Acoustic sensor versus electrocardiographically derived respiratory rate in unstable trauma patients

Abstract

Respiratory rate (RR) is important in many patient care settings; however, direct observation of RR is cumbersome and often inaccurate, and electrocardiogram-derived RR (RR_ECG) is unreliable. We asked how data derived from the first 15 min of RR recording after trauma center admission using a novel acoustic sensor (RR_a) would compare to RR_ECG and to end-tidal carbon dioxide-based RR ([Formula: see text]) from intubated patients, the "gold standard" in predicting life-saving interventions in unstable trauma patients. In a convenience sample subset of trauma patients admitted to our Level 1 trauma center, enrolled in the ONPOINT study, and monitored with RR_ECG, some of whom also had [Formula: see text] data, we collected RRa using an adhesive sensor with an integrated acoustic transducer (Masimo RRa™). Using Bland-Altman analysis of area under the receiver operating characteristic (AUROC) curves, we compared the first 15 min of continuous RRa and RR_ECG to [Formula: see text] and assessed the performance of these three parameters compared to the Revised Trauma Score (RTS) in predicting blood transfusion 3, 6, and 12 h after admission. Of the 1200 patients enrolled in ONPOINT from December 2011 to May 2013, 1191 had RR_ECG data recorded in the first 15 min, 358 had acoustic monitoring, and 14 of the latter also had [Formula: see text]. The three groups did not differ demographically or in mechanism of injury. RR_a showed less bias (0.8 vs. 6.9) and better agreement than RR_ECG when compared to [Formula: see text]. At [Formula: see text] 10-29 breaths per minute, RR_a was more likely to be the same as [Formula: see text] and assign the same RTS. In predicting transfusion, features derived from RR_a and RR_ECG gave AUROCs 0.59-0.66 but with true positive rate 0.70-0.89. RR_a monitoring is a non-invasive option to glean valid RR data to assist clinical decision making and could contribute to prediction models in non-intubated unstable trauma patients.

Keywords: Acoustic sensor; Monitoring; Respiratory rate; Transfusion prediction; Trauma.