Optimising the identification of acute deterioration and sepsis through digital technology

Abstract

Sepsis is a global health concern with a high mortality. For every hour’s delay in sepsis treatment, mortality increases by 7.6%. Vital sign changes are used to detect alterations in a patient’s physiology and detect deterioration. Novel wearable sensors, which measure vital signs continuously, may identify unwell patients and sepsis sooner. We aimed to determine if such sensors could reliably identify unwell patients and those with sepsis sooner.

A sensor was selected which measures multiple vital signs simultaneously. The sensor measures heart rate, respiratory rate and temperature every 2 minutes. Sensor and nursing ward data were compared, patient data and staff feedback was collected.

A 500-patient study is complete with admitted patients wearing the sensor in additional to standard clinical monitoring. Comparisons between sensor readings and ward observations taken by nurses were made. A comparison of sensor time windows for alert generation was made and the 10 minute window selected. The correlation coefficient between Sensium data and ward observation data for Heart Rate (HR) was 0.79 (p=0.0001) and a mean difference of less than 3 beats per minute. The correlation coefficient between Sensium data and ward observation data for Respiratory Rate (RR) was 0.3 (p=0.0001) and a mean difference of less than 2 breaths per minute. The correlation coefficient between Sensium data and ward observation data for temperature was 0.43 (p=0.0001) and a mean difference of less than 0.5° C. In 82% of cases of HR and 81% of RR the sensor was faster than ward observations in time to detection. Patient feedback was positive (453 questionnaires analysed) with 86% agreeing to wear the sensor again and 85% agreeing it was comfortable. Staff groups were interviewed; feedback was positive with problems in current monitoring highlighted.

Wearable sensors may cause a paradigm shift in future patient monitoring with safer, smarter wards.