Content uploaded by Timothy N Harwood
Author content
All content in this area was uploaded by Timothy N Harwood on Aug 16, 2018
Content may be subject to copyright.
Vol.:(0123456789)
1 3
Journal of Clinical Monitoring and Computing
https://doi.org/10.1007/s10877-017-0085-0
ORIGINAL RESEARCH
Evaluation ofawireless, portable, wearable multi-parameter vital
signs monitor inhospitalized neurological andneurosurgical patients
RobertS.Weller1· KristinaL.Foard2· TimothyN.Harwood1,3
Received: 12 January 2017 / Accepted: 28 November 2017
© Springer Science+Business Media B.V., part of Springer Nature 2017
Abstract
Unrecognized changes in patients’ vital signs can result in preventable deaths in hospitalized patients. Few publications or
studies instituting routine patient monitoring have described implementation and the setting of alarm parameters for vital
signs. We wanted to determine if continuous multi-parameter patient monitoring can be accomplished with an alarm rate
that is acceptable to hospital floor nurses and to compare the rate of patient deterioration events to those observed with rou-
tine vital sign monitoring. We conducted a prospective, observational, 5-month pilot study in a 26-bed adult, neurological/
neurosurgical unit (non-ICU) in an academic medical center. A patient surveillance system employing a wireless body-worn
vital signs monitor with automated nursing notification of alarms via smartphones was used to gather data. Data collected
included: alarm rates, rapid response team (RRT) calls, intensive care unit (ICU) transfers, and unplanned deaths before
and during the pilot study. Average alarm rate for all alarms (SpO2, HR, RR, NIBP) was 2.3 alarms/patient/day. The RRT
call rate was significantly reduced (p < 0.05) from 189 to 158 per 1000 discharges. ICU transfers per 1000 discharges were
insignificantly reduced from 53 to 40 compared to the previous 5-month period in the same unit. Similar measures of com-
parison units did not change over the same period. Although unplanned patient deaths in the study unit were also reduced
during the intervention period, this was not statistically significant. Continual, multi-parameter vital signs monitoring can
be customized to reduce a high alarm rates, and may reduce rapid response team calls.
Keywords Critical care· Deterioration· Early warning score· Electronic health records· Patient safety· Physiologic
monitoring
1 Introduction
Historically, the major focus on reducing perioperative mor-
bidity and mortality has been on identifying and reducing
risk factors for anesthesia and surgery. Much less empha-
sis has been placed on elucidating risks during the post-
operative period. One method in which to determine signs
of adverse events in the hospitalized patient, including the
postoperative patient, has been collection, recording, and
tracking of vital signs (VS) data. Traditionally, this infor-
mation is thought to be of great importance in detecting
deterioration and monitoring inpatient medical conditions.
The standard VS [temperature, heart rate (HR), blood pres-
sure (BP), and respiratory rate (RR)] and oxygen satura-
tion (SpO2) are routinely but infrequently recorded outside
of critical care areas. Continuous monitoring on a general
medical-surgical unit may be associated with a decrease in
total length of stay in both hospital and intensive care unit
days, as well as a lower incidence of cardiac arrest [1].
On the other hand, manually entered pulse oximetry data
may not truly reflect more granular, continuous pulse oxi-
metry data that can be automatically captured on the gen-
eral care floor [2]. It seems likely that serious postoperative
cardiopulmonary complications may also be preceded by
changes in HR, BP, and/or RR [3]. We therefore set out to
evaluate a patient surveillance system based on continual
monitoring of pulse oximetry, HR, RR, and non-invasive
* Timothy N. Harwood
tharwood@wakehealth.edu
1 Department ofAnesthesiology, Wake Forest
University School ofMedicine, Medical Center Blvd.,
Winston-Salem27157-1009, NC, USA
2 Department ofNursing, North Carolina Baptist Hospital,
Winston-Salem, NC, USA
3 Department ofAnesthesiology, Wake Forest University
School ofMedicine, Medical Center Boulevard,
Winston-Salem, NC27157-1009, USA
Journal of Clinical Monitoring and Computing
1 3
blood pressure (NIBP) that employs direct nursing notifica-
tion of violation of alarm limits via a wireless phone.
A previous study demonstrated the benefit of continual,
multi-parameter VS monitoring in general non-ICU patients
[4]. One of the concerns about implementing multi-param-
eter continual VS monitoring on a general medical or med-
ical-surgical unit is alarm fatigue. The high rate of alarms,
and their sometimes limited clinical relevance has been well
documented in the operating room (OR) [5] and intensive
care unit (ICU) [6, 7], and the concern about alarm fatigue
is real [5, 6, 8]. In a large study in both the ICU and general
medicine ward of a major children’s medical center, investi-
gators determined that most alarms were non-actionable, and
response time increased as non-actionable alarm exposure
increased [7]. Alarm fatigue could explain these findings. In
general medical units, the nurse to patient ratio is lower than
in the OR and ICU, so continual monitoring in this environ-
ment has the potential for creating a significant burden on
nurses.
The alarm limits used for continual monitoring on the
general care floor may be different than alarm rates used in
OR and ICU settings. Alarm limits in the OR are appropri-
ate in a 1:1 procedure care setting when a provider’s full
attention can be directed to identifying and responding to
legitimately dangerous conditions. However, in a general
care setting with a 1:5 nurse to patient ratio, an alarm redi-
rects nurse attention from other important tasks, and a high
frequency of alarms will desensitize staff, potentially leading
to delayed responses. One approach to reduce the number of
non-actionable alarms is to combine alarm thresholds with
annunciation delays (a delay between when an alarm thresh-
old has been crossed and when the monitor and network alert
is sounded or displayed) [3, 9–13]. This approach, based
on examining the distribution of VS measurements in the
general floor patient population, has been shown to have
the potential to minimize the alarm rate during continual,
multi-parameter VS monitoring [14].
Because of what we perceived were medication-related
respiratory and cardiac arrests occurring in some of our
non-ICU hospital units, one of the anesthesiologists (RW)
was assigned to oversee these adverse events and consider
monitoring interventions. After a review of several moni-
toring systems, we asked for demonstrations of VS alert
systems (VSAS) from two different companies. After the
demonstrations, we chose to deploy the Sotera VisiMobile
system in a short-term lease arrangement. Implementation
costs were borne primarily from hospital clinical equipment
and nursing education budgets. Sotera did provide technical
assistance in establishing wireless network communication
between their portable monitoring devices, their server, and
our electronic health record system.
We wished to evaluate two hypotheses: (1) using an
automated VSAS monitoring multiple VS would allow us
to “titrate” the VS settings (VS parameter and time) that
triggered alarms and alerted the nursing staff, and (2) use
of an electronic system would help reduce RRT calls and
ICU transfers. We then established a study of an electronic,
automated VS monitoring system in one nursing unit during
an evaluation of several networked VS monitoring systems.
This paper will also describe the method used to arrive at
alarm threshold and annunciation delays that were accept-
able to the clinical care team.
Once the alarm thresholds and delays were finalized, the
complete system was evaluated throughout an entire care
unit for 5 months to determine the daily per patient total
alarm rate for all VS and the level of acceptance by the non-
ICU nursing staff. The rate of patient deterioration events
was determined during the 5-month pilot of continual, multi-
parameter VS monitoring and compared to that for the same
unit during the previous 5-month period, when we employed
standard manual VS measurements (every 4h pulse check,
respiratory rate, and non-invasive BP monitoring obtained
by a nursing assistant). Pulse oximeters could be ordered
if desired but measurements were not more frequent than
every 2h.
No surveillance monitoring capabilities were available on
our general care units prior to the pilot. Centralized telem-
etry was available for physicians to order on appropriate
patients during the time prior to the pilot as well as during
the pilot. If the patient required cardiac monitoring, both
centralized telemetry and the surveillance monitoring sys-
tem were applied, although the usual proportion of patients
monitored in this fashion was less than 5%.
2 Methods
2.1 Implementation
The site of study was a 26-bed neurologic/neurosurgical
unit with an average of 200 patient-days and 53 patient dis-
charges per week. The nurse-to-patient ratio was 1:5 with a
mostly elderly population undergoing either postoperative
neurosurgical surveillance and standard acute neurologic
care with use of postoperative opioids and other centrally-
acting CNS suppressants. During the 6months antecedent
to study period, vital signs were checked manually and inter-
mittently by nurse aides as ordered by the admitting physi-
cian team. If patients’ vital signs fell into a specific range
determined by the Department of Nursing’s Early Warning
System (EWS), the nurse aide would contact the staff RN
and the rapid response team (RRT) would be called to evalu-
ate and treat the patient. For patients who exhibited risk for
hypoxemia, spot check SpO2 readings using portable pulse
oximeters could be ordered up to every 1h. Since VS data
entry into the EPIC EHR was performed by nursing staff,
Journal of Clinical Monitoring and Computing
1 3
no automated VS entry occurred and the EWS results were
evaluated by the staff nurse with manual implementation of
responses to high EWS scores.
During the period involving evaluation of automated VS
alert systems, we evaluated an FDA-cleared VSAS (ViSi
Mobile System, Sotera Wireless, Inc., San Diego, CA) on
all patients admitted to the same 26-bed Neurosurgery/
Neurology hospital unit (non-ICU) from January 16, 2015
to June 30, 2015. This system continually provided all VS
measurements (BP, HR, RR, SpO2) required for patient care.
The system was connected to the hospital’s existing wireless
network to enable VS information and alarms to be distrib-
uted to a central station and to the nurses’ hospital-supplied
phones (ASCOM, Morrisville, NC). We provided training
to the approximately 23 nurses and 20 nursing assistants
covering all shifts in the study unit. This training included
in-service training on VSAS device use for which the man-
ufacturer provided technical instructors. Clinical safety
personnel in our institution conducted a discussion of the
problem of unrecognized deteriorations, a description of the
alarm threshold policy, and provided two weeks of daily
rounding to identify and correct problems. No additional
staff was added to the existing care team.
The study leadership team (composed of two anesthesi-
ologists and two nurse managers in technical consultation
with representatives of the manufacturer) used an iterative
process to determine alarm threshold limits and annunci-
ation delay (alarm settings) for each VS. Additionally, if
stable patients were found to have VS that were normally
outside of the set parameters, the physician was notified and
the alarm parameters were personalized for that patient to
avoid non-actionable calls to the nursing staff.
For the purposes of evaluating alarm frequency rates dur-
ing the course of the pilot, all VS data were sent to a third-
party cloud database. Our clinicians had access to the cloud
data and used it to determine the alarm rate for the alarm
settings used in the pilot, but also to simulate and provide
interactive feedback on the impact changes in alarm thresh-
old and annunciation delay would have on total alarm rate
[14]. Based on nursing feedback and data analysis every
few days, the clinical manager team then initiated settings
in an effort to reduce the overall alarm burden. Our goal
was to reduce the alarm rate with a soft target of 2alarms/
patient/day (Table1). Although hard evidence does not
exist to establish this as a universal target for hospitalized
patients, our clinical managers in concert with the nursing
staff expressed this target as a manageable but safe goal. This
team evaluated 4 iterations of alarm settings during the first
month of the pilot study before arriving at the final alarm
settings (Table2).
2.2 Data collection
This VSAS device uses proprietary code to sense artefactual
signals and reduce false readings. For example, for SpO2,
if a poor or loss of signal occurred, data was transmitted to
the server as “xx”. The software is setup to then reset the
sequential counting of alarm periods when this occurs. For
example, if an SpO2 of 84% occurred during a 15s epoch
but was followed by an artefact in the next 15s epoch, the
alarm sequence would not restart until another out of range
data point occurred. This was setup to reduce false alarms
with the hope that we would still not miss many true out of
range data points.
Table 1 Overall description and comments regarding alarm parameter changes by time period
APD alarms per patient per day
Date instituted Description Comments APD
1/13/2015 to 1/15/2015 Go live N/A 11.41
1/16/2015 to 2/3/2015 Modify configuration to VSAS defaults Overall reduction of non-actionable alarms 7.09
2/4/2015 to 2/9/2015 Increase high systolic BP alarm limit Reduce non-actionable NIBP alarms—this population kept at
higher systolic
5.94
2/10/2015 to end of study Increase BP and SpO2 alarm delays Further reduction of non-actionable alarms and optimization with
extension distributive alarming
2.01
Table 2 Final vital signs parameters
SpO2 pulse oximeter oxygen saturation, HR heart rate, PR pulse rate,
RR respiratory rate, SBP systolic blood pressure, MAP mean arterial
pressure
Alarm Threshold Delay (s)
SpO2 low 85 90
HR high 150 15
HR low 39 15
PR high 150 60
PR low 39 60
RR high 35 120
RR low 4 120
SBP high 200 240
MAP low 58 60
Journal of Clinical Monitoring and Computing
1 3
The system monitored a total patient of 61,823h with a
maximum session length of 204h and median session length
of 40h. The stored parameter value for each 15-s record was
the median calculated during the 15s of original data.
We also collected patient outcome data: RRT calls, ICU
transfers, and unexpected deaths (not compassionate or hos-
pice care). In addition, we sampled the nursing staff and
patients regarding their impressions about usability of the
VSAS devices and system.
2.3 Data analysis
We analyzed data at daily to weekly intervals during the imple-
mentation of the VSAS on the study nursing unit. The entire
monitoring period ran from January 13, 2015 to June 30, 2015.
The pre-study comparison time period prior to the intervention
of monitoring ran from August 1, 2014 to December 31, 2014.
We compared this site with another other unit that cared for
surgical patients for purposes of trend analysis.
We evaluated our choice of alarm scenario by analyzing the
following variables: frequency of alarm, post-alarm VS at each
15-s interval for 5min, RRT calls, and transfers to the ICU.
All alerts were reviewed, and only alarms meeting the
trigger criteria described earlier were recorded in the data-
base as rescue events. For comparison purposes, we tracked
rescue events per 1000 discharges for patients wearing
VSAS units. We tracked transfers to the ICU as transfers
per 1000 patient-days for all units (as the most commonly
used denominator for patient transfers). We compared all
parametric data outcomes with Students’ t tests, with the
Mann–Whitney test for nonparametric data examining
medians, and with Chi-Square for proportional analysis. We
considered an alpha level of 0.05 as statistically significant.
The statistical software we used was MatLab (MathWorks,
Natick, Massachusetts, USA).
We considered the possibility of hospital care trends that
could affect our dependent variables (e.g., higher awareness
of when to call RRT, etc.) To analyze for temporal patient
monitoring or nursing care trends that could simultane-
ously occur during the study period, we analyzed the same
events in a similar non-ICU medical ward during the same
pre-study and study periods. In that the other unit did not
compare closely to our study unit in terms of patient types,
the acuity levels were similar as measured by the Charlson
Comorbidiy Index (CCI).
3 Results
3.1 Demographics
Patient demographics including age, gender, and acuity level
were similar between the test periods (Table2).
3.2 Vital sign/alarm results
We analyzed a total of 736 patients and accumulated over
30,000h of monitoring, averaging 41h per patient. In terms
of oxygen saturation (SpO2) estimation, the estimated SaO2
from the pulse oximeter produced unusable data in 7.4% of
epochs. Nine hundred and sixty of 1675 (57.3%) of sessions
demonstrated an alarm at some point during the monitor-
ing periods. To a large extent (> 99% of all alarms), alarms
occurred for singular parameters out of range. By default, if
SpO2 data error occurred during measurements (artifact or
poor signal resulting in an inability to calculate the SpO2),
the alarm parameter would restart timing commencing with
the next good signal. Thus, any alarm would be based upon
acceptable signal quality, and would avoid “false” alarms.
The largest number of alarm episodes involved HR, followed
by SpO2.
3.3 Establishing alarm parameters
We set alarm parameters initial to the monitor’s default
mode and adjusted limits for SpO2 and time spent below the
low limit prior to achieving alarm state. Our initial effort
in implementation of the alert system was to limit alarms/
patient/day (APDs) to what our clinical staff regarded as
unobtrusive yet still sensitive enough to avoid false negatives
(significantly altered VS without alarms).
Regarding the monitoring optimization, the process was
that every few days, the clinical managers reviewed alarm
data that resulting from prior thresholds and delays. For each
parameter, the APDs werereviewed, and then the impact of
changes on either threshold or delay were modeled by our
data analysts (with the impact of changes on the prior total
VS dataset). The clinical managers then discussed whether
widening parameters would create a meaningful reduction
of alarm rates and be clinically acceptable considering vari-
ous deterioration scenarios. In other words, would widening
the alarm window still be likely to detect and allow nurses
to intervene on a patient before serious harm would occur.
The managers considered that hypotension, bradycardia,
and hypoxemia would be tolerated for a much shorter time
than tachycardia or hypertension, unless the tachycardia
resulted in hypotension. We usually considered the scenario
in which we required delay before alarm enunciation, and
then a further 90-s delay before escalating via phone alert.
With our historical VS monitoring consisting of every 4h
checks, we regarded that even these apparently wider alarm
limits and delays would still be more likely to pick up dete-
rioration compared to the prior standard.
At the initiation of the study we considered the APDs too
frequent (> 11 APDs) (see Table1.) After day 3, our clini-
cal managers agreed to liberalize the alarm settings. Over
the next 3–4weeks we titrated the alarm limits to gradually
Journal of Clinical Monitoring and Computing
1 3
more acceptable APD results. We considered 2 overall APDs
as desirable. Anecdotally, we simultaneously determined
that false “no alarms” were not occurring on a regular basis.
Our final alarm scenario (see Tables1, 2) resulted in
an overall alarm rate of 2.01alarms/patient/day (APD).
Approximately half of these were from SpO2 alarms (0.97
APD) (Table3). We focused on the oxygen desaturation
results in this project because that was the primary motiva-
tion for our pilot study. Overall, hypoxia (oxygen desatura-
tion: SpO2 < 85%, > 90s) occurred in approximately 0.4% of
15-s epochs recorded, and thus was infrequent in occurrence,
although this resulted in 114h of desaturation in our cadre
of over 700 patients.
Regarding an impression that we were “pushing the limit”
of SpO2 alarms, detailed analysis of post-alarm parameters
indicated that most oxygen saturation levels at 1min after
alarm were higher. Only in 0.6% of SpO2 alarm cases did the
SpO2 not rise above the alarm limit at 1-min post-alarm, and
no cases had a 5-min post-alarm SpO2 less than the alarm
limit of 85%.
Since nursing records were not integrated with the alarm
system, cause and effect cannot be determined. However, we
surmise that this indicates either timely self-rescue or nurs-
ing/healthcare worker rescue. In addition, nursing staff found
the alarm rate to be acceptable in terms of their workload.
3.4 Patient rescue/transfer outcomes
Overall, length of stays (LOSs) were similar between pre-
pilot and intra-pilot study periods (Table4). Rapid Response
Team events during the study time decreased significantly
after implementation of the VSAS (Table5). Transfers to the
ICU also declined, albeit insignificantly, after implementa-
tion of the system. In our comparison unit examining tempo-
ral trends during the same time period, the comparison unit’s
RRT event rates did not significantly change. As in the study
unit, the comparison unit’s ICU transfers also decreased,
although insignificantly.
3.5 Patient deaths
Observed deaths during the VSAS period dropped both in
absolute and relative terms but were not statistically differ-
ent (Table5.) These deaths include those that occurred on
the non-ICU ward and after transfer to ICU. Deaths were
included only if they were not categorized as “Compassion-
ate Care”, otherwise known as In-Hospital Hospice care.
These decreases were not significant statistically in the num-
ber of deaths per 1000 discharges.
4 Discussion
The final stage of implementation of this VSAS continuously
monitored pulse oximetry, HR, RR, NIBP and resulted in a
total alarm rate of 2.3alarms/patient/day which was accepta-
ble to the nursing staff. Patient outcomes during the 5-month
evaluation of the system were compared to outcomes in the
same unit in the 5months preceding the study, where vital
signs were intermittently monitor. The primary finding is
that during the period with continuous monitoring of all vital
signs with the VSAS, we discovered fewer RRT calls and a
decreased need to escalate care to the ICU.
Table 3 Specific SpO2 alarm settings and results by time period
APD alarms per patient per day
Date Low threshold Low delay (s) SpO2 APD
1/13/2015 to 1/15/2015 85 30 1.05
1/16/2015 to 2/3/2015 85 30 2.92
2/4/2015 to 2/9/2015 85 30 2.91
2/10/2015 to end of
study
85 90 0.97
Table 4 Pre- and intra-pilot patient/unit characteristics
M/F male/female, LOS length of stay, W/B/O white, black, other, E/U/R emergency, unscheduled, routine, CCI Charlson Comorbidity Index
(proportions of index = 0,1,2,≥3)
Time period Patient
dis-
charges
Total
patient
days
Gender
(M%/F
%)
Age (year;
median/
SD)
LOS (days;
median/SD)
Race (W/B/O) Admit type (E/U/R) CCI (0,1,2,≥3)
Study unit
5months prior 889 5469 58/42 59.3/15.5 3.0/5.0 81/14/5 60/21/19 0.68/0.12/0.06/0.14
5months intra-pilot 1069 5662 54/46 60.5/14.7 3.0/5.0 80/14/6 63/27/11 0.73/0.09/0.05/0.13
Statistical analysis n/a n/a NS NS NS NS NS NS
Comparison unit
5months prior 1019 6022 57/43 61.3/14.5 6.04/6.87 83/11/6 55/36/9 0.43/0.22/0.08/0.26
5months intra-pilot 979 5108 52/48 60.1/15.5 5.32/5.09 85/10/5 53/37/10 0.42/0.22/0.09/0.22
Statistical analysis n/a n/a NS NS NS NS NS NS
Journal of Clinical Monitoring and Computing
1 3
Deployment of the VSAS was associated with a signifi-
cant drop of RRT calls from 189 to 158 per 1000 patient dis-
charges. In our 26-bed unit, this means an effect size change
from 70 events annualized. ICU transfers declined from 53
to 40 per 1000 discharges, although this was not statistically
significant.
We did detect a diminished death rate during the trial;
however, due to the low number of deaths and minor abso-
lute differences in deaths between the time periods, this did
not achieve statistical significance.
One of our concerns centered on the reduction of alarm
rates while still maintaining safety for our patients. The set-
ting of ≤ 85% SpO2 for > 90s results in an acceptable alarm
rate while maintaining the ability of nurses to rescue patients
who are experiencing episodes of postoperative hypoxemia.
Oxygen saturation levels after alarm onset were higher, indi-
cating either that patients improved on their own or that a
nursing/healthcare worker intervened.
The current standard of care for hospital inpatients is the
sampling of intermittent vital signs and clinical examina-
tions with additional condition monitoring for patients con-
sidered to be at high risk for adverse events. Low nurse-
patient ratios demand a different balance of sensitivity and
specificity when compared with higher staffing ratio sites
such as ORs or ICUs. Continuous patient surveillance can
only be successful if it is not an additional workload to
limited personnel resources while maintaining or improv-
ing efficacy in identifying patient deterioration. Many false
positive alarms (nuisance alarms) will lead staff to become
desensitized, as has been observed in other reports [7, 11].
In contrast, our current work demonstrates that, clinically
actionable alarms gained system acceptance and adoption by
the nursing staff, while providing identification of vital signs
changes that may have led to a decrease in rescue events.
Although the alarm limits are clearly different than those
typically used in the OR, ICU, or condition monitoring situa-
tion—and might appear counterintuitive—they are based on
the fundamentally different approach of triaging and surveil-
lance monitoring in the general care setting.
Staff on the implementation floor were aware of the
ongoing data collection, so that any changes due to the
Hawthorne effect (where performance may improve in
the company of observation) should be similar across
the dataset. The training received by the members of the
VSAS was limited to the use of the new technology and
did not introduce any new general interventional or diag-
nostic techniques. The most common nursing comment is
of a sense of increased knowledge about the status of their
patient-based on the SpO2 and HR information visible on
the in-room monitor (along with remote notification), rein-
forcing the likelihood that any increased nursing attention
is a direct result of the new system, not a by-product of the
guided implementation of the new process.
Monitoring systems other than continuous pulse oxime-
try such as those found in the system we studied, including
CO2 or RR monitoring, are commercially available. Evi-
dence exists that in procedural arenas, capnography, when
used in a 1:1 monitored environment, can help assure air-
way patency [15]. However, several studies with these
devices have found that patient tolerance and compliance
can be too low for them to be used as continuous monitors
in the ward setting and evidence is not strong that they
result in the desired combination of low alarm rates and
reduced care escalation to RRT or ICU status [16–18].
Acoustic respiratory monitoring may offer promise but is
limited to a single VS parameter only [15]. Likewise, con-
tinuous pulse oximetry reports indicate conflicting study
results, resulting in questionable value of monitoring with
pulse oximetry as judged by improved reliable outcomes,
effectiveness, and efficiency [19].
Limitations of our study include a lack of granularity
concerning exact causes of the RRT calls and deaths. In
addition, although we include data from a “comparison
unit”, the surgical subspecialties and types of patients
on that unit vary significantly from our study unit. We
hope that aggregate outcomes indicate an improvement
in care, this is largely empirical since we did not conduct
a randomized, controlled trial examining the use of this
monitoring system. We consider the evidence presented
here as preliminary findings only, and understand that the
final judgement on employing these systems remains to
be achieved.
Table 5 RRT events, ICU transfers, unplanned deaths
RRT rapid response team, ICU intensive care unit
Time period Patient discharges RRT events/1000 dis-
charges
ICU transfers/1000 dis-
charges
Unplanned deaths/1000
discharges
Study unit Comp unit Study unit Comp unit Study unit Comp unit Study unit Comp unit
5months prior 889 1053 189 149 52.9 50.3 4.92 1.68
5 months pilot 1069 1000 158 139 40.2 43.0 2.60 1.04
Statistical significance (Z
score for two proportions)
n/a n/a P < 0.05
P = 0.036
NS NS
P = 0.09
NS NS NS
Journal of Clinical Monitoring and Computing
1 3
4.1 Conclusion
In conclusion, our results demonstrate that continuous
patient surveillance can detect alterations in VS, while main-
taining a low rate of alarms, and keep patient outcomes at
least as safe, if not safer, compared to standard intermittent
vital signs monitoring in a neurological/neurosurgical ward
setting. We plan to more fully implement VSAS monitoring
in our acute care settings, leading to larger-scale studies. We
plan further research on patient and event outcomes such
as LOS, urgent rescues by RRT, reductions in escalation
to ICU, and both in-hospital and post-discharge mortality.
Acknowledgements We would also like to thank the hard-working
nurses and other staff on the study unit that endured our staged imple-
mentation, fine-tuning of technical details, and providing feedback
during the study.
Compliance with ethical standards
Conflict of interest The authors have no conflicts of interest to dis-
close, financial or otherwise.
Ethical approval This study was approved by the Wake Forest Univer-
sity Health Sciences Institutional Review Board.
Informed consent Since no PHI information was stored, Informed
consent was waived by the WFUHS IRB.
References
1. Brown H, Terrence J, Vasquez P, Bates DW, Zimlichman E. Con-
tinuous monitoring in an inpatient medical-surgical unit: a con-
trolled clinical trial. Am J Med. 2014;127(3):226–32.
2. Taenzer AH, Pyke J, Herrick MD, Dodds TM, McGrath SP. A
comparison of oxygen saturation data in inpatients with low oxy-
gen saturation using automated continuous monitoring and inter-
mittent manual data charting. Anesth Analg. 2014;118:326 – 31.
3. Taenzer AH, Pyke JB, McGrath SP, Blike GT. Impact of pulse
oximetry surveillance on rescue events and intensive care unit
transfers: a before-and-after concurrence study. Anesthesiology.
2010;112:282–7.
4. Zimlichman E, Szyper-Kravitz M, Shinar Z, Klap T, Levkovich S,
Unterman A, Rozenblum R, Rothschild JM, Amital H, Shoenfeld
Y. (2012) Early recognition of acutely deteriorating patients in
non-intensive care units: assessment of an innovative monitoring
technology. J Hosp Med. 7:628–33.
5. de Man FR, Greuters S, Boer C, Veerman DP, Loer SA. (2013)
Intra-operative monitoring—many alarms with minor impact.
Anaesthesia. 68:804–10.
6. Görges M, Markewitz BA, Westenskow DR. Improving alarm
performance in the medical intensive care unit using delays and
clinical context. Anesth Analg. 2009;108:1546–52.
7. Drew BJ, Harris P, Zègre-Hemsey JK, Mammone T, Schindler D,
Salas-Boni R, Bai Y, Tinoco A, Ding Q, Hu X (2014) Insights into
the problem of alarm fatigue with physiologic monitor devices: a
comprehensive observational study of consecutive intensive care
unit patients. PLoS ONE. 9:e110274.
8. Bonafide CP, Lin R, Zander M, Graham CS, Paine DW, Rock
W, Rich A, Roberts KE, Fortino M, Nadkarni VM, Localio AR,
Keren R. (2015) Association between exposure to nonactionable
physiologic monitor alarms and response time in a children’s hos-
pital. J Hosp Med. 10:345–51.
9. Rheineck-Leyssius AT, Kalkman CJ. Influence of pulse oximeter
settings on the frequency of alarms and detection of hypoxemia:
Theoretical effects of artifact rejection, alarm delay, averaging,
median filtering or a lower setting of the alarm limit. J Clin Monit
Comput. 1998;14:151–6.
10. Schmid F, Goepfert MS, Franz F, Laule D, Reiter B, Goetz AE,
Reuter DA. Reduction of clinically irrelevant alarms in patient
monitoring by adaptive time delays. J Clin Monit Comput. 2015.
https://doi.org/10.1007/s10877-015-9808-2.
11. Graham KC, Cvach M. Monitor alarm fatigue: standardizing use
of physiological monitors and decreasing nuisance alarms. Am J
Crit Care. 2010;19:28–34.
12. Burgess LP, Herdman TH, Berg BW, Feaster WW, Hebsur S.
Alarm limit settings for early warning systems to identify at-risk
patients. J Adv Nurs. 2009;65:1844–52.
13. Welch J. (2011) An evidence-based approach to reduce nuisance
alarms and alarm fatigue. Biomed Instrum Technol Spring.
45(s1):46–52.
14. Welch J, Kanter B, Skora B, McCombie S, Henry I, McCombie
D, Kennedy R, Soller B. Multi-parameter vital sign database to
assist in alarm optimization for general care units. J Clin Monit
Comput. 2016;30:895–900.
15. Waugh JB, Epps CA, Khodneva YA. (2011) Capnography
enhances surveillance of respiratory events during procedural
sedation: a meta-analysis. J Clin Anesth. 23:189–96.
16. Richardson M, Moulton K, Rabb D, Kindopp S, Pishe T, Yan C,
Akpinar I, Tsoi B, Chuck A. (2016) Capnography for Monitoring
End-Tidal CO2 in Hospital and Pre-hospital Settings: A Health
Technology Assessment. Ottawa: Canadian Agency for Drugs
and Technologies in Health. http://www.ncbi.nlm.nih.gov/books/
NBK362374/.
17. Patino M, Redford DT, Quigley TW, Mahmoud M, Kurth CD,
Szmuk P. Accuracy of acoustic respiration rate monitoring in
pediatric patients. Paediatr Anaesth. 2013;23:1166–73.
18. Kodali BS. Capnography outside the operating rooms. Anesthe-
siology. 2013;118:192–201.
19. Pedersen T, Nicholson A, Hovhannisyan K, Møller AM, Smith
AF, Lewis SR. (2014) Pulse oximetry for perioperative monitor-
ing. Cochrane Database Syst Rev. 3:CD002013.
A preview of this full-text is provided by Springer Nature.
Content available from Journal of Clinical Monitoring and Computing
This content is subject to copyright. Terms and conditions apply.
