Early recognition of deterioration in high-risk patients using continuous remote wireless monitoring: a clinical observation study with the Checkpoint Cardio system

High-risk hospitalized patients * either admitted for acute medical conditions
or recovering from major surgery - are at risk for postoperative complications
that could result in prolonged hospitalization, hospital readmissions, injury
or even death (Sweeney, 2013). Observational studies suggest that unexpected
in-hospital ICU admissions or ICU readmissions, cardiopulmonary arrests and
death are usually preceded by changes in vital signs 6 to 24 hours prior to
these adverse events (Schein, Hazday, Pena, Ruben, & Sprung, 1990). Obvious
clinical indicators of deterioration, such as changes in respiratory rate or
heart rate are often overlooked and not detected in time (Goldhill, 2004).
Known problems are a low nurse-to-patient ratio and infrequent vital signs
monitoring. On general wards the current standard is intermittent manual
measurement of vital signs, typically only once every nurse shift (every 8 h).
In contrast, in high-care settings such as the Intensive Care unit, it is very
unusual for acute patient deterioration to be missed, because nurse staffing
ratio*s are much higher (1:1, 1:2) and vital signs are continuously monitored
using conventional bedside wired monitors. After hospital discharge, vital
signs are no longer monitored at all, and the likelihood of delayed recognition
of vital instability increases again. Although the risk of sudden patient
deterioration typically decreases towards the day of hospital discharge and
beyond, the time span that patient deterioration goes unnoticed increases at
each successive care transition from ICU to home..

In an attempt to address the issue of of critical illness and death resulting
from missed physiological deterioration, many hospitals have no implemented
early warning scoring (EWS) on regular patient wards. In the Netherlands, using
EWS is mandatory since 2007. The EWS is designed to alert first-line caregivers
to impending adverse events and trigger them to call for help (Bokhari et al.,
2010). Despite the potential benefits of early warning scores, many
deteriorating patients are still not recognized in time, resulting in
significant adverse events and poor outcomes which might have been prevented if
the deterioration had been identified earlier (Goldhill, White, & Sumner,
1999). A 2019 consensus-based guideline by the International Society of Rapid
Response Systems suggests that hospitals with a well-functioning EWS and Rapid
Response System should have near-zero rates of in-hospital cardiac arrests on
general wards and recommends that each hospital tracks their in-house cardiac
arrests, potential predictability, and timeliness of escalation (Subbe et al.,
2019).

*
The Nightingale project has to date resulted in the Checkpoint Cardio prototype
system designed to improve safety on general wards by monitoring vital signs
continuously using wireless wearable vital signs sensors and combining these
data with a clinical decision support system, optimized to timely alert
caregivers of patient deterioration.

We previously evaluated the performance of this prototype wireless sensor to be
used in this study in adult healthy volunteers (METC 18/579; CCMO:
NL67031.041.1). The sensor was highly accurate for heart rate, and accurate for
respiratory rate. Continuous noninvase blood pressure tracked exercise-induced
small increases in blood pressure. Oxygen saturation (ear probe) was in the
normal range, but its performace during hypoxemia could not be evaluated within
healthy volunteers breathing room air.

However, despite these encouraging initial results in healthy volunteers,
performance in clinically ill patients might differ * there could be more or
less artefacts and/or the duration of periods with missing data could be
different - for example during episodes of profuse sweating or delirium. We
consider accurate *24/7* tracking of heart rate and respiratory rate a minimum
requirement for being able to start using these devices clinically as a patient
safety monitor.

In the present study we will test the performance of the Checkpoint Cardio
sensor solution for the first time in a clinical setting in a limited number of
patients at high risk of deterioration. There are three main study objectives:

1. to assess the agreement between vital signs measured with the new sensor and
a clinical monitoring system currently used on high-care wards (ICU, Medium
Care, operation rooms) in our hospital.

2. to evaluate the preliminary accuracy oof the whole system (wireless sensor +
clinical decision support software) for the detection of patient deterioration

3. to assess the usability of the system after the initial five days at home
after hospital discharge from a patient perspective

Design: This is a prospective cohort study in which patients receive additional
wireless monitoring on top of usual care at the hospital ward and in the first
5 days after hospital discharge. Wireless monitoring is used in a passive way:
no warnings are generated by the wireless remote monitoring system nor will it
be used to change clinical decisions of patients.

Objective 1 (agreement between sensor values and routine hospital monitors):
We will apply different *methods comparison* techniques to assess the agreement
between the wireless sensor values and those obtained with routine hospital
monitors. To this end we will make use of the fact that a large proportion of
high-risk patients is treated at some point during their hospital stay in a
high-care setting (operation room, recovery room, medium care unit, intensive
care unit). In these settings continuous (wired) vital signs monitoring is used
routinely as a standard of care. We will collect the vital signs data recorded
by the routine bedside monitor, which will allow methods comparison techniques
such as Bland-Altman analysis, Clarke-Error grids, Four-Quadrant plots) to
evaluate the agreement between sensor values and the reference standard (the
routinely used wired patient monitor) in *real life* clinical scenarios.

Objective 2 (accuracy to detect clinical deterioration):
We will compare off-line generated *deterioration probabilities* and
(theoretical) alerts using various alarm thresholds alerts in relation to the
presence or absence of to actual prospectively documented events for
prospectively collected observational patient datareflecting relevant clinical
deterioration.
The clinical decision support engine of the Checkpoint Cardio wireless monitor
system provides a *deterioration probability* and can provide a binary output
(*alert*) when the probability of deterioration exceeds a user-defined
threshold (in future clinical use this alert threshold will be set by the
caregiver, e.g., *send alert IF probability of deterioration > 40% or > 60%*).
This data will be collected both in-hospital and during the first five days
after hospital discharge. In addition, data from both the hospital and home
period will be retrospectively analysed to evaluate the number of alerts (while
varying the probability of deterioration that exceeds certain tresholds) in
patients without occurrence of events.
Next to inputs from the wireless vital signs sensor, the clinical decision
support engine can also receive inputs from the patient, the nurse, and
informal carer, as well as a pre-defined set of inputs from the hospital
electronic medical record (age, primary condition, comorbidities, selected most
recent laboratory values, as well as trends in those values).
Neither vital signs recorded by the sensor nor any outputs from the clinical
decision support system will be visible to the patient and the care team; data
will occasionally be visible to study personnel during daytime hours.

Objective 3 (usability assessment):
We will assess the experiences of patients regarding the *usability* of the
system, including the more subtle aspects of *wearability* of the sensor. We
will also evaluate the ease of sensor application and removal by study
personnel, as well as the process of battery change or recharging of the
removable battery. Usability will be assessed with patient interviews (please
see appendix F1: usability questionnaire) by phone after completion of the
study (after the initial five days at home after hospital discharge).

In this observational diagnostic study the systems will not send out alerts to
caregivers and, as a result, there will not be any additional diagnostic or
therapeutic interventions triggered by the system. Hence, all patients will
undergo the same routine monitoring by nurses in a systematic manner, and all
treatment decisions will be based on current clinical practice (clinical
reasoning with inputs from manually recorded vital signs, direct patient
observation/examination by nurses and/or doctors, laboratory values and imaging
results).

The position of the system used: an adhesive patch sensor placed on the
patient*s chest allows measuring vital signs without the inconvenience of
physical attachment to immobile monitoring systems. This sensor is lightweight,
wireless (no need for cables), and allows untethered measuring of vital signs
while in bed or at home. As with any adhesive plaster/tape, it is possible that
patients may experience skin irritation in response to the adhesives used, in
which case the sensor will be removed. Since physical discomfort from the
sensors is unlikely, patient burden will be minimal.