Blackwell Science, LtdOxford, UKJEPJournal of Evaluation in Clinical Practice1356-1294Blackwell Publishing Ltd 2004112133138Original ArticlePatient information system and MEWSC.P.J.

Quarterman
et al.

Journal of Evaluation in Clinical Practice, 11, 2, 133–138

Use of a patient information system to audit the introduction of

modified early warning scoring
C. P. J. Quarterman,1 A. N. Thomas MB BS FRCA,2 M. McKenna BA (Hons)3 and R. McNamee MSc PhD4
1
Medical Student, University of Manchester, Manchester, UK
2
Consultant in Intensive Care, University of Manchester, Manchester, UK
3
Electronic Patient Record Project Manager, University of Manchester, Manchester, UK
4
Senior Lecturer in Biostatistics, University of Manchester, Manchester, UK

Correspondence Abstract
Dr Antony N. Thomas Modified early warning scoring (MEWS) uses abnormalities in routine
Intensive Care Unit
observations to identify patients at risk of critical illness. Nurses recorded
Hope Hospital
Salford M6 8HD scores at or above the medical response score of 3 on a hospital clinical
UK information system during the first year of introducing MEWS to 10 wards
E-mail: Tony.Thomas@srht.nhs.uk in a university hospital. A total of 619 triggers were recorded in 365 patients.
Fifty-nine required intensive care unit (ICU)/high dependency unit (HDU)
Keywords: audit, critical illness, early
warning scoring, information systems, care; 71 died. Survival was significantly worse for initial scores > 4 (35/104
physiological deterioration, vital signs patients died) than for scores 3–4 (P < 0.004). Multivariant analysis showed
age (P < 0.001) and trigger score (P < 0.001) but not ward specialty
Accepted for publication: (P = 0.1) predicted death. Mean ages of survivors and non-survivors were
31 March 2004
64 years (SD 18) and 74 years (SD 17), respectively. Addition of a score for
age did not significantly increase the area under a receiver operator char-
acteristic curve for the predictive value of MEWS scores. The study shows
that increasing MEWS score is associated with worse outcome across a
range of specialties and that nursing staff will use a patient information
system to audit MEWS scores.

ical care services (Department of Health 2000). Their

Introduction
recommended scoring system is similar to that set out
Intensive care admissions and in-hospital cardiac in Table 1. We aimed to use a clinical information sys-
arrests are often preceded for hours or even days by tem to audit and support the introduction of a mod-
gross abnormalities in routine observations (Schein ified early warning scoring (MEWS) system into a
et al. 1990; Smith & Wood 1998; Goldhill et al. 1999; number of medical and surgical wards in a university
Hillman et al. 2002). Modified early warning scoring teaching hospital. We also aimed to use data
(Stenhouse et al. 2000) attributes scores to these recorded in the system to investigate the relationship
abnormal observations and requires nursing staff to between patient outcome, trigger score age and med-
summon help when a trigger score is reached. This ical speciality.
gives us a theoretical opportunity to intervene to
avoid intensive care unit (ICU) admissions or in-
Method
hospital cardiac arrests or to improve the outcome if
ICU admission is inevitable (McQuillan et al. 1998). In early 2002 we started to introduce an MEWS at
This process was strongly supported in ‘comprehen- Hope Hospital, Salford, an 885-bedded inner-city
sive critical care’ by the expert panel commissioned teaching hospital. The system was introduced without
by the National Health Service (NHS) to review crit- additional funding, and was championed by a steer-

© 2005 Blackwell Publishing Ltd 133

C.P.J. Quarterman et al.
Table 1 Scores allocated to bedside observations
3 2
Airway Threatened
Respiratory rate <6 6–8 9–17
Systolic blood pressure (mmHg) <70 71–80 81–100 101–199
Heart rate (bpm) <40 40–50 51–100
AVPU score Alert
Temperature (∞C) <35.0 35.0–38.0
Urine output (mL h-1) <10 <35
AVPU is a simple score of level of consciousness, defined as awake, responding to verbal stimulus, responding to pain and unresponsive. The lower
limit for the first point of respiratory rate was increased from 14 to 18 after the third month of the project. A trigger score of 3 was taken as the trigger
score for the nursing staff to summon medical assistance.
ing group of consultants and senior nurses from med-
ical, surgical and critical care disciplines as well as the
resuscitation officer and the critical care manager.
Over the next 12 months (June 2002 to May 2003) the
system was introduced into three medical, four sur-
gical and three orthopaedic wards. Training was pro-
vided to nursing staff in teaching sessions held in the
overlap between early and late shifts. Medical staff
received induction training about MEWS and addi-
tional training in speciality teaching sessions; in addi-
tion, opportunistic ward teaching was provided to
all staff when possible by members of the steering
group. Nursing staff were trained to score observa-
tions using the scoring system set out in Table 1. They
were also asked to call the house staff when the
MEWS score was 3 or more. The failure of house
staff to correct the score required them to refer up to
more senior members of staff within their team; these
staff could then contact the ICU registrar or consult-
ant if required.
We decided to audit the introduction of MEWS
using our patient information system [Sunrise Clini-
cal Manager 3.03 (2002), iSoft Manchester, UK]. This
system allows staff to review results of tests and
investigations as well as standard documents and
patient demographics. The system is available from
terminals on all wards. It was introduced in August
2000 and all ward medical and most nursing staff
used the system regularly. The decision to use this
patient information system was based partly on a
lack of funding to provide any staff to collect audit
data. We also felt the system would allow cross-
134
Score
1 0 1 2 3
18–20 21–29 >30
>200
101–110 111–129 >130
Verbal stimulus Pain response Unresponsive
38.1–38.5 ≥38.6
referencing of recorded trigger scores with other
information collected by the system; for example,
patient demographics, length of stay, transfer
between wards and patient outcome. Staff entering a
trigger score on the system carry out three steps to
get to an MEWS significant event field as shown in
Figure 1. This field then allows them to classify an
MEWS trigger score as a number between 3 and 6 or
more than 6. The opportunity also exists for ward
medical staff or ICU staff to log their attendance.
There is a free-text field and staff can also review pre-
vious MEWS trigger score entries.
The system was used to develop monthly reports
detailing number of trigger scores per ward and out-
come by ward as well as other information about
ICU and high dependency admissions and times that
MEWS scores were entered. These reports were cir-
culated to individual wards and consultants to allow
monthly feedback on progress. Free-text fields were
reviewed and details of all individual patient entries
were sent to the patient’s ward manager and special-
ity clinical governance lead, so they could then act on
organizational failings reviled by the entries. After
12 months, and with ethics committee approval, it
was decided to review all trigger score entries logged
into the system; this review forms the basis of the
results presented below. The information system was
interrogated using standard query language to iden-
tify all MEWS entries by ward, time of entry and
patient demographics and outcome; free-text entries
were also retrieved. Anonimized information was
then exported into an Access database for subse-
© 2005 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice, 11, 2, 133–138
 Patient information system and MEWS

Figure 1 Screen print to show options for recording of trigger scores or attendance by ward or intensive care unit (ICU)
medical staff. There is also a section to record free text. Three simple steps are required to obtain this screen.

quence statistical analysis using an SPSS program
(SPSS 11.5, SPSS Inc, Chicago, IL).
Where a patient was admitted on more than one
occasion, these admissions were classified as separate
patient episodes. Significant end-points were defined
as death, admission to HDU or ICU and hospital
discharge.
Descriptive analysis of data was performed fol-
lowed by production of Kaplan Meier survival curves
to allow comparison of outcome by grouped trigger
scores (3 or 4, or 5 or above). Outcome was defined
as death or live discharge, with the time taken from
first score to outcome. Multivariant analysis was
also performed using the Cox proportional hazards
model to look at the individual risk factors of trigger
score, age and medical speciality by ward as pooled
variables to determine which factors predicted mor-
tality. For this analysis trigger scores were treated as
individual scores except for scores above 6, which
were grouped as a single score. Medical specialty was
defined as medicine, general surgery or orthopaedics,
depending on the ward where the patient triggered
their score. Age was determined in years. The predic-
tive value of combining trigger scores with age was
then investigated by grouping ages into less than
50 years (0 point), 50–70 years (1 point), or over
70 years (2 points), producing receiver operator
characteristic (ROC) curves and then calculating the
area under these curves with and with out age as a
covariable, using similar method to previous pub-
lished work (Subbe et al. 2001). Significance for all
tests was set at the normal 5% level.
Results
A total of 619 trigger scores were recorded in 365
admissions between 1 June 2002 and 31 May 2003.
In addition to this, there were only 21 entries made
by ward medical staff and 41 entries made by ICU
medical staff. Twenty-four patients remained in hos-
pital at the end of the study period and were there-
fore excluded from survival analysis. The mean ages
of survivors and non-survivors were 64 years (SD
17) and 74 years (SD 12), respectively; 63% of the
patients were male. Of the 341 patients followed to
hospital discharge, 34 patients required ICU care,
25 patients required HDU care and 71 patients
died.

© 2005 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice, 11, 2, 133–138 135
C.P.J. Quarterman et al.

Eighty-two per cent of nurse entries had additional
free-text information and 68% contained the individ-
ual physiological variables that allowed direct calcu-
lation of the early warning score. All scores of more
than 6 could be ascribed using free-text entries
recorded in the clinical information system. Com-
ments that were useful for audit were made in many
of the entries. Forty-three per cent of triggers scores
were recorded between 06:00 and 14:00, 49%
between 14:00 and 22:00, and 8% between 22:00 and
06:00.
Figure 2 shows the distribution of trigger scores for
the first and highest scores recorded per patient. Sev-
enty per cent of reported scores were 3 or 4 and 30%
were 5 or more. Further details of the number of
scores and survival of patients by ward are shown in
Table 2 together with the number of months that the
system was on use on each ward. Figure 3 shows the
relationship between trigger score and negative end-
points of death, ICU or HDU admission. Although
scores of 5 or above occurred in only 30% of patients,
these patients had 47% of all negative end-points.
Figure 4 shows Kaplan Meier survival plots for the
cumulative survival of patients with MEWS scores
of 3 or 4 compared with patients with scores of 5 or
more. Log rank tests showed a significant difference
between curves with an increasingly poor prognosis
for higher scores (P < 0.004). Multivariant analysis
suggested that increasing age (P < 0.001) and increas-
ing MEWS score (P < 0.001) were associated with

180

160 155

140
126 Initial
120 Maximum
Frequency

100
82 78
80 76
66
60

40 32
21 17 Figure 2 Distribution of trigger scores
20 10
5 8 for the first and highest scores
1 1 0 0 0 1 1 2 recorded per patient across all wards
0
3 4 5 6 7 8 9 10 11 12 during the study period. MEWS,
MEWS score modified early warning scoring.

70
% death
60 % ICU
% HDU
50
% Frequency

40

30

20

10 Figure 3 The relationship between

first trigger score and negative end-
0 points of death, intensive care unit
3 4 5 6 7 8 (ICU) or HDU admission. MEWS,
MEWS score modified early warning scoring.

136 © 2005 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice, 11, 2, 133–138
 Patient information system and MEWS

Table 2 The number of scores and survival of patients by ward with the number of months that the system was on use
on each ward, and the type of wards involved in the study

% mortality of trigger
patients (n = number of
Number of MEWS triggers patients)
Time of MEWS
Type of ward use (months) 3 and 4 5 or more 3 and 4 5 or more

Surgical 12 34 13 5.9% 30.8%

(Mixed sex, 25 beds) (n = 2) (n = 4)
Surgical 12 29 26 13.8% 23.1%
(Mixed sex, 25 beds) (n = 4) (n = 6)
Surgical 12 10 1 10% 0
(Mixed sex, 14 beds) (n = 1)
Orthopaedic 4 13 4 23.1% 25%
(Mixed sex, 25 beds) (n = 3) (n = 1)
Orthopaedic 12 28 20 14.3% 40%
(Mixed sex, 25 beds) (n = 4) (n = 8)
Medical 2 5 1 20% 0
(Female, 22 beds) (n = 1)
Medical 12 87 30 26.4% 40%
(Male, 23 beds) (n = 23) (n = 12)
Medical Admissions Unit, 2 26 8 11.5% 50%
(Mixed sex, 21 beds) (n = 3) (n = 4)
Orthopaedic 2 4 2 0 0
(Mixed sex, 17 beds)
Totals 70 236 105 17.4% 33.3%
(n = 41) (n = 35)

MEWS, modified early warning scoring.

1.2 poor prognosis. Patient speciality as defined by med-

icine, surgery or orthopaedics ward was, however,
1.0 not predictive of outcome (P = 0.1). Addition of the
Cumulative survival

MEWS score age score to the MEWS score increased the area
0.8 Lower line
under the curve for the ROCs from 0.623 to 0.697, a
Score of 5, 6 or
greater difference that was not significant (P = 0.1362).
0.6
Score of 5, 6 or
greater – censored
0.4 Discussion
Upper line
Score of 3 or 4
0.2 The study clearly demonstrates that a clinical infor-
Score of 3 or 4
– censored mation system can be used to gather information
0.0 about patients triggering MEWS scores and that this
0 100 200
Time from score to death (days) information can be used to audit the introduction of
MEWS. The study unfortunately suggests that medi-
Figure 4 Kaplan Meier survival plots for the cumulative
cal staff would require specific incentives to make
survival of patients with initial MEWS scores of 3 or 4
compared with patients with scores of 5 or more. MEWS, them use the system. We are unable to tell how many
modified early warning scoring. patients triggering MEWS scores did not have their
triggers entered on to the information system; how-
ever, the number of trigger scores were similar to
those reported in other studies in surgical wards (Pit-

© 2005 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice, 11, 2, 133–138 137
C.P.J. Quarterman et al.
tard 2003). This would suggest that compliance with
the system at least gave a reasonable sample size of
trigger scores.
The results of our study show a significant relation-
ship between trigger score and patient outcome. This
has previously been described in patients on a med-
ical admissions unit (Subbe et al. 2001), but we are
not aware of this having been observed across several
surgical and medical specialties. We could not con-
firm previous work that suggested that the predictive
value of MEWS triggers may be enhanced by com-
bining scores with the patient’s age (Subbe et al.
2001). We were also unable to demonstrate a predic-
tive value of medical speciality, although patients
triggering the system were much more common on
medical than on surgical wards.
Although recommended by the expert group and
embraced by many UK hospitals, the introduction of
MEWS and outreach services has been criticized as
lacking in an evidence base (Cuthbertson 2003).
Although audit data have suggested improvement in
outcome in surgical patients (Pittard 2003), this has
yet to be confirmed in medical patients (Subbe et al.
2003). Unfortunately the nature of the intervention
would make a randomized control trial of MEWS
almost impossible. Many colleagues are also so influ-
enced by individual patient episodes before and after
MEWS scoring that MEWS will continue to be
increasingly used in UK hospitals. In this situation
the collection of audit data about patient outcome
and comparison with different outreach services will
benefit patient care. Patient benefit will be maxi-
mized where these audit data are fed into a continu-
ous quality improvement process (Thijs et al. 1997;
Clemmer et al. 1999) to improve the systems of care
of critically ill patients on general wards. The feed-
back process should also be extremely important for
individual doctors who, with the introduction of shift
working, may have no other feedback about the deci-
sions that they have made concerning patient care
(Ende 1983). Regular feedback to ward staff about
outcomes of patients who have triggered MEWS will
also improve compliance with this process.
In summary, we have demonstrated a relationship
between MEWS trigger score and patient outcome,
and also that patient information systems may be
used in audit systems to help in the introduction of
modified early warning scoring.
138 © 2005 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice, 11, 2, 133–138
References
Clemmer T.P., Spuhler V.J., Oniki T.A. & Horn S.D.
(1999) Results of a collaborative quality improvement
program on outcomes and costs in a tertiary critical care
unit. Critical Care Medicine 27 (9), 1768–1774.
Cuthbertson B.H. (2003) Outreach critical care – cash for
no questions? British Journal of Anaesthesia 90 (1), 4–6.
Department of Health (2000) Comprehensive critical care –
a review of adult critical care services (report).
Department of Health, London.
Ende J. (1983) Feedback in clinical medical-education.
Journal of the American Medical Association 250 (6),
777–781.
Goldhill D.R., White S.A. & Sumner A. (1999) Physiolog-
ical variables and procedures in the 24 hours before ICU
admission from the ward. Anaesthesia 54, 529–534.
Hillman K.M., Bristow P.J., Chey T., Daffurn K., Jacques
T., Norman S.L., Bishop G.F. & Simmons G. (2002)
Duration of life-threatening antecedents prior to inten-
sive care admission. Intensive Care Medicine 28 (11),
1629–1634.
McQuillan P., Pilkington S., Allan A., Taylor B., Short A.,
Morgan G., Neilsen M., Barrett D. & Smith G. (1998)
Confidential enquiry into quality of care before admission
to intensive care. British Medical Journal 316, 1853–1858.
Pittard A.J. (2003) Out of our reach? Assessing the impact
of introducing a critical care outreach service. Anaesthe-
sia 58 (9), 882–885.
Schein R.M.H., Hazday N., Pena M., Ruben B.H. &
Sprung C.L. (1990) Clinical antecedents to in-hospital
cardiopulmonary arrest. Chest 98 (6), 1388–1392.
Smith A.F. & Wood J. (1998) Can some in-hospital cardio-
respiratory arrests be prevented? A prospective survey.
Resuscitation 37, 133–137.
Stenhouse C., Coates S., Tivey M., Allsop P. & Parker T.
(2000) Prospective evaluation of a modified early warn-
ing scoring score to aid detection of patients developing
critical illness on a surgical ward. British Journal of
Anaesthesia 84, 663.
Subbe C.P., Kruger M., Rutherford P. & Gemmell L.
(2001) Validation of a modified Early Warning Score in
medical admissions. Monthly Journal of the Association
of Physicians 94 (10), 521–526.
Subbe C.P., Davies R.G., Williams E., Rutherford P. &
Gemmell L. (2003) Effect of introducing the Modified
Early Warning score on clinical outcomes, cardio-
pulmonary arrests and intensive care utilisation in acute
medical admissions. Anaesthesia 58 (8), 797–802.
Thijs L.G., Baltpoulos G., Bion J.K. et al. (1997) Continu-
ous quality improvement in the ICU: general guidelines.
Intensive Care Medicine 23 (1), 125–127.