The increased use of computer and home monitoring technologies requires caregivers to assess increasing amounts of clinical information. We report preliminary results from experiments designed to determine how information display format affects decision accuracy and time in the detection of infection/rejection events from physiologic and symptom data, drawn from participants in the Lung Transplant Home Monitoring Program (LTHMP) at the University of Minnesota.
In two experiments, physicians and nurses from transplant medicine and related specialties (N = 6 and 10, respectively) were shown a series of cases consisting of pulmonary and symptom information from the LTHMP databases (C = 12 and 20, respectively, divided equally between infection/rejection events and nonevents). In both experiments, cases were shown, in random order, in each of several display formats. Experiment 1 included paper- and computer screen-based graphical and tabular displays. Experiment 2 included computer screen-based graphical, table, hybrid graphical/table, and control chart formats. Clinician readers provided probability statements of infection/rejection status for each display, which were used in constructing receiver operating characteristic (ROC) curves. Times required for making decisions and clinician preference rankings for the different display formats were also recorded.
No statistically significant differences were found in decision accuracies for any of the display formats (Area-under-ROC ranges: 0.725-0.783, P > 0.5; 0.704-0.753, P > 0.5 for Experiments 1 and 2, respectively). Decision times were not significantly different at the 0.05 level (ranges: 20.1-26.8 seconds, P = 0.12; 33.8-40.5 seconds, P = 0.10, respectively). Readers overwhelmingly preferred the graphical display formats over the tables.CONCLUSIONS AND CLINICAL IMPLICATIONS: We conclude that screen-based, graphical displays of transplant monitoring information would be efficacious, well-accepted tools in clinical practice. Further research is being conducted to explore the effects of interactivity (i.e., reader ability to manipulate views of data) on this clinical decision-making task.
D.S. Pieczkiewicz, None.