Informatics in Medicine Unlocked (Jan 2020)
Comparison of health record vitals and continuously acquired vitals data identifies key differences in clinical impression
Abstract
This study aimed to investigate how intermittent ICU data documented by nursing in the electronic medical record (EMR) corresponds to that obtained from continuous data acquisition (CDA) and the differences in perception by medical providers. Both descriptive statistics and impact on decision-making were compared. We performed a secondary analysis of de-identified physiologic data collected as part of a study exploring the impact of dexmedetomidine adjunctive sedation on ICP and primary sedation dose (C3PO study). Data was in two sets, one derived from nurse charting in the health record, the other was continuous digital acquisition. We compared the number and timing of clinical events within each dataset that were outside the clinical treatment goals for that parameter. We also made graphs of concurrent 3-h periods from the two datasets and surveyed providers about their impressions and the need for intervention based on the visual review of the trends. Data consisted of 74,143 measures of ICP compiled from 40 neurocritically ill patients. We found no significant difference in overall means. The CDA dataset captured significantly more ICP elevation events defined as ICP >20 mmHg. Survey of ICU providers revealed that CDA data segments were perceived as “less controlled” and were more likely to be considered as “not meeting goals” for the vital parameter. We conclude that there are no significant differences in mean ICP measures between EMR and CDA vitals data, however, EMR data underreported the number of ICP elevation events and may not represent the overall patient exposure to out of goal vitals measures. We conclude that potentially significant differences exist between EMR and CDA data for ICU patients and that further consideration of this is needed in ICU trial design and as more machine learning and surveillance algorithms are developed for patient monitoring and predictions of deterioration.
