Association of Vital Sign and Laboratory Abnormalities Detected via Remote Monitoring with ICU Readmission and Mortality: An Observational Cohort Study.

BACKGROUND

Post-intensive care unit (ICU) patients remain at high risk for adverse outcomes, and early identification of instability may improve outcomes and reduce costs. This study measures the association between eight vital sign and laboratory abnormalities ("flags") on surgical wards and the odds of ICU readmission or death in surgical patients after ICU transfer. Flags captured intuitive indications for ICU admission, including tachycardia, hypotension, hypoxemia, hyperkalemia, hyperglycemia, hypoglycemia, and anemia, which may be missed with infrequent ward evaluation.

METHODS

This single-center, observational cohort study included all adult surgical patients transferred from the ICU to a general care ward between May 2017 and March 2019 pre-COVID-19 pandemic. A mixed effects model evaluated the relationship between virtual flags and adverse discharge within six hours, adjusted for demographics and confounders. Modeling was inferential and not intended for diagnostic or screening prediction.

RESULTS

A total of 391 surgical ICU admissions were enrolled, comprising 138,623 rolling one-hour observation windows. After ICU discharge, 73 (18.7%) transfers died or were readmitted to the ICU. A flag on the wards was associated with 11.44-fold increased odds of ICU readmission or death within six hours of the flag (95% CI: 7.49-17.49) and 3.36-fold increased odds within 48 hours (95% CI: 2.23-5.05). A tachycardia flag raised the odds by 25.25 (95% CI: 13.34-47.82). Flags occurred on average every 2.6 hours per hundred ward patients monitored.

CONCLUSIONS

The presence of eight actionable vital sign and laboratory abnormalities which were detected in an automated system was strongly associated with ICU readmission or death after ICU transfer. Incorporating such monitoring on surgical wards could enable timely interventions, improve patient safety, and reduce mortality without alarm fatigue.