African Journal of Emergency Medicine (2012) 2, 103107

African Federation for Emergency Medicine

African Journal of Emergency Medicine

www.afjem.com
www.sciencedirect.com

Warning scores in triage Is there any point?

Les scores dalerte dans le triage: Est-ce vraiment utile?
Sean B. Gottschalk
a
b
c

a,*

, Chris Warner b, Vanessa C. Burch c, Lee A. Wallis

Department of Emergency Medicine, University of Cape Town, Cape Town 8005, South Africa
Delft Community Health Centre, Cape Town, South Africa
Department of Medicine, University of Cape Town, Cape Town, South Africa

Received 9 August 2011; revised 10 April 2012; accepted 10 April 2012

Available online 20 July 2012

KEYWORDS
Emergency Centre;
Warning score;
MEWS;
Triage

Abstract Introduction: The South African Triage Scale (SATS), a novel triage system for Emergency Centres, was initially proposed in 2006. The system incorporates an adapted version of the
Modied Early Warning Score (MEWS).
Methods: A prospective study was conducted to evaluate the use of the MEWS as a triage tool in
EC settings in the Western Cape, South Africa. A total of 1867 cases were prospectively assessed.
The MEWS was correlated with Emergency Centre outcome
Results: The data show clear potential for use of the MEWS as a triage instrument for medical
patients. Its use for trauma cases is more limited.
Conclusion: The MEWS in its un-adapted form is unsuitable as a unied triage scoring system for
both medical and trauma cases in Emergency Centres.
2012 African Federation for Emergency Medicine. Production and hosting by Elsevier B.V. All rights
reserved.

* Corresponding author. Address: Department of Emergency Medicine, University of Cape Town, Ravenscraig Road, Green Point, Cape
Town 8005, South Africa. Tel.: +27 835800619.
E-mail address: chalk1@cybersmart.co.za (S.B. Gottschalk).
Peer review under responsibility of African Federation for Emergency
Medicine.

Production and hosting by Elsevier

2211-419X 2012 African Federation for Emergency Medicine. Production and hosting by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.afjem.2012.04.004

104

S.B. Gottschalk et al.

KEYWORDS

Abstract Introduction: Lechelle de triage dAfrique du Sud (SATS), un nouveau syste`me de triage pour les centres durgence, a eteinitiallementproposee en 2006. Le syste`meinte`greune version
adaptee du syste`medalerteprecoce de modication (CHME).
Methodes: Uneetude prospective a eterealiseeandevaluerlutilisation du CHME en tantquoutil
de triage dans les centres durgenceetablisdans la province du Cap Occidental, en Afrique du
Sud. Un total de 1 867 cas a eteevalueprospectivement. La notation du CHME etait en correlationavec lesresultats des centres durgence.
Resultats: Les donneesmontrentunpotentielevident pour lutilisation de la notation du CHME en
tantquoutil de triage pour les patients en medecinegenerale. Son utilisation pour les cas de traumatismesest plus limite.
Conclusion: Le CHME, danssaforme non-adaptee, ne convient pas en tantquesyste`me de notation
de triage unie a` la fois pour la medecinegenerale et pour les cas de traumatismesdans les centres
durgence.

Emergency Centre;
Warning score;
MEWS;
Triage

2012 African Federation for Emergency Medicine. Production and hosting by Elsevier B.V. All rights
reserved.

African relevance







MEWS is easy to calculate.

MEWS uses simple parameters only.
Basic equipment only is required for the score calculation.
Junior staff are able to calculate the score.
MEWS has been incorporated in the South African Triage
Scale.

Whats new?

 MEWS is useful for medical Emergency Centre (EC) triage.

 MEWS is predictive of hospital admission from the EC
(medical cases).
 MEWS is not useful for trauma EC triage.
 MEWS cannot be used for combined (medical + trauma)
EC triage.
 MEWS needs to be adapted for combined EC triage purposes (medicine + trauma).
Introduction
Triage in South Africa
Pre-hospital ambulance personnel in South Africa have been
using a four-level colour-based triage system for many years.1
The system advocates prioritisation based on physiological
instability. Wide discrepancies in triaging were common, however, owing to lack of robust parameters dening instability
and varying levels of personnel clinical expertise.2 Similarly,

Table 1

in hospital Emergency Centres (ECs) there was no standardised triage system; decisions regarding urgency of care were left
to nursing staff and even receptionists (rst point of contact in
ECs). Given these limitations, the Cape Triage Group (later
the South African Triage Group) set about developing a
simple triage tool for use in the pre-hospital and EC setting.3
The South African Triage Scale4 uses a two-stage stepwise
approach to assigning triage priority colour; this is based on
simple physiological parameters as well as a list of readily identiable clinical presentations (e.g. chest pain). The physiological parameters used are based on the Modied Early Warning
Score (MEWS) (Table 1). The MEWS has been shown to identify medical in-patients at risk of acute deterioration and inhospital death57 as well as medical patients requiring admission.8 A concern regarding the MEWS as the basis for a triage
tool is that it has not been adequately evaluated in ECs admitting both trauma and non-trauma (incl. medical and surgical)
emergencies.9 In a recent study it was shown that MEWS did
not add any additional information to the Manchester Triage
Score10 when used in the EC setting; however, a comparison of
the performance of the MEWS in trauma and non-trauma patients was not reported.11
The purpose of this study was to evaluate the utility of the
MEWS in an EC setting, analysing its predictive ability in
trauma and non-trauma patients.
Methods
Study design and setting
A prospective observational study was conducted over a
2-month period in the ECs of two privately funded hospitals

Modied Early Warning Score (MEWS).

Systolic BP
Heart rate
Respiratory rate
Temperature
AVPU score

<70

7180
<40
<9
<35

81100
4150

101199
51100
914
3538.5
Alert

1
101110
1520
Reacting to voice

2
P200
111129
2129
P38.5
Reacting to pain

3
P130
P30
Unresponsive

Warning scores in triage Is there any point?

105

in Cape Town, South Africa.12 Both hospitals admit both trauma and non-trauma patients to a single EC setting. MEWS on
EC admission was correlated with EC outcome (in-hospital
ward admission or discharge) for both trauma and non-trauma
cases.
Study population
Patients were eligible for inclusion in the study if they were
16 years or older and presented as rst visit. Follow-up cases,
direct referrals for hospital admission and patients booked for
minor surgical procedures were excluded. Data were prospectively collected over 2 months.
Data collection
Nursing staff captured the following information on admission
to the EC: respiratory rate, systolic blood pressure (DINAMAP, GE Healthcare, ST Giles, UK), pulse, temperature
(Sure-Temp, Welch-Allyn, USA) and level of consciousness
(standard AVPU scale). MEWS was calculated based on these
physiological parameters (Table 1). The primary endpoints of
the study were EC discharge or admission to hospital, or death
in the ED.
Statistical analysis
An internet database site was developed enabling multiple
data-entry sites using a web-based interface. Data les were reviewed by the principal investigator to check data capture
completeness. Data were imported into Excel (Microsoft corporation, USA) and statistical analysis performed using Statistica V7 (Statsoft, Tulsa, USA).

MEWSs were analysed in four endpoint subgroups: trauma

admissions, trauma discharges, non-trauma admissions and
non-trauma discharges.
The KrusskalWallis (KW) test was used to determine if
there was a difference between any of the four subgroups. A
p-value of 0.05 or less was considered to be signicant.
The MannWhitney U (MWU) test was then used to compare each of the four subgroups against one another. A p-value
of 0.008 or less was considered signicant (0.05 divided by 6 as
six tests performed comparing each of the subgroups against
one another).
Ethical considerations
Approval to conduct the study was granted by both the private
hospital group as well as the Human Research Ethics Committee of the University of Cape Town.
Results
Demographics
A total of 1875 case records were captured on the database.
Eight patients refused hospital treatment and were excluded
from the study leaving 1867 records for analysis (1175 hospital A and 692 hospital B). Fifty-one percent of the sample
was men; the median age was 39 years (range 1694). A total
of 74% of cases were between the ages of 2059 years. A
total of 562 cases (30%) were trauma-related EC visits.
Twenty-seven percent of the total sample was admitted.
There were 3 EC deaths which were grouped with the
admissions.

5.0

4.5

4.0

3.5

Score

3.0
2.7

2.5

2.0
1.8
1.6

1.5

1.0

1.1

0.5

0.0
Trauma D/C

Trauma Admit

Fig. 1

Non-Trauma D/C

Patient outcome vs. MEWS.

Non-Trauma Admit

106
Patient outcome vs. MEWS
The mean MEWS for trauma admissions was 1.8 vs. 1.1 for
discharges; the non-trauma admissions scored a mean MEWS
of 2.8 vs. a discharge mean of 1.6 points. The mean MEWS for
admissions was signicantly greater than for discharged
patients in both the trauma and non-trauma groups (p <
0.001). Comparison of the mean MEWS of trauma admissions
and non-trauma admissions shows that trauma cases scored
signicantly less than non-trauma (p < 0.001); this was also
the case comparing the trauma vs. non-trauma discharge
groups (p < 0.001). Furthermore, the mean MEWS of trauma
admissions was similar to that of non-trauma discharges from
the EC (p = 0.037) (Fig. 1).

S.B. Gottschalk et al.

 This study was done on cases age 16 or older. The MEWS is
not a paediatric score in its current format. A modied version is used in the child SATS.4
Conclusions
In the private sector setting in South Africa, MEWS predicts
in-patient admission from the Emergency Centre. This works
reasonably well for medical patients. It is less effective for trauma cases. In its original form MEWS is not suitable as a combined triage early warning tool.
Competing interests
None.

Percentage endpoints reached

The percentage endpoints achieved (i.e. admissions) climbed
progressively with an increasing MEWS. Endpoints reached
were 100% from a MEWS of 7 upwards.

Contributors
S.G. Conceived the idea, wrote the rst draft; C.W. technical data analysis; All authors contributed to the nal draft;
S.G. is the guarantor of the paper.

Discussion

Funding

The value of the MEWS in triage has been questioned,11 particularly as an addition to current systems such as the Manchester Triage Score (MTS).10 However, the MTS is being
used in a mature system with different stafng and case-mix
to our environment. The South African Triage Group set
out to design a concise triage tool which junior staff were capable of using, hence the interest in the MEWS.
In this study, the EC outcome was used as a proxy marker
of illness acuity. The original MEWS on EC presentation was
correlated with the EC outcome. The MEWS was originally
designed as a warning tool for medical in-patients5; its extension to non-trauma cases in the EC seems to be a viable option.
An under-triage rate of 15% in rural South Africa has recently
been reported for the MEWS in triage for cases requiring
admission.13 The MEWS was not originally designed to assess
trauma cases; it certainly seems to be less successful in this regard (overall underscoring). If used in a mixed EC environment as a lone triage tool, the trauma cases would be
underscored and therefore under-triaged in relation to the
medical cases. The tool can therefore not be used as a unied
instrument for both medical and trauma cases in triage in its
original format. However, a graduated increase in admission
rate with increasing MEWS is evident, an important quality
for a scoring system in a triage setting.
Despite potential as a medical triage tool, further work
needs to be done to adapt the MEWS to function effectively
in a mixed Emergency Centre environment.
Due to the weakness of the MEWS in its original form,
there appears to be a need to modify it for use in triage. The
MEWS has been signicantly modied in the SATS(4); further
improvements have been proposed and will be described in
future.

There was no external funding.

Limitations
 As outcome measures were grouped, denitive conclusions
may be difcult to draw from small numbers of critically
ill cases.

3. In its unaltered format, the MEWS can be used as a triage

tool for Emergency Centres
a. True
b. False

Acknowledgements
The Medi-Clinic Group and their respective heads of unit, Drs.
Bonner and Kow, for participation in this study.
Mr. A. Melzer of Eighty20 data management company for setting up the website and hosting the database website: http://
www.eighty20.co.za/ accessed December 2010.
Mr. A.R. Sayed, Senior Biostatistician, Faculty of Health Sciences, University of Cape Town for assistance on statistical
data analysis.
Welch-Allyn for sponsorship of Sure-Temp thermometers.

Appendix A. Short answer questions

1. Warning Scores are used primarily to detect deterioration
of the following group of patients:
a. EC trauma cases
b. All EC cases
c. Medical hospital in-patients
d. Medical staff
e. EC procedural cases
2. The MEWS in EC triage is shown to be a predictive tool in
the following:
a. EC Hospital admission of trauma cases
b. EC Hospital admission of medical cases
c. EC discharge of trauma cases
d. EC hospital admission of all cases
e. EC ICU admission of trauma cases

Warning scores in triage Is there any point?

References
1. Mac Mahon AG. Sorting out triage in urban disasters. SAMJ
1985;67(4):5556.
2. Gottschalk SB. Triage a South African perspective. CME SA J
CPD Emerg Med 2004;22(6):3257.
3. Gottschalk SB, Wood D, DeVries S, et al. The Cape triage score:
a new triage system for South Africa. Proposal from the Cape
triage group. Emerg Med J 2006;23(2):14953. http://dx.doi.org/
10.1136/emj.2005.028332.
4. Emergency Medicine society of South Africa homepage on the
Internet. Cape Town, SA; 2009 [cited February 2012]. South
African Triage Scale. Available from: <http://emssa.org.za/sats/>.
5. Subbe CP, Kruger M, Rutherford P, Gemmel L. Validation of a
modied early warning score in medical admissions. QJM
2001;94(10):5216.
6. Buist M, Berbard S, Nguyen TV, et al. Association between
clinically abnormal observations and subsequent in-hospital
mortality: a prospective study. Resuscitation 2004;62(2):
13741.

107
7. Goldhill DR, McNarry AF. Physiological abnormalities in early
warning scores are related to mortality in adult patients. Br J
Anaesth 2004;92(6):8824.
8. Burch VC, Tarr G, Morroni C. Modied early warning score
predicts the need for hospital admission and in hospital mortality.
Emerg Med J 2008;25(10):6748.
9. Bruijns SR, Wallis LA, Burch VC. A prospective evaluation of the
Cape triage score in the emergency department of an urban public
hospital in South Africa. Emerg Med J 2008;25(7):398402.
10. Manchester Triage Group. Emergency triage. Manchester: BMJ
Publishing Group; 1997.
11. Subbe CP, Slater A, Menon D, et al. Validation of physiological
scoring systems in the accident and emergency department. Emerg
Med J 2006;23(11):8415.
12. Gottschalk SB. Evaluation of the modied early warning score as
a triage instrument in the Western Cape private health care setting.
University of Cape Town, M.Phil. degree; 2007.
13. Rosedale K, Smith AZ, Wood D. The effectiveness of the South
African Triage Score (SATS) in a rural emergency department.
SAMJ 2011;101:53740.