Early detection of respiratory deterioration using continuous respiratory rate monitoring

<bold>Background:</bold> Respirasense is a wearable patch which continuously measures respiratory rate by detecting chest wall movement. We aimed to assess its accuracy, and potential for earlier detection of respiratory deterioration. <bold>Methods:</bold> Respirasense was installed on the Respiratory Assessment Unit at Nottingham University Hospitals NHS Trust. Ward nurses were free to choose which patients required monitoring. Manual and continuous respiratory rate measurements (mRR and cRR respectively) were extracted from 30^th November 2022 to 22^nd July 2023. The study was approved by the Wales Research Ethics Committee 6 (23/WA/0160). <bold>Results:</bold> The dataset included 403 admission episodes, with a total of 2606 synchronous mRR and cRR measurements. 298 respiratory deterioration events occurred, defined as a rise in manually counted respiratory rate by at least 4 breaths per minute between two consecutive measurements, with the second measurement being at least 21 breaths per minute. The mean bias of cRR with respect to mRR was 1.42 breaths/minute, and the 95% limits of agreement were -9.51 to 12.36 breaths/minute. The average difference between cRR and the most recent mRR measurement (cRR-mRR) gradually increased in the run-up to respiratory deterioration events. An alarm threshold of cRR-mRR ≥ 4 had a sensitivity of 79.9% and a specificity of 58.3% for the early detection of respiratory deterioration. The area under the receiver operating characteristic curve was 0.756. Early detection of respiratory deterioration occurred on average 1.7 hours prior to the next mRR measurement. <bold>Conclusion:</bold> Continuous respiratory rate monitoring may lead to earlier detection of physiological deterioration on general respiratory wards.