused moderate ischemia due to onset and worsening of adult respiratory distress syndrome. The simulated patient’s vital signs returned to the scenario’s baseline values with administration of 100% Fio2 and positive-end-expiratory-pressure. The scenario duration was 7 min.
|
Footnotes |
|---|
Supported, in part, by General Electric Healthcare, Applied Medical Visualizations L.L.C., and NIH RO1 EB00294.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONCONCLUSIONAPPENDIX: SCENARIO DESCRIPTIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
