A graphical object display improves anesthesiologists' performance on a simulated diagnostic task
G T Blike1, S D Surgenor, K Whalen
1Department of Anesthesiology, Dartmouth-Hithcock Medical Center, Lebanon, NH 03756-0001, USA. George.Blike@Hitchcock.org
Objective:
This study tests the hypothesis that a graphical object display (a data display consisting of meaningful shapes) will affect the ability of anesthesiologists to perform a diagnostic task rapidly and correctly. The diagnostic tasks studied were recognition and differentiation of five etiologies of shock--anaphylaxis, bradycardia, myocardial ischemia, hypovolemia, pulmonary embolus.
Methods:
Data sets consisting of HR, Systemic Arterial BP, Pulmonary Arterial BP, CVP, and Cardiac Output were generated for five shock states and five non-shock states. The resulting 10 data sets were presented on a computer monitor to study subjects twice (first in an alpha-numeric format and then in the object format) for a total of twenty decision screens. Subjects used soft-buttons on a computer touch-screen monitor to: a) advance to the next display; b) differentiate a non-shock state from a shock state; and, c) select the etiology of shock state represented by the display (Figure 2). Data collection was automatic, using the internal clock and memory of the computer.
Results:
Eleven anesthesiologists participated in this study. They completed a total of 3060 diagnostic decisions, half with each display format. Performance measures were time to decision and diagnostic accuracy. The object display improved no-shock recognition by 1.0 second and shock etiology determination by 1.4 seconds (p < 0.05). The object display also significantly improved accuracy for shock recognition by 1.4% and etiology determination by 4.1% (p < 0.05). Testing was completed in a time interval of <45 min per 10 trials.
Conclusions:
The primary finding of this study was that anesthesiologists using the object display format committed significantly fewer diagnostic errors when interpreting physiologic data. In addition, both the recognition of no-shock and the diagnosis of shock etiology were completed more rapidly when the object display was used. The major limitation of this initial trial is the simplicity of the test. Future investigation of the impact of the display on clinical decision making will require more realistic clinical scenarios with partial or full simulation to better understand the potential clinical impact.
