View as “PDF version”
Patient Safety Tip of the Week
New Early
Warning System Integrates Nurse Surveillance Patterns
A variety of early warning systems have been
utilized over the years to detect patients at risk of clinical deterioration.
Most have relied on trends detected using physiological data. Many have
utilized the Modified Early Warning Score (MEWS)
for Clinical Deterioration. Our March 2012 What's New in the Patient Safety
World column “Value
of an Expanded Early Warning System Score” highlighted
an expanded version of the MEWS that was introduced in the Netherlands in 2009.
A study (Smith
2012) reported the impact
of that score in predicting clinical deterioration in patients admitted to
general or trauma surgery wards. The tool included the basic parameters
included in earlier versions of the MEWS (heart rate, systolic BP, respiratory
rate, oxygen saturation, temperature, and level of consciousness) but added
some new parameters. One was urinary output. The other was a more subjective
parameter: the nurse’s level of concern about the patient’s condition.
We commented that, despite all the potential
merits of technological solutions, we liked the idea that the expanded MEWS in
the Netherlands study also used what we consider a most valuable measure: the
nurse’s bedside gestalt of the patient’s condition!
A new study amplifies the importance of adding
nursing observations to an early warning system (Rossetti
2025). The COmmunicating Narrative Concerns Entered by RNs (CONCERN)
early warning system (EWS) uses real-time nursing surveillance documentation
patterns in its machine learning algorithm to identify deterioration risk.
The researchers compared outcomes between
patients whose care teams were and patients whose care teams were not informed
by the CONCERN EWS. There were 60,893 hospital encounters at multiple sites (33,024
in the intervention group, 27,869 in the usual care group). Patients in the intervention
group had a 35.6% decreased risk of death (adjusted hazard ratio 0.64), an 11.2%
decreased length of stay (adjusted incidence rate ratio 0.91), a 7.5% decreased
risk of sepsis (aHR 0.93) and a 24.9% increased risk
of unanticipated intensive care unit transfer (aHR
1.25) compared with usual-care group encounters (all results statistically
significant).
The CONCERN EWS model
uses electronic health record (EHR) metadata (for example, date and time
stamps, and data type) of nursing surveillance activities. Their previous study
notes it identifies all-cause deterioration up to 42 hours earlier than
models reliant on only physiological indicators (Rossetti 2021). Therefore, CONCERN EWS can be used as
clinical decision support to make the care team aware of deterioration much
earlier so that more timely interventions can be performed.
To objectively measure and test nurses’
concern levels in predicting patient deterioration, the authors created a
machine learning-based predictive model that processes nurse surveillance
patterns from metadata of nurse-entered documentation, with a small additional
signal from natural language processing of ‘mentions’ of concern in nurses’ narrative
notes.
The tool uses nurses’ concern levels reflected
by nurses' increased surveillance, such as
1.
increased frequency of assessments (for example,
respiratory rate checked every 2 hours for a non-intensive care unit (ICU) acute
care floor patient)
2.
assessments performed at uncommon times (for example,
checking vital signs in the middle of the night for a non-ICU acute care floor
patient)
3.
nursing medication administration interventions, such
as not administering a scheduled medication when it is due (typically because
the patient is clinically unstable)
After calculating the nurses’ concern level,
the model assigns a categorical deterioration risk score of green (low), yellow
(increased) or red (high), updates the score hourly and presents the score to
care team members on the CONCERN clinical-decision-support EWS display in the EHR.
We love this concept. Utilizing nursing
patterns of care to reflect concerns that nurses have about patients can thus
be captured readily by CONCERN EWS to help identify early clinical deterioration
and improve patient outcomes.
Some
of our other columns on MEWS or recognition of clinical deterioration:
·
February 26, 2008 “Nightmares:
The Hospital at Night”
·
April 2009 “Early
Emergency Team Calls Reduce Serious Adverse Events”
·
December 15, 2009 “The
Weekend Effect”
·
December 29, 2009 “Recognizing
Deteriorating Patients”
·
February
22, 2011 “Rethinking Alarms”
·
March 15, 2011 “Early
Warnings for Sepsis”
·
October 18, 2011 “High
Risk Surgical Patients”
·
March 2012 “Value
of an Expanded Early Warning System Score”
·
September
11, 2012 “In Search of the Ideal Early Warning Score”
·
May 2013
“Ireland First to Adopt National Early Warning
Score”
·
September
17, 2013 “First MEWS, Now PEWS”
·
January
2014 “It MEOWS But Doesn’t Purr”
·
March
11, 2014 “We Miss the Graphic Flowchart!”
·
July 15,
2014 “Barriers to Success of Early Warning Systems”
·
November
11, 2014 “Early Detection of Clinical Deterioration”
·
February
2015 “Detecting Clinical Deterioration: Don’t
Neglect Clinical Impression”
·
April
28, 2015 “Failure to Escalate”
·
September
8, 2015 “TREWScore for Early Recognition of Sepsis”
·
October
2015 “Even Earlier Recognition of Severe Sepsis”
·
December
15, 2015 “Vital Sign Monitoring at Night”
·
June
2016 “An
EMR-Based Early Warning Score”
·
May 2018
“Pediatric
Early Warning System Fails”
·
May 26,
2020 “Early Warning Scores”
·
December
1, 2020 “An Early Warning System and
Response System That Work”
·
July
2021 “EPIC Sepsis Prediction Tool
Falls Short”
·
July 13,
2021 “The Skinny on Rapid Response
Teams”
·
December
2021 “Can AI Triage Postoperative
Patients More Appropriately?”
·
April
26, 2022 “Challenges with Early
Warning Systems”
References:
Smith T, Den Hartog D, Moerman
T, et al. Accuracy of an expanded early warning score for patients in
general and trauma surgery wards. British Journal of Surgery 2012; 99(2): 192-197
https://academic.oup.com/bjs/article-abstract/99/2/192/6140962?redirectedFrom=fulltext
Rossetti SC. Dykes PC.,
Knaplund C, et al. Real-time surveillance system for
patient deterioration: a pragmatic cluster-randomized controlled trial. Nature
Medicine 2025; Published: 02 April 2025
https://www.nature.com/articles/s41591-025-03609-7
Rossetti SC, Knaplund
C, Albers D, et al. Healthcare process modeling to phenotype clinician
behaviors for exploiting the signal gain of clinical expertise (HPM-ExpertSignals): development and evaluation of a conceptual
framework. J Am Med Inform Assoc 2021; 28, 1242-1251
https://academic.oup.com/jamia/article/28/6/1242/6149011
Print “PDF version”
http://www.patientsafetysolutions.com/