1. Title
Focus on Respiratory Rate: Improved reliability of measurements in acutely unwell
adults by using electronic sensors?
2. Authors
Ali Virk1, Naveed Khan1, Sophie Dorsey1, Anne Bonnici Mallia1, Mori Jones1, Elaine
Hardman1, Joanne Roche1, Chris Subbe2, Bernd Duller3
3. Institution
1
Ysbyty Gwynedd, Penrhosgarnedd, Bangor, UK
School of Medical Sciences, Bangor University, Bangor, UK
3
Philips Healthcare, Stuttgart, Germany
2
4. Introduction
Respiratory Rate is the strongest predictor of adverse outcomes such as cardiac arrests
or unscheduled critical care admission.
We aimed to establish whether electronic respiratory sensors
[a] would be able to monitor respiratory rate reliably in acutely unwell patients
[b] would lead to changes in the values observed
5. Methods
VITAL II is an interventional trial of an integrated monitoring system with automated
notification of deteriorating patients.
The Philips wireless IntelliVue CL Respiration Pod (Fig 1) is a sensor that uses 3D
accelerometer technology to measure respiration. An attachment for the sensor allows
placement to the patient’s chest for signal acquisition.
The setting: 28 bedded ward in a district general hospital with a mixture of gastroenterological and general medical patients. The majority of patients were emergency
admissions with acute illness.
Reproducibility study: Respiratory rate was measured simultaneously in an unselected
patient cohort over one minute by a trained clinician and the sensor. Measurements by
the sensor were every 12 seconds and averaged over one minute.
Frequency distribution during clinical application: We compared respiratory rates that
were manually obtained as part of routine surveillance of vital signs and respiratory
rates that were recorded from sensors in the same patient cohort during May 2014.
6. Results
Reproducibility study: We recorded 140 datasets in 31 patients (Mean age 66 years,
mean bodymass index 25). Respiratory rates measured by clinicians and by sensors had
a high level of agreement (linear correlation 0.91) with a mean of 15.8 by clinician and
15.5 by sensor. The average of differences (mean bias) was -0.31 rpm (SD for
differences 1.91 rpm, t=2,292, p=0,0229, t-test for paired samples).
Frequency distribution: We analysed data from 1513 manual measurements and 103
measurement by the sensor. Analysis of distribution showed clustering of respiratory
rates around the values of 16 (26%), 18 (33%), 20 (22%) in the manual cohort.
Frequencies in the cohort measured by the Sensor were 16 (5%), 18 (9%) and 20 (5%).
(Fig 2).
7. Discussion
Electronic sensors are able to measure respiratory rates reliably in acutely unwell
patients. Even in our environment where over 90% of sets of vital signs contain a
respiratory rate in emergencies or with staff under pressure a significant number of
values are possibly extrapolated or based on clinical impression rather than
measurement. Faulty measurements could underestimate the severity of acute
deterioration and delay escalation.
The impact on clinical outcomes requires further evaluation.
8. Conflicts of interest
The study was part of the VITAL II study, a clinical trial sponsored by Philips
Healthcare. CPS is PI of the VITAL II study, and has received speaker fees from
Philips. BD is a consultant for Philips.
Fig 1: Respiratory rate sensor with additional capability to record heart rate and posture
of a patient
Fig 2: Respiratory rate distribution in manual (top panel) and electronic (bottom panel)
acquisition in acutely unwell patients