(1352) Evaluation of Remote Early Warning Score Derived From Wristband-Type Sensors in Intensive Care Unit

Introduction: Activation of the rapid response system depends on changes in vital signs or scores based on vital signs. However, bedside monitors are not feasible for all hospitalized patients. A wearable device was recently utilized to capture changes in vital signs. This study investigated the performance of a prototype wristband-type sensor developed by Nihon Kohden for detecting patients in severe clinical conditions by remote early warning score: REWS, which is calculated using heart rate: HR and respiratory rate: RR (van der Stam JA et al., European Journal of Surgical Oncology, 2023).

Methods: Among the patients admitted to the intensive care unit with bedside monitors between March 2020 and August 2020, this study recruited those who agreed to wear the wristband sensor. We excluded the patients on the mechanical ventilation for this analysis. We calculated REWS from HR, RR values measured by the wristband sensor, and National Early Warning Score: NEWS from four vital sign values (HR, RR, SpO2, and Blood Pressure) measured by the bedside monitor without consciousness level, supplemental oxygen, and temperature every 1 minute. We defined the time points with NEWS >= 7 as severe clinical conditions and evaluated the performance of REWS to detect them.

Results: A total of 24 cases (13 postoperative,１sepsis, ７stroke, and ３others)　and 9,367-minute measurements were included in the analysis. The mean patient age was 66.8±16.7 years; 14 patients were male. In the case of NEWS >= 7 (n = 55) and NEWS < 7 (n = 9,312), the median REWS values were 1 (IQR: 1-2) and 1 (IQR: 0-1), respectively. The AUC for REWS to detect NEWS >= 7 was 0.73. The performance of REWS at 2 was sensitivity: 0.436, specificity: 0.937, positive predictive value: 0.039 and negative predictive value: 0.996.

Conclusions: The wristband sensor-derived REWS using HR and RR could capture the severe critical conditions defined by the bedside monitor-derived NEWS. The wristband sensor could enable the early automatic detection of a ward patient's condition deterioration without manual spot vital sign measurements.