British Journal of Anaesthesia, 132 (3): 519e527 (2024)

doi: 10.1016/j.bja.2023.11.040
Advance Access Publication Date: 21 December 2023
Clinical Investigation

CRITICAL CARE

Impact of continuous and wireless monitoring of vital signs on

clinical outcomes: a propensity-matched observational study of
surgical ward patients
Bradley A. Rowland1, Vida Motamedi2,4,7, Frederic Michard3 , Amit K. Saha4,5,6 and
4,5,6,
Ashish K. Khanna *
1
Department of Internal Medicine, Wake Forest University School of Medicine, Winston-Salem, NC, USA, 2Wake Forest
University School of Medicine, Winston-Salem, NC, USA, 3MiCo, Vallamand, Switzerland, 4Department of
Anesthesiology, Wake Forest University School of Medicine, Winston-Salem, NC, USA, 5Perioperative Outcomes and
Informatics Collaborative (POIC), Winston-Salem, NC, USA, 6Outcomes Research Consortium, Cleveland, OH, USA and
7
Department of Anesthesiology, Vanderbilt School of Medicine, Nashville, TN, USA

*Corresponding author. E-mail: akhanna@wakehealth.edu

Abstract
Background: Continuous and wireless vital sign monitoring is superior to intermittent monitoring in detecting vital sign
abnormalities; however, the impact on clinical outcomes has not been established.
Methods: We performed a propensity-matched analysis of data describing patients admitted to general surgical wards
between January 2018 and December 2019 at a single, tertiary medical centre in the USA. The primary outcome was a
composite of in-hospital mortality or ICU transfer during hospitalisation. Secondary outcomes were the odds of indi-
vidual components of the primary outcome, and heart failure, myocardial infarction, acute kidney injury, and rapid
response team activations. Data are presented as odds ratios (ORs) with 95% confidence intervals (CIs) and n (%).
Results: We initially screened a population of 34,636 patients (mean age 58.3 (Range 18e101) yr, 16,456 (47.5%) women.
After propensity matching, intermittent monitoring (n¼12 345) was associated with increased risk of a composite of
mortality or ICU admission (OR 3.42, 95% CI 3.19e3.67; P<0.001), and heart failure (OR 1.48, 95% CI 1.21e1.81; P<0.001),
myocardial infarction (OR 3.87, 95% CI 2.71e5.71; P<0.001), and acute kidney injury (OR 1.32, 95% CI 1.09e1.57; P<0.001)
compared with continuous wireless monitoring (n¼7955). The odds of rapid response team intervention were similar in
both groups (OR 0.86, 95% CI 0.79e1.06; P¼0.726).
Conclusions: Patients who received continuous ward monitoring were less likely to die or be admitted to ICU than those
who received intermittent monitoring. These findings should be confirmed in prospective randomised trials.

Keywords: continuous; monitoring; mortality; postoperative; outcomes; vital signs; wireless monitoring

Editor’s key points  The findings of this propensity-matched analysis of

 Regular monitoring of vital signs is key to early
detection of physiological deterioration of patients
on hospital wards.
 The combination of wireless ‘wearable tech’ and, in
future, machine learning algorithms allows continuous
monitoring of vital signs and earlier detection of the
deteriorating patient with minimal impact on staffing.
observational data suggest that this technology could
lead to improved patient outcomes.
 Observational analyses of this type are vulnerable to
bias. Confirmation in a randomised trial with sup-
porting health economic analysis would accelerate
widespread implementation.

Received: 4 July 2023; Accepted: 23 November 2023

© 2023 British Journal of Anaesthesia. Published by Elsevier Ltd. All rights reserved.
For Permissions, please email: permissions@elsevier.com

519
 520 - Rowland et al.
There have been significant reductions in intraoperative mor-
tality secondary to advancements in anaesthesia, surgical
technique, and monitoring.1 Yet, all-cause postoperative mor-
tality, defined as death within the first 30 days of surgery, re-
mains about 100 times as common, affecting approximately
1e2% of all patients.2,3 The majority of these postoperative
deaths occur during index hospitalisation and under our direct
clinical supervision.4,5 Myocardial injury after noncardiac sur-
gery (MINS), bleeding, and sepsis are the three most common
causes of attributable mortality after noncardiac surgery in
adults older than 45 yr.5 Other common aetiologies of post-
operative clinical deterioration include respiratory failure, car-
diac arrhythmia, and acute renal failure.6 The Joint Commission
provision of care standard PC.02.01.19 requires that hospitals
recognise and respond to changes in a patient’s condition, and
even beyond the guidelines it is our ethical responsibility to
respond to our patients. Thus, the timely recognition of clinical
deterioration within the inpatient setting remains critical to
early intervention and prevention of further clinical decline.6
Traditional practice at most hospital ward systems in-
volves intermittent monitoring of vital signs every 4e8 h7 and
is often combined with Early Warning Scores (EWS) to identify
and risk stratify deteriorating patients.8e10 Perturbations in
vital signs, including oxygenation, ventilation, and haemody-
namic parameters, may be sustained for long periods of time
in between intermittent vital sign checks.11e14 Undetected
vital sign perturbations have significant consequences.14e17
For example, even transient postoperative hypotension is
strongly associated with myocardial injury (MI), and acute
kidney injury (AKI).15e18 Vital signs on hospital wards may
show subtle pattern changes for hours before clinical deteri-
oration.13,19e21 Intermittent monitoring can entirely miss or
fail to detect these vital sign abnormalities, thereby reducing
lead time for effective mitigating interventions. Thus, the
advent of continuous and wireless vital sign monitoring
outside of the critical care setting using wearable devices
provides more granular detection capabilities to capture clin-
ical deterioration in a timely fashion.
Safety in healthcare has shifted in recent decades towards
a systems approach including a socio-technical system where
the interaction of humans and technology is essential to
achieving the desired goals. Safety may be further understood
as Safety-I and Safety-II, where the latter aims to improve
overall performance through the development of resilient
systems by leveraging data and technology. Continuous
monitoring in this context is a driver of Safety-2, allowing
clinical teams to respond more rapidly to physiological dete-
rioration, improving the quality of care and system
performance.22
Published literature suggests the clinical benefit of contin-
uous monitoring compared with historical controls using
before and after comparison studies.23e26 At the Atrium Health
Wake Forest Baptist Medical Center, continuous ward moni-
toring using wearable and portable monitoring device tech-
nology has been in place since 2015 and is coupled with vital
sign threshold-based alarms and interventions based on best
practices.27 An earlier report from our institution demon-
strated a significant decrease in rapid response calls after
implementation.27 We have also reported substantial changes
in heart rate (HR) and blood pressure (BP) detected by these
monitoring devices that would have gone undetected with
traditional intermittent monitoring.28
We sought to evaluate and compare clinical outcomes in a
large propensity-matched, postoperative population of
surgical patients with contemporaneous controls, admitted to
a single institution who received either continuous or inter-
mittent monitoring every 4e6 h. Our hypothesis was that with
earlier detection of vital sign aberrations continuously, use of
this monitoring modality would reduce in-hospital mortality
and ICU transfers compared with intermittent monitoring.
Other outcomes including heart failure (HF), MI and AKI, and
length of hospital and ICU stay were also examined.
Methods
We conducted a single-centre, retrospective analysis using data
from electronic medical records (EMR). We first identified adult
surgical patients (aged 18 yr) admitted to a tertiary care centre,
(Atrium Health Wake Forest Baptist Medical Center main
campus) over a 2-yr period (between January 1, 2018, and
December 31, 2019), undergoing noncardiac surgical procedures.
Surgical services admitting these patients included ortho-
paedics, urology, general surgery, neurosurgery, gynaecology,
vascular, otolaryngology, cardiothoracic surgery, trauma, and
others. Other subspecialties included hand, gastroenterology,
haematology/oncology, interventional radiology, oral surgery,
plastics, podiatry, and ophthalmology. Race and ethnicity
were self-described by patients on admission and then retro-
spectively collected with other patient characteristics from the
EMR. Our study and a priori defined statistical analysis plan
were approved by the Wake Forest School of Medicine Insti-
tutional Review Board (IRB00083520).
Monitoring protocols
Patients admitted to the surgical wards received either (1) vital
sign monitoring with a continuous wireless monitoring de-
vice, ViSi Mobile, (Sotera® Wireless, Inc. San Diego, CA, USA)
or (2) standard intermittent monitoring (every 4e6 h). Briefly,
the ViSi Mobile System is a portable wrist-mounted system
that received clearance by the Food and Drug Administration
in 2012. These ViSi devices were purchased and implemented
by Wake Forest Baptist hospital system in 2015. Today, ViSi
monitors are on several different medical and surgical units;
however, not all units and patients have access to this moni-
toring. If hospital units do not have ViSi monitors or within a
unit if patients decline to wear the monitor, then intermittent
monitoring is used. Hospital units with ViSi devices are staffed
by the same number of personnel as those not doing this
monitoring. Nurse-to-patient ratios on our floors range from
1:3 to 1:6. For this analysis, patients were not randomised into
continuous or intermittent monitoring. The type of moni-
toring was dependent upon the floor a patient was transferred
to after operation, which was dependent on bed availability
and not patient’s clinical status. Based on clinical experience,
there would not have been significant crossover between
groups, that is, once a patient arrived at the floor and were
allocated to either continuous or intermittent monitoring,
then they would most likely continue that strategy until
discharge; unless for example, a patient was transferred to a
different unit and received a different type of monitoring
strategy there.
The ViSi device records oxygen saturation (SpO2), HR,
ventilatory frequency, continuous noninvasive BP, and either
3-lead or 5-lead ECG. Continuous BP measurements are esti-
mated from pulse arrival time, specifically the time that
elapses between the R wave being detected by the ECG and the
arrival of the resulting pulse at the SpO2 finger sensor.

According to the manufacturer, the estimated maximal mean
error is 5 mm Hg with a standard deviation 8 mm Hg.29 ViSi
monitors were calibrated daily, connected to the hospital’s
wireless network, and measurements were distributed to the
central nurse’s station for that unit and continuously dis-
played. Vital sign abnormalities exceeding established
thresholds for the local hospital system generated alerts that
were distributed to a central station and to the nurses’
hospital-supplied phones (ASCOM, Morrisville, NC, USA).28
Vital sign alarms not addressed by the primary nurses are
escalated to other floor nurses, then to the unit manager.
The EWS at our institution are based on the patient’s
weighted vital signs consisting of HR, ventilatory frequency,
SpO2, temperature, BP, rate/percentage of inspired oxygen, and
level of awareness as determined by the patient’s nurse. EWS
are calculated by the EMR every 4 h based on data input from
ViSi monitors which can be collected from the central display or
at the bedside or those collected by intermittent monitoring.
Total scores range from 0 to 20 and can be calculated from vital
sign measurements collected via ViSi monitor or intermittent
monitoring. Ad hoc EWS may be generated by the clinical team
in response to changes in a patient’s clinical status through
either monitoring strategy. For example, if abnormal vital signs
are detected by ViSi monitoring and an alarm is generated at
the central nurse’s station, the clinical team may assess the
patient and generate an updated EWS. A rapid response team
(RRT) is triggered if EWS is 8.
Outcomes
Our primary outcome was the risk of a composite of post-
operative in-hospital mortality or ICU admission during the
hospital encounter. Secondary outcomes were the individual
components of the primary outcome (in-hospital mortality
and ICU admission), and risk of HF, MI, AKI, and RRT activa-
tions throughout the hospitalisation. Exploratory outcomes
including hospital length of stay (LOS), ICU LOS and median
EWS values for the entire LOS were accessed using the ordi-
nary least squares regression model. Diagnosis of HF, MI, and
AKI were identified using International Statistical Classifica-
tion of Diseases and Related Health Problems (ICD-10) billing
codes available in the EMR.
Statistical analysis
Statistical analysis was performed in R v3.6.1 (R Foundation
for Statistical Computing, Vienna, Austria) using RStudio
environment v1.1.456 (RStudio, Boston, MA, USA). The par-
ticipants were first stratified into ViSi continuous monitoring
and contemporaneous controls as intermittent, standard vi-
tal sign monitoring groups. Sequential random nearest
neighbour matching with a propensity score ratio of 1:3 and
with a caliper of 0.2530,31 was used to balance the potential
factors affecting the outcome including selection bias. To
create propensity score-matched pairs, we performed
random one-to-three matching using the MatchIt package in
R, in which the control variables in our study were used
(race, sex, ethnicity, age, quarter-year of surgery, surgical
area, BMI, surgical duration in hours, ASA physical status,
Charlson Comorbidity Index [CCI] and patient’s primary in-
surance type).32 The variable selection for propensity score
matching was based on feedback from clinical expertise. To
adjust for the effect of time and variability of the continuous
wireless monitoring device usage, the quarter-year of the
Continuous versus intermittent vital signs - 521
surgery was used as a covariate in the model. Standardised
differences were used as indicators of intergroup balance. If
the standardised differences were less than 10%, the cova-
riates between the two groups were considered well-
balanced.
For summary statistics, categorical variables were
described as numbers and percentages and non-normally
distributed continuous variables were described as median
and interquartile ranges (IQRs). Primary, secondary, and
exploratory outcomes were compared between groups and
subgroups. Multivariable logistic regression was performed to
detect differences in outcomes between groups. Initially
restricted cubic splines were included with age and CCI to
relax the linearity assumptions of the multivariable logistic
regression model. As the non-linear components of age and
CCI were not statistically significant, the splines were removed
from variables. The multivariable logistic regression adjusted
model were controlled for race, sex, ethnicity, age, BMI, sur-
gical duration, surgical area, ASA physical status, and CCI. The
P-value for significance level within the primary and second-
ary outcome analyses was set at 0.004 for each group of ana-
lyses after Bonferroni correction for multiple comparisons.
To test the influence of unobserved confounding, we used
an E-value analysis to quantify the potential implications of
unmeasured confounders. E-value is defined as the minimum
strength of association on the risk ratio scale that an unmea-
sured confounder would need to have with both the exposure
and the outcome, conditional on the measured covariates, to
fully explain away a specific exposureeoutcome association.
We further performed multivariable logistic regression for the
composite outcome on the whole, unmatched, study cohort
(n¼34 636) to assess the effect of continuous monitoring after
controlling for race, sex, ethnicity, age, BMI, surgical duration,
surgical area, ASA physical status, and CCI as a sensitivity
analysis.
Results
We identified 34,636 patients who met inclusion criteria
(Fig. 1). Over the study period, 11,176 patients received
continuous wireless monitoring compared with 23,460
contemporaneous controls who received intermittent moni-
toring. Based on propensity matching, 7955 (71.2%) patients
receiving continuous wireless monitoring were matched with
12,345 (52.6%) contemporaneous patients receiving intermit-
tent monitoring. When propensity matching was not possible,
they were excluded from propensity score-matched analysis.
Table 1 presents patient and clinical characteristics in the
unmatched and propensity-matched models. Furthermore,
8902 (37.9%) patients receiving intermittent monitoring had
in-hospital mortality or ICU admission compared with 1428
(12.8%) patients receiving continuous wireless monitoring
among the whole study cohort. Patients who had intermittent
vital sign monitoring had greater odds of the primary com-
posite outcome than those who were on continuous moni-
toring (odds ratio [OR] 3.42, 95% confidence interval [CI]
3.19e3.67; P<0.001; Table 2, Fig. 2). The primary composite
outcome was primarily driven by ICU admission (in-hospital
mortality 57 [0.7%] vs 242 [1.9%]; ICU admission 804 [10.1%] vs
4342 [35.1%]). The remainder of all individual secondary out-
comes (in-hospital mortality, ICU admission, HF, MI, AKI, and
RRT activation) are shown in Table 2 and Figure 2. Intermit-
tently monitored patients had a significantly increased risk of
in-hospital mortality, ICU admission, HF, MI, and AKI
 522 - Rowland et al.

Adult patients receiving noncardiac surgical

procedures at Atrium Health Wake Forest Baptist
Hospital, Winston-Salem, NC, USA main campus
requiring a length of stay of ≥48 h, between
January 1, 2018 and December 31, 2019,
(n=34 636)

Unmatched analysis Propensity score-matched analysis

x Patients who received IM, standard vital sign x Patients who received IM, standard vital sign
monitoring, (n=23 460) monitoring, (n=12 345)
x Patients who received CVSM with Sotera® x Patients who received CVSM with Sotera®
Wireless ‘VISI’ wearable device, (n=11 176) Wireless ‘VISI’ wearable device, (n=7,955)

Fig 1. Study flow diagram for patient selection. CVSM, continuous wireless vital sign monitoring; IM, intermittent monitoring.

compared with continuous wireless monitoring, except for
RRT activations which were similar in both groups (Table 2,
Fig. 2). E-values for each of the associations of exposure and
outcome are also reported in Table 2. For the composite
outcome of in-hospital mortality or ICU admission, an un-
measured confounder would need to have a strength of as-
sociation of a risk ratio of 3.10 with both the type of monitoring
and the outcome, conditional on the measured covariates, to
fully explain away the OR of 3.42 in favour of continuous
monitoring.
Supplementary Table S1 shows the characteristics of pa-
tients in each group that matched and those that were
excluded because of no match. Absolute values of the primary
and secondary outcomes within the populations in each of the
monitoring groups are shown in Supplementary Table S2.
Supplementary Figure 1a shows covariate balance plot before
and after propensity score matching between continuous
wireless monitoring cohort with the control and intermittent
monitoring cohort. Caliper width of 0.25 is used for propensity
match. Supplementary Figure 1b shows a kernel density dis-
tribution of propensity score before and after matching be-
tween the continuous monitoring cohort and the intermittent
monitoring cohort.
Within the propensity-matched model, patients who
received intermittent monitoring had a longer hospital LOS
median [IQR] (4.19 [1.50e9.01] vs 3.31 [1.58e6.85]; P<0.001) and
a longer ICU LOS 2.27 [1.19e4.89] vs 1.91 [0.91e3.73] days;
P<0.001) (Table 3). Although median EWS values were not
dissimilar, maximal EWS values were much higher in the
intermittent monitoring group 1.5 (0, 13) than the continuous
monitoring group 2 (0, 7) (Table 3). Supplementary Figure 1c
shows EWS distribution and spread for both unmatched and
propensity-matched cohort.
As part of sensitivity analysis, in the whole, unmatched
study cohort (n¼34,636), we measured effects of an intermit-
tent monitoring after adjustment for all other cofounders and
showed significant odds (OR 1.57, 95% CI 1.34e1.87; P<0.001)
for in-hospital mortality with an event rate of 2.6% among the
intermittent monitoring group vs 0.6% among the continuous
monitoring group. The adjusted odds of ICU admission were
significant (OR 4.16, 95% CI 3.90e4.43; P<0.001) for intermit-
tent monitoring group compared with continuous monitoring
group with an event rate of 36.4% among the intermittent
monitoring group vs 8.7% among the continuous monitoring
group. The composite outcome adjusted odds were signifi-
cantly greater for the intermittent monitoring group (OR 4.17,
95% CI 3.92e4.43; P<0.001) than for the continuous monitoring
group.
Discussion
In this propensity-matched analysis, continuous and wireless
monitoring of adult surgical ward patients was associated
with a reduced risk of in-hospital mortality and ICU admis-
sion, compared with traditional intermittent monitoring. This
improvement in patient outcomes associated with continuous
wireless monitoring may be secondary to the decreased risk of
severe adverse events, consequent to earlier detection of vital
signs changes that lead up to these conditions. Patients
receiving intermittent monitoring were as likely to have RRT
activation but greater odds of ICU admission than those
receiving continuous monitoring. We hypothesise that this is
because of lack of detection of progressive changes in vital
signs leading to greater clinical deterioration at RRT activation
in the intermittent vital signs cohort that would have other-
wise been detected and received intervention earlier, thereby
preventing ICU admission in the continuous monitoring
cohort.
Patients who received continuous wireless monitoring had
a reduced risk of new onset of HF, MI, and AKI. Each of these
disease states is often preceded by vital sign abnormalities
that if undetected and unrecognised may lead to further
decompensation.33 Continuous wireless monitoring provides
instantaneous streams of data for the clinical care team as
compared with the ‘spot checks’ of intermittent monitoring,
which continues to be the standard of care with a bedside
nurse or provider checking in on a patient every 4e6 h.
Therefore, the time paradigm between the two monitoring
strategies can be upwards of several hours, which, in patients
with organ system failure secondary to haemodynamic
 Continuous versus intermittent vital signs - 523

Table 1 Patient and clinical characteristics. IQR, interquartile range; SMD, standard mean deviation.

Unmatched model Propensity-matched model

Continuous Intermittent SMD Continuous Intermittent SMD

wireless monitoring wireless monitoring
monitoring (n¼23 460) monitoring (n¼12345)
(n¼11 176) (n¼7955)

Quartile (Q) - Year (%) 1.192 0.094

Q1-2018 1976 (17.7) 1996 (8.5) 1347 (16.9) 1738 (14.1)
Q2-2018 2518 (22.5) 1815 (7.7) 1453 (18.3) 1723 (14.0)
Q3-2018 2329 (20.8) 1927 (8.2) 1469 (18.5) 1830 (14.8)
Q4-2018 1323 (11.8) 3009 (12.8) 1320 (16.6) 2393 (19.4)
Q1-2019 241 (2.2) 3988 (17.0) 241 (3.0) 850 (6.9)
Q2-2019 206 (1.8) 4180 (17.8) 206 (2.6) 664 (5.4)
Q3-2019 392 (3.5) 4079 (17.4) 392 (4.9) 1067 (8.6)
Q4-2019 2191 (19.6) 2466 (10.5) 1527 (19.2) 2080 (16.8)
Age (yr), median (IQR) 60.00 (48.00e70.00) 61.00 (48.00e71.00) 0.036 60.00 (48.00e70.00) 60.00 (48.00e70.00) 0.024
Surgical subspecialty (%) 0.537 0.021
Orthopaedics 2344 (21.0) 3648 (15.5) 1378 (17.3) 1894 (15.3)
Urology 1338 (12.0) 1604 (6.8) 800 (10.1) 1032 (8.4)
General 2336 (20.9) 3212 (13.7) 1497 (18.8) 2047 (16.6)
Neurosurgery 1154 (10.3) 2360 (10.1) 907 (11.4) 1276 (10.3)
Gynaecology 718 (6.4) 948 (4.0) 404 (5.1) 555 (4.5)
Other 1591 (14.2) 5243 (22.3) 1559 (19.6) 2719 (22.0)
Vascular 162 (1.4) 1597 (6.8) 162 (2.0) 561 (4.5)
Otolaryngology 803 (7.2) 1240 (5.3) 0 (0.0) 0 (0.0)
Thoracic surgery 433 (3.9) 2314 (9.9) 542 (6.8) 754 (6.1)
Trauma 297 (2.7) 1294 (5.5) 429 (5.4) 963 (7.8)
ASA physical status (%) 0.528 0.074
1 207 (1.9) 717 (3.1) 202 (2.5) 374 (3.0)
2 2785 (24.9) 3997 (17.0) 1729 (21.7) 2398 (19.4)
3 7097 (63.5) 11983 (51.1) 4940 (62.1) 6976 (56.5)
4 1054 (9.4) 5990 (25.5) 1051 (13.2) 2503 (20.3)
5 33 (0.3) 749 (3.2) 33 (0.4) 94 (0.8)
6 0 (0.0) 24 (0.1) 0 (0.0) 0 (0.0)
Sex (%) 0.075 0.033
Male 5584 (50.0) 12,596 (53.7) 4032 (50.7) 6458 (52.3)
Female 5592 (50.0) 10,864 (46.3) 3923 (49.3) 5887 (47.7)
Ethnicity, 414 (3.7) 896 (3.8) 0.006 309 (3.9) 459 (3.7) 0.009
Hispanic/Latino (%)
Race (%) 0.009 0.012
White 8703 (77.9) 18,214 (77.6) 6206 (78.0) 9603 (77.8)
Black or African 1901 (17.0) 4000 (17.1) 1328 (16.7) 2109 (17.1)
American
Other 572 (5.1) 1246 (5.3) 421 (5.3) 633 (5.1)
Surgical duration (h), 2.95 (1.97e3.78) 3.03 (1.85e3.83) 0.039 3.02 (1.95e3.78) 3.08 (1.90e3.88) 0.071
median (IQR)
Charlson Comorbidity 3.00 (1.00e4.00) 3.00 (1.00e5.00) 0.113 3.00 (1.00e4.00) 3.00 (1.00e5.00) 0.059
Index, median (IQR)
Hypertension (%) 4662 (41.7) 10,044 (42.8) 0.022 3307 (41.6) 5099 (41.3) 0.005
BMI, median (IQR) 29.05 (24.75e33.45) 28.53 (24.26e33.04) 0.017 28.89 (24.55e33.20) 28.68 (24.29e33.09) <0.001
Insurance class (%) 0.105 0.029
Commercial 3130 (28.0) 6110 (26.0) 2188 (27.5) 3347 (27.1)
Government 6460 (57.8) 14,665 (62.5) 4719 (59.3) 7475 (60.6)
Other 1586 (14.2) 2685 (11.4) 1048 (13.2) 1523 (12.3)

insults, as an example, would increase the risk of poor out-
comes. Recent work from nearly 15,000 patients on our hos-
pital wards has also shown that continuously measured HR
and BP changes are common and likely missed with inter-
mittent monitoring.28
Similarly, patients with HF, MI, and AKI may have vital sign
aberrations, such as prolonged periods of hypotension pre-
ceding these events and early detection may allow for timely
intervention and a reduction in the occurrence and progres-
sion of these.34e36 Our work also demonstrated that the
downstream effect of lack of continuous monitoring could
translate into longer hospital LOS and when transferred to the
ICU as a possible unplanned admission, a longer stay in that
unit as well. This adds to healthcare resource burden and
strain.37 Patients receiving intermittent monitoring may have
initially been sicker or more continuous wireless monitoring
usage may have been deployed over a time course to a
healthier group. To account for these differences, cohorts were
matched on a combination of baseline covariates including
several comorbidities present on admission, intraoperative
duration of surgery, along with every quarter-year of surgery
to adjust for time variance of continuous wireless monitoring
 524 - Rowland et al.

Composite outcome

In-hospital mortality

ICU admission

Heart failure

Myocardial infarction

Acute kidney injury

RRT activation

–4 –2 –1 0 1 2 4 8
Favours intermittent monitoring Favours continuous monitoring
Odds ratio (95% confidence interval)
Propensity-matched cohort Unmatched cohort

Fig 2. Forest plot of odds ratio (95% confidence interval) of primary and secondary outcomes for patients receiving intermittent monitoring
vs continuous wireless monitoring. RRT, rapid response team.

usage throughout our large hospital system. Our reported
E-values for unobserved confounding are consistently on the
higher side that provide further reassurance for our results.
Clinical outcomes in our results are in concert with other
recent literature for wireless continuous wireless monitoring,
including this specific technology, that has examined some of
these outcomes, such as hospital LOS,38 ICU admissions,23 ICU
and in-hospital mortality,39 either as before and after or
observational cohort designs. The risk of RRT activations was
not different in the two groups examined. This may be because
RRT activations are driven from threshold EWS, which are
calculated every 4e6 h at our institution whether patients are
monitored continuously or otherwise. It is therefore plausible
that the bedside team using continuous wireless monitoring
was able to provide proactive care for patients with vital sign
changes and achieve lower risk of in-hospital mortality and
ICU admissions, whereas EWS-driven RRT presence was not
different. This is substantiated by the fact that median EWS

Table 2 Primary and secondary outcomes for intermittent monitoring vs continuous wireless monitoring unmatched and matched
models. Values are presented as odds ratios (95% CI). CI, confidence interval; RRT, rapid response team.

Outcomes Unmatched model Propensity-matched model E-value

Composite 4.17 (95% CI: 3.92e4.43), P<0.001 3.42 (95% CI: 3.19e3.67), P<0.001 3.10
In-hospital mortality 1.57 (95% CI: 1.34e1.87), P<0.001 1.25 (95% CI: 1.12e1.64), P<0.001 1.81
ICU admission 4.16 (95% CI: 3.90e4.43), P<0.001 3.43 (95% CI: 3.21e3.68), P<0.001 6.14
Heart failure 2.28 (95% CI: 1.91e2.72), P<0.001 1.48 (95% CI: 1.21e1.81), P<0.001 2.32
Myocardial infarction 5.61 (95% CI: 4.05e7.74), P<0.001 3.87 (95% CI: 2.71e5.71), P<0.001 7.21
Acute kidney injury 1.66 (95% CI: 1.43e1.93), P<0.001 1.32 (95% CI: 1.09e1.57), P<0.001 1.97
RRT activation 0.98 (95% CI: 0.89e1.09), P¼0.758 0.86 (95% CI: 0.79e1.06), P¼0.726 1.61

Table 3 Exploratory outcomes for propensity-matched cohorts. EWS, Early Warning Score; IQR, interquartile range; LOS, length of stay.
Continuous wireless
monitoring (n¼7955)
Hospital LOS (days), median (IQR) 3.31 (1.58e6.85)
ICU LOS (days), median (IQR) 1.91 (0.91e3.73)
EWS median value during LOS, 2 (0e7)
median (minimumemaximum)
values were not dissimilar in either group. However, the
intermittent monitoring group had a wider range and a higher
maximum EWS than the continuous monitoring group.
Our work was a propensity-matched cohort from a hospital
system where continuous wireless monitoring was imple-
mented and used in conjunction with vital signs alarms and
interventions. Previous studies were before and after com-
parison studies, and their results may be confounded by an
improvement of quality of care over time. We used a
contemporaneous comparator control group that was
matched on several characteristics that would have influ-
enced the choice of monitoring and these outcomes. We also
looked at a large set of outcomes that may be considered
clinically relevant effects of the type of ward monitoring.
Prospective interventional trials are now needed to confirm
the effect and the magnitude of benefit of continuous wireless
monitoring on clinical outcomes.
Our study was limited by its retrospective approach.
Although propensity matching allowed us to match patients
on baseline covariates, we were unable to match based on
ongoing changes in a patient’s hospital course. Our model
included an array of baseline comorbidity burden factors,
intraoperative duration of surgical procedure, and accounted
for the effect of time as quartile of year. Although the use of
continuous wireless monitoring or intermittent monitoring
was not based on patient condition, we can speculate that
patients after complicated surgery may have been placed on
continuous wireless monitoring as a preference. Similarly,
patients who refused continuous monitoring received inter-
mittent monitoring. In the absence of granular data for rea-
sons for refusal of continuous monitoring, we can speculate
that ambulatory and stable patients were more likely to
receive intermittent monitoring. However, all these factors, if
anything, should not have driven outcomes in favour
of continuous wireless monitoring. Our approach with pro-
pensity matching was set at a caliper of 0.25 and the drop off
for continuously monitored patients does demonstrate that
patient groups may be different at baseline within the defined
caliper. However, we did achieve adequate matching and an
increase of caliper from 0.25 to 0.35 did not show any signifi-
cant change in matched cohort number. We relied on reported
rates of conditions from discharge summaries and adminis-
trative codes which have inherent flaws. We did not evaluate
changes in providers, each with different clinical management
habits. Most importantly, this very large administrative data-
set does not have granular information that would allow
deeper understanding of the mechanisms of these favourable
outcomes for continuous monitoring. This includes the num-
ber of alarms, false alarms, or situations where certain alarms
were not responded to and needed to be escalated within the
nursing workforce or whether nursing staff training and
Continuous versus intermittent vital signs - 525
Intermittent monitoring P-value
(n¼12 345)
4.19 (1.50e9.01) <0.001
2.27 (1.19e4.89) <0.001
1.5 (0e13) <0.001
behaviour influenced the response to monitoring. Similarly,
we have limited information on EWS values at the time of RRT
activation, specifically if an EWS check was per normal pro-
tocol or an escalation of care measurement. Therefore, our
reported EWS values (which include median, minimum,
maximum, and overall spread along with absolute numbers of
RRT activations) for the entire LOS for both groups should be
interpreted with caution. A smaller pilot from the same
institution and using the same monitoring devices reported a
rate of 2.3 alarms per patient per day and a significant decrease
in RRT calls from 189 to 158 per 1000 discharges after
implementation.27
Conclusions
In this propensity-matched analysis of observational data,
patients receiving intermittent monitoring were at greater risk
of transfer to ICU or death during their index hospitalisation
than those receiving continuous wireless monitoring. Pro-
spective randomised trials are necessary to confirm the
impact of continuous monitoring on morbidity and mortality.
Authors’ contributions
Conceptualisation: BAR, AKK
Conceptualising statistical analysis: AKS
Writing, editing, reviewing statistical analysis and results: all
authors
Declaration of interest
AKK consults for Medtronic, Edwards Life Sciences, Philips
Research North America, Baxter, GE Healthcare, Potrero Med-
ical, Retia Medical, and Caretaker Medical. The Department of
Anesthesiology at Wake Forest University School of Medicine
is supported by Edwards Lifesciences under a master clinical
trial agreement and receives grant funding from Masimo and
Medtronic. FM is the founder and managing director of MiCo, a
consulting and research firm based in Switzerland. He has no
relationship with Sotera Wireless.
Acknowledgements
This work was supported by funding at the Department of
Anesthesiology, Wake Forest School of Medicine, Winston-
Salem, NC, USA.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.bja.2023.11.040.
 526 - Rowland et al.
References
1. Braz LG, Braz DG, Cruz DS, Fernandes LA, Modolo NS,
Braz JR. Mortality in anesthesia: a systematic review.
Clinics (Sao Paulo) 2009; 64: 999e1006
2. Fecho K, Lunney AT, Boysen PG, Rock P, Norfleet EA.
Postoperative mortality after inpatient surgery: incidence
and risk factors. Ther Clin Risk Manag 2008; 4: 681e8
3. Semel ME, Lipsitz SR, Funk LM, Bader AM, Weiser TG,
Gawande AA. Rates and patterns of death after surgery in
the United States, 1996 and 2006. Surgery 2012; 151:
171e82
4. de Vries S, Jeffe DB, Davidson NO, Deshpande AD,
Schootman M. Postoperative 30-day mortality in patients
undergoing surgery for colorectal cancer: development of
a prognostic model using administrative claims data.
Cancer Causes Control 2014; 25: 1503e12
5. Vascular Events in Noncardiac Surgery Patients Cohort
Evaluation (VISION) Study Investigators, Spence J, LeMa-
nach Y, et al. Association between complications and
death within 30 days after noncardiac surgery. CMAJ 2019;
191: E830e7
6. Petit C, Bezemer R, Atallah L. A review of recent advances
in data analytics for post-operative patient deterioration
detection. J Clin Monit Comput 2018; 32: 391e402
7. Khanna AK, Hoppe P, Saugel B. Automated continuous
noninvasive ward monitoring: future directions and
challenges. Crit Care 2019; 23: 194
8. Kovacs C, Jarvis SW, Prytherch DR, et al. Comparison of
the National early warning score in non-elective medical
and surgical patients. Br J Surg 2016; 103: 1385e93
9. Smith ME, Chiovaro JC, O’Neil M, et al. Early warning
system scores for clinical deterioration in hospitalized
patients: a systematic review. Ann Am Thorac Soc 2014; 11:
1454e65
10. Michard F, Thiele RH, Saugel B, et al. Wireless wearables
for postoperative surveillance on surgical wards: a survey
of 1158 anaesthesiologists in Western Europe and the
USA. BJA Open 2022; 1, 100002
11. Sun Z, Sessler DI, Dalton JE, et al. Postoperative hypox-
emia is common and persistent: a prospective blinded
observational study. Anesth Analg 2015; 121: 709e15
12. Turan A, Chang C, Cohen B, et al. Incidence, severity, and
detection of blood pressure perturbations after abdominal
surgery: a prospective blinded observational study. Anes-
thesiology 2019; 130: 550e9
13. Saab R, Wu BP, Rivas E, et al. Failure to detect ward
hypoxaemia and hypotension: contributions of insuffi-
cient assessment frequency and patient arousal during
nursing assessments. Br J Anaesth 2021; 127: 760e8
14. Khanna AK, Bergese SD, Jungquist CR, et al. Prediction of
opioid-induced respiratory depression on inpatient wards
using continuous capnography and oximetry: an inter-
national prospective, observational trial. Anesth Analg
2020; 131: 1012e24
15. Jones AE, Yiannibas V, Johnson C, Kline JA. Emergency
department hypotension predicts sudden unexpected in-
hospital mortality: a prospective cohort study. Chest
2006; 130: 941e6
16. Liem VGB, Hoeks SE, Mol K, et al. Postoperative hypoten-
sion after noncardiac surgery and the association with
myocardial injury. Anesthesiology 2020; 133: 510e22
17. Khanna AK, Shaw AD, Stapelfeldt WH, et al. Postoperative
hypotension and adverse clinical outcomes in patients
without intraoperative hypotension, after noncardiac
surgery. Anesth Analg 2021; 132: 1410e20
18. Sessler DI, Meyhoff CS, Zimmerman NM, et al. Period-
dependent associations between hypotension during and
for four days after noncardiac surgery and a composite of
myocardial infarction and death: a substudy of the POISE-
2 trial. Anesthesiology 2018; 128: 317e27
19. Khanna AK, Ahuja S, Weller RS, Harwood TN. Post-
operative ward monitoring - why and what now? Best
Pract Res Clin Anaesthesiol 2019; 33: 229e45
20. Brekke IJ, Puntervoll LH, Pedersen PB, Kellett J,
Brabrand M. The value of vital sign trends in predicting
and monitoring clinical deterioration: a systematic re-
view. PLoS One 2019; 14, e0210875
21. Sessler DI, Saugel B. Beyond ’failure to rescue’: the time
has come for continuous ward monitoring. Br J Anaesth
2019; 122: 304e6
22. Webster CS, Mahajan R, Weller JM. Anaesthesia and pa-
tient safety in the socio-technical operating theatre: a
narrative review spanning a century. Br J Anaesth 2023;
131: 397e406
23. Eddahchouri Y, Peelen RV, Koeneman M, Touw HRW, van
Goor H, Bredie SJH. Effect of continuous wireless vital sign
monitoring on unplanned ICU admissions and rapid
response team calls: a before-and-after study. Br J Anaesth
2022; 128: 857e63
24. Areia C, Biggs C, Santos M, et al. The impact of wearable
continuous vital sign monitoring on deterioration detec-
tion and clinical outcomes in hospitalised patients: a
systematic review and meta-analysis. Crit Care 2021; 25:
351
25. Cardona-Morrell M, Prgomet M, Turner RM, Nicholson M,
Hillman K. Effectiveness of continuous or intermittent
vital signs monitoring in preventing adverse events on
general wards: a systematic review and meta-analysis. Int
J Clin Pract 2016; 70: 806e24
26. Michard F, Kalkman CJ. Rethinking patient surveillance on
hospital wards. Anesthesiology 2021; 135: 531e40
27. Weller RS, Foard KL, Harwood TN. Evaluation of a wire-
less, portable, wearable multi-parameter vital signs
monitor in hospitalized neurological and neurosurgical
patients. J Clin Monit Comput 2018; 32: 945e51
28. Khanna AK, O’Connell NS, Ahuja S, et al. Incidence,
severity and detection of blood pressure and heart rate
perturbations in postoperative ward patients after
noncardiac surgery. J Clin Anesth 2023; 89, 111159
29. ViSi Mobile System Technical Specifications. Visi Mobile -
Sotera wireless 2015
30. Austin PC. Optimal caliper widths for propensity-score
matching when estimating differences in means and dif-
ferences in proportions in observational studies. Pharm
Stat 2011; 10: 150e61
31. Rassen JA, Shelat AA, Myers J, Glynn RJ, Rothman KJ,
Schneeweiss S. One-to-many propensity score matching
in cohort studies. Pharmacoepidemiol Drug Saf 2012;
21(Suppl 2): 69e80
32. Stuart EA, King G, Imai K, Ho D. MatchIt: nonparametric
preprocessing for parametric causal inference. J Stat Softw
2011; 42(8)

33. Kenzaka T, Okayama M, Kuroki S, et al. Importance of
vital signs to the early diagnosis and severity of sepsis:
association between vital signs and sequential organ
failure assessment score in patients with sepsis. Intern
Med 2012; 51: 871e6
34. Ancion A, Tridetti J, Nguyen Trung ML, Oury C,
Lancellotti P. Serial heart rate measurement and mortality
after acute heart failure. ESC Heart Fail 2020; 7: 103e6
35. Matsue Y, Sama IE, Postmus D, et al. Association of early
blood pressure decrease and renal function with prog-
nosis in acute heart failure. JACC Heart Fail 2021; 9:
890e903
Continuous versus intermittent vital signs - 527
36. Sato R, Luthe SK, Nasu M. Blood pressure and acute kid-
ney injury. Crit Care 2017; 21: 28
37. Halpern NA, Pastores SM. Critical care medicine beds, use,
occupancy, and costs in the United States: a methodo-
logical review. Crit Care Med 2015; 43: 2452e9
38. Brown H, Terrence J, Vasquez P, Bates DW,
Zimlichman E. Continuous monitoring in an inpatient
medical-surgical unit: a controlled clinical trial. Am J Med
2014; 127: 226e32
39. Subbe CP, Duller B, Bellomo R. Effect of an automated
notification system for deteriorating ward patients on
clinical outcomes. Crit Care 2017; 21: 52
Handling Editor: Rupert Pearse