burns 37 (2011) 12881295

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/burns

Review

Prognostic scoring systems in burns: A review

N.N. Sheppard *, S. Hemington-Gorse, O.P. Shelley, B. Philp, P. Dziewulski
St. Andrews Centre for Burns and Reconstructive Surgery, Broomfield, Chelmsford, United Kingdom

article info

abstract

Article history:

Survival after burn has steadily improved over the last few decades. Patient mortality is,

Accepted 19 July 2011

however, still the primary outcome measure for burn care. Scoring systems aim to use the
most predictive premorbid and injury factors to yield an expected likelihood of death for a

Keywords:

given patient. Age, burn surface area and inhalational injury remain the mainstays of burn

Burns

prognostication, but their relative weighting varies between scoring systems. Biochemical

Prognosis

markers may hold the key to predicting outcomes in burns. Alternatively, the incorporation

Scoring system

of global scales such as those used in the general intensive care unit may have relevance in

Mortality

burn patients. Outcomes other than mortality are increasingly relevant, especially as
mortality after burns continues to improve.
The evolution of prognostic scoring in burns is reviewed with specific reference to the
more widely regarded measures. Alternative approaches to burn prognostication are
reviewed along with evidence for the use of outcomes other than mortality. The purpose
and utility of prognostic scoring in general is discussed with relevance to its potential uses in
audit, research and at the bedside.
# 2011 Elsevier Ltd and ISBI. All rights reserved.

Contents
1.
2.

3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Materials and methods . . . . . . . . . . . . . . . . . . . . . .
2.1. Abbreviated burn severity index (ABSI) 1982
2.2. Cape Town modified burns score 1998 . . . . .
2.3. Ryan 1998. . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4. Burd 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5. Belgian outcome in burn injury (BOBI) 2009 .
2.6. Laboratory-based prognostic scoring . . . . . .
2.7. Non burn-specific measures . . . . . . . . . . . . .
2.8. Predicting other outcomes. . . . . . . . . . . . . . .
2.8.1. Length of stay . . . . . . . . . . . . . . . . . .
2.8.2. Quality of life . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1. Designing a scoring system. . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

* Corresponding author at: 34 Tithe Close, Gazeley, Newmarket, Suffolk, CB8 8RS, United Kingdom. Tel.: +44 1371830454;
fax: +44 07980901395.
E-mail address: nicknsheppard@hotmail.com (N.N. Sheppard).
0305-4179/$36.00 # 2011 Elsevier Ltd and ISBI. All rights reserved.
doi:10.1016/j.burns.2011.07.017

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.

1289
1289
1289
1290
1290
1290
1291
1291
1291
1292
1292
1292
1292
1292

1289

burns 37 (2011) 12881295

4.

1.

3.1.1. Mortality prediction in practice . . . . . . . . .

3.1.2. Predicting death at the bedside . . . . . . . . .
3.1.3. Predicting outcome for research purposes.
3.1.4. Internal audit . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Introduction

Survival after burn has steadily improved over the last few
decades. With overall death rates in the range of 515%,
patient mortality is still at the forefront of outcome
measures for burn care. Numerous factors combine to form
a clinical picture and dictate interventions and outcome.
Scoring systems aim to use the most predictive of these
factors to yield an expected likelihood of death for a given
patient.
This prediction has a number of potential uses. The
standardised mortality ratio (SMR) is the ratio of observed
to predicted deaths and can be used as an index of a units
overall performance. A scoring system enables a plot of
performance over time, which takes into account variations in
case-mix. A standardized tool allows for improved research
and comparison of therapeutic interventions. The SMR is used
as the endpoint for trial of an intervention. The ideal scoring
system would extend beyond research and audit. A robust
schema should improve our ability to provide a bedside
prognosis and even plan delivery of treatment, provision of
services and identify the patients in whom a palliative
approach should be taken over one of aggressive intervention
[1]. Some, however, would argue that no score should
determine medical interventions [2].
Any of the above may be possible for a system that is locally
standardised, but the final step is for a scoring tool to retain
accuracy and precision when applied to different patient
populations in differing units with inevitably differing casemixes. For this a multi-centre approach is required.
The search for understanding of the factors contributing to
mortality has been an integral part of burns research from
early in its nascence as a specialized area of practice. Studies
from Copenhagen [3] and Toronto [4] paved the way for an
analysis by Bull and Fisher [5], which yielded a prognostic
nomogram based on age and percentage area burned.
It was Baux in 1961 who first described the oft quoted ruleof-thumb that:
Mortality rate age percentage area burned
This was only published in a thesis in French, but gained
wide international acceptance [6]. A later modification was the
prognostic burn index (PBI) below which takes account of the
effects of differing burn thickness [7].
1
PBI TBSA full thickness TBSA partial thickness age
2
Much of the work over the subsequent decades centred
entirely around the two variables of age and burn surface area.
During this time burn mortality fell with each step of progress
in general intensive care and burns surgery [8]. Fluid

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

1293
1293
1293
1293
1294
1294

resuscitation, improved intensive care techniques and early

surgical excision are just a few of many advances which have
improved the prognosis for the burned patients.
Inhalational injury was subsequently found to be highly
predictive and yielded the equation by Clark [9]
ez
p
1 ez
where p = probability of mortality, z =  7  9 + 0.78(respiratory
symptom score) + 0.094(TBSA) + 0.34(age).
The need for a more refined approach to burns care
analysis has prompted many attempts to generate predictive
models and considerable debate as to the merits of this
approach. We look in detail at the work undertaken to better
elucidate the relationship between the patient, their injury
and their risk of subsequent mortality.

2.

Materials and methods

Pubmed and medline databases were searched using the

terms burn, score, outcome, prediction and mortality. Subsearches were performed for intensive care outcome, mortality, quality and score and each resultant scoring systems name
cross-referenced with the term burn. Numerous papers have
looked at prognostic factors alone, but we have focussed on
those which have generated a scoring system per se. All named
burns scoring systems are included, or those general ITU
systems for which there is evidence or opinion relating them
to burns patients. The better-known, most recent and more
widely evidenced systems are the subject of more detailed
discussion.

2.1.

Abbreviated burn severity index (ABSI) 1982

Published in 1982 the ABSI is in widespread use [10]. The

system was based on analysis of over 1300 patient records. The
system used multivariate logistic regression to determine the
power of 5 variables (gender, age, inhalation injury, %TBSA
and presence of FT burn) to predict mortality. Each variable
was then assigned a numerical value which varies according
to severity. The sum of these values is used to predict
mortality (Table 1).
Tobiasen et al. applied the score to an estimation group and
found it to be accurate at predicting risk. It was then validated
using data collected from a combination of community
hospitals and burn centres and once again found to be
accurate. This scoring system is easy to use and based upon
simple clinical findings. It gives a range for survival which may
be of benefit when discussing outcomes with family members.

1290

burns 37 (2011) 12881295

Table 1 Abbreviated burn severity index.

Variable

Patient characteristic

Sex

Female
Male
020
2140
4160
6180
81100

Age in years

Inhalation injury
Full thickness burn
TBSA %
110
1120
2130
3140
4150
5160
6170
7180
8190
91100
Threat to life
Total burn score
23
Very low
Moderate
45
Moderately severe
67
Serious
89
Severe
1011
Maximum
1213

Score
1
0
1
2
3
4
5
1
1
1
2
3
4
5
6
7
8
9
10
Survival probablity
>99
98
8090
5070
2040
<10

Its accuracy has been confirmed by validation both internally

and externally. It does however need to be charted and visible
within a burns unit as it is difficult to remember the value of all
components. Once again it acts only as a guide for outcome
and is not an absolute predictor of mortality.

2.2.

Cape Town modified burns score 1998

The main aim of the Cape Town system was to improve the
Baux score and create a triage-type scoring system to ensure
that burn injured patients are directed to the appropriate
centre (i.e. intensive care or ward) for management [11]. The
authors acknowledge that %TBSA and age have an equal effect
on outcome but recognize the importance of the presence or
absence of an inhalation injury. They attempt to refine the
Baux score by assigning a numeric value to the degree of
inhalation injury. As in the ABSI score multiple logistic
regression is utilized to determine the impact of %TBSA, age
and inhalation injury on mortality concentrating in particular
on inhalation injury. The Cape Town group noted that they
encountered 28 deaths which were not predicted by the Baux
score. Of these 24 had an inhalation injury. Using multiple
regression analysis it was determined that each grade of
inhalation injury (mild, moderate, severe, score 1, 2, 3
respectively) had 20  the effect on mortality. The modified
burn score thus created: age + %TBSA + (20  inhalation score)
was found to markedly improve the predictive value from 43%
with Baux to 84% in their patient group.
The disadvantage of the modified score was a fall in
specificity.
Once again the score is easy to remember and apply
however it is not accurate at predicting outcome. Its increased
sensitivity is at the expense of a high rate of false predictions

of mortality. It is however a useful resource for triage when

mass causalities may exceed capacity for treatment.

2.3.

Ryan 1998

This study was based on a retrospective review of 1665 patients

admitted to Massachusetts General Hospital and the Shriners
burns institute between 1990 and 1994 [12]. It was published in
1998. Age greater than 60, TBSA greater than 40% and
inhalational injury were identified as risk factors for mortality.
Gender, admission and discharge dates, type of burn, and need
for escharotomy were analysed and found not to be predictive.
Their calculations led to a simple scoring system based on
how many of the above three risk factors were present. Hence,
no risk factors gave a mortality of 0.3%, 1 factor 3%, 2 risk
factors 33% and all 3 risk factors had a mortality of 90%. This
held for patients under the age of 90.
The widely held criticism of this system is that the
mortality rates for the higher-risk burns with scores of 2 or
3 were based on groups of 111 and 22 patients, respectively. It
has also been noted that few authors have been able to report
an overall mortality rate as low as in this cohort (4%) [10,11,13]
and the scoring system has been found to under-predict
mortality in severely burned patient groups [14]. All of this
suggests that the system is based too heavily on lower risk
burns and fails on accuracy with larger, high-risk burns in
whom mortality is more likely.

2.4.

Burd 2002

This study sought to isolate burns of greater than 10% from the
broader burn population and yield a predictive model for
major burns in Hong Kong [15]. 286 major burns were seen over
a 7 year period with a median burn size of 18%. Stepwise
logistical regression was used for the analysis. Independent
risk factors for death were inhalational injury, age and TBSA.
LOS was, in addition to these, predicted by gender. The
formulas are given below.
Probability of death 1 ey1

where e 2:718;

y 3:6  1:7inhalation injury 0:001age2

0:001TBSA of burn2 ;  inhalation injury no 1; yes 0
LOS 8:7 2:1TBSA of burn  0:0018TBSA of burn2
16:7inhalation injury  9:4Sexy ;
 inhalation injuryno 0; yes 1;
y sexfemale 1; male 2
In essence, the mortality component of this study is very
similar to the much earlier Clark formula [9]. It demonstrates
the usefulness of multivariate analysis in yielding a
manageable line-of-best fit, the equation for which can
be put into practice as a unit-specific predictive measure. In
this case their formula proved 93% positively predictive
although it is not clear whether this was carried out on a
separate study population or on the population from which it
was derived.
In this study, it is of note that length of stay is not predicted
by age but by gender; they do not go on to elucidate exactly
how predictive this formula subsequently proved to be.

1291

burns 37 (2011) 12881295

Table 2 Belgian outcome in burn injury scale.

Age (years)
TBSA (%)
Inhalation injury

<50
<20
No

5064
2039

6579
4059

>80
6079
Yes

>80

Total score

Predicted mortality (%)

2.5.

10

0.1

1.5

10

20

30

50

75

85

95

99

Belgian outcome in burn injury (BOBI) 2009

The BOBI scoring system is the most recent addition to the

severity scores [16]. The group attempted to develop a model
that accurately predicted the probability of mortality based
upon the clinical parameters set out by Ryan et al. in essence
this scoring system attempts to increase the predictive value
of Ryans scoring system by subdividing patients according to
age and TBSA (Table 2).
The model was derived using data from prospectively
recorded databases for more than 5000 patients between 1999
and 2003. As expected the majority of patients fell into the low
risk mortality groups with 270 patients scoring 5 and above.
This might lead one to believe that as with Ryan the scoring
system is flawed however this does not appear to be the case.
Prior to initial publication the scoring system was validated in
a further 981 patients and was found to accurate predicting 40
of 42 deaths. To add weight to its value the model has been
tested with a different population for external validation [17].
The system once again demonstrated a high predictive value.
As with all scoring systems there are drawbacks and
limitations. A major criticism levelled at all such systems is
that the impact of complications and co-morbidities is often
ignored. The Belgian group acknowledged this but rightly
point out that the impact of co-morbidities is too small to sway
the prediction score. They also point out that many of the
complications which adversely affect survival occur in the
most severe injuries who would score highly on mortality
prediction. It is also therefore assumed that this does not
further impact outcome.
The Belgian scoring system has clear advantages over the
Ryan and ABSI. It refines the Ryan score hence increasing
predictive value [14,17] and is more simple and easier to
remember than ABSI for bedside use. Its place however in
prospective prediction of poor outcome is yet to be determined.

2.6.

Laboratory-based prognostic scoring

Taking the process a step further from the bedside are the
studies which examine biochemical and inflammatory markers and their correlation with burn mortality. The unique
burn inflammatory response may combine with variables
such as age and inhalation injury to yield a specific change in
biochemical constitution which can readily be assessed by
simple laboratory assays.
The prognostic inflammatory and nutritional index (PINI) is
one such system [18]. Plasma levels of albumin, prealbumin,
orosomucoid and C-reactive protein (CRP) were found to
correlate with burn severity. Subsequent work showed the

PINI to correlate with mortality [19] in 60 patients with burns of

mean 44.7% TBSA.
Early changes in renal function have been shown to
independently predict mortality [20]. This readily measurable
parameter may reflect much of the systemic effect of the
inflammatory response and hypoperfusion which goes on to
bring about death after a burn. In a similar vein, base deficit and
lactate changes over the first 24 h correlate with mortality [21].
Tumour necrosis factor (TNF) has been found to correlate
with mortality and to be independent of burn area and age. It
was more common in, but not limited to those with sepsis [22].
The study results include unexplained anomalies and lacked
the statistical power to definitively confirm predictive merit in
the assay, but the correlation has been borne out since. It was
shown that admission TNF level did not correlate with burn
surface area or mortality rates, but the maximum level
attained during the admission did [23].
Other circulating inflammatory factors released after burn
have been the subject of intense research. IL-6, IL-8 and IL-10
have all been demonstrated to increase following burns, but
there are conflicting studies as to their correlation with
subsequent mortality [24,25].

2.7.

Non burn-specific measures

The APACHE or acute physiology and chronic health evaluation score was developed in the early 1980s by Knaus et al.
[26]. His initial 34 variables were narrowed down to 12 in 1985
with the development of APACHE II [27]. This system assigns
points for aberrant variables occurring within the first 24 h of
ITU admission. As such it goes further than any of the existing
burn-specific systems in using the patients response to initial
goal-directed therapy as part of their prognostic determination. It was updated to APACHE III in 1991 and is a highly
regarded tool used throughout the literature [28].
Apache II score on admission was found to be associated
with subsequent mortality in burns [29]. Apache III has been
formally tested on a burn population and found to correlate
with mortality and PBI [30]. The correlation coefficient for
APACHE III score plotted against PBI was .74. However, with a
sample size of just 74 with completed APACHE scores of which
14 died the study does not allow us to draw conclusions as to
whether this proved to be the better predictor of mortality
than the burn-specific models.
These findings yielded the appropriately titled FLAMES
score (fatality by longevity, APACHE II score, measured extent
of burn, and gender). In 1439 patients attending a single unit,
this achieved areas under the receiver operating characteristic
(ROC) curve of 0.97 in the population from which it was

1292

burns 37 (2011) 12881295

developed and 0.93 in a validation population [31]. Thus, it

performed better than APACHE or burn indexes alone.
APACHE III score and burn severity (in terms of full
thickness area) have been shown to provide more reliable
prediction of mortality in burn patients when combined than
can either score alone [32]. The study lacks the numbers
required to fully validate a scoring system, but it demonstrates
integration of burn-specific factors with the APACHE score and
paves the way for a larger investigation of this combination.
APACHE IV is the most recent incarnation [33].
The Sepsis-Related Organ Failure Assessment tool specifically looks at organ-failure in the ITU population [34]. SOFA
scores were investigated in 439 burns intensive care patients.
The SOFA score on days 04 and the rate-of-change of this
score are predictive of mortality and this is independent of
age, sex, TBSA or inhalational injury [35]. Organ failure has
been investigated in a burns population. The Thermal Injury
Organ Failure Score was developed in a study looking at organ
failure and performed better than APACHE II in predicting
mortality [36].
SAPS (simplified acute physiology score) was developed as
an offshoot of APACHE I and was updated in 1993 to SAPS II
[37]. Both SOFA and SAPS scores were used as part of a study
looking at mortality and quality-of-life (QoL) correlates. Both
scoring systems were found to correlate with mortality [38],
but with numbers of just 50 and with only 19 QoL 1-year
follow-ups, again, lasting conclusions with respect to these
scoring systems cannot yet be made.
The paediatric intensive care population was addressed
specifically by the PRISM (paediatric risk of mortality) score.
This was first developed in 1988 [39] and has subsequently
been revised to PRISM III-APS (Acute Physiology Score). It was
originally a condensation of the physiologic stability index to
14 routine physiologic variables based on 116 deaths in 1415
patients across 4 units. PRISM scores were combined with
operative status and age to provide a good predictive
performance. The revision to PRISM III was performed on a
larger data set and takes into account readings at 12 and 24 h
following admission [40]. It performs better than its predecessor for a general ITU population. The system has not been
looked at for a burn-specific paediatric population.
While the use of non-burn specific measures may prove
valid in a general ITU population, none take into account the
profound physiological effects of the burn itself. In isolation,
they cannot compare to measures that do take into account
the burn. However, the rigor with which they have been
formulated and the number of variables that have proved
worthy of inclusion through thorough statistical analysis, may
provide a standard to which we should be striving in the
development of a robust burns scoring system.

2.8.

Predicting other outcomes

It has long been argued that mortality should no longer be the

only yardstick against which burn care is measured. In the
short term mortality may be valid. However, consideration
should be given to medium-term measures of persistent
hypermetabolic response and length of stay. In the long term,
quality of life and exercise tolerance may be relevant
endpoints for consideration [41].

2.8.1.

Length of stay

Length of stay has been looked at in some of the studies in

which mortality was the primary burn outcome [12,15,16].
There is a rule-of-thumb for which a reference could not be
found that length of stay in days is roughly equal to value for
TBSA. Ryan found that patients could be subdivided by burn
size to give predictions of a probable range in which the length
of stay would fall. These ranges are broad, but loosely
correspond to the crude rule above. If this relationship is
expressed as the ratio of length of stay in days/TBSA, we can
aim for a value of 1 [42].
The Hong Kong study discussed earlier [12] yielded an
equation for prediction of length of stay and crucially found
that sex was more predictive than age. The equation has not
been the subject of subsequent prospective analysis. Further
study of these parameters with larger data sets and stratification rather than a simple line of best fit analysis may give a
clearer picture of this relationship.

2.8.2.

Quality of life

While the predictive factors for quality of life after burns have
been examined in a number of studies, no predictive scoring
system has yet been developed.
The burn specific health scale (BSHS) is the current standard
measure for quality of life outcomes after burns [43,44]. It is a
laborious questionnaire which considers 114 items divided
into 8 sub-scales. Many variations, including abbreviated
versions have been developed, each with their merits and
were reviewed in detail by Yoder et al. [45]. This paper also
summarizes the data from numerous studies which identified
factors affecting eventual quality of life scores on this scale.
BSHS scores have been shown to correlate with extent of burn
and burn location (particularly hand and face) along with
length of stay and eventual mobility [46,47]. Furthermore, it
was shown that psychosocial support, both existing pre burn
and that offered by friends, family and professionals post burn
is a key predictive factor for subsequent rehabilitation [48].
The Short-Form 36 survey has been used in one study which
found two factors in addition to the established variables for
mortality prediction. In the physical section of the survey, hand
function at follow up correlated strongly with overall QOL score.
For the mental section, pre-existing social support was found to
be predictive of outcome [43]. Psychological well-being has been
found to be independent of burn severity [49]. In the same study
of 34 burn patients, subsequent psychopathology was found to
correlate with pre-injury psychological state, coping mechanisms and a threat of death at the time of the injury.
An all-encompassing approach to prediction modelling for
the burn patient should consider these factors in its analysis,
but as yet none has gone this far and until quality of life
endpoints can be reliably assessed, scoring systems for their
prediction will lag some way behind.

3.

Discussion

3.1.

Designing a scoring system

Certain parameters clearly have a place in any prognostic

index. The inclusion of age and burn area are widely agreed,

burns 37 (2011) 12881295

but there remains disagreement in the literature as to the

relative merits of total burn area versus the area of full
thickness burn. In an Australian cohort of 228 patients, full
thickness area was found to correlate with mortality where
partial thickness area did not [32]. The Smith score adds
considerable weight to the presence of absence of inhalational
injury [50] and the relevance of this parameter in prognostic
scoring is now widely accepted. However, it has been
suggested that the need for invasive ventilation may be more
predictive than inhalational injury alone [51]. There is also the
question of how inhalational injury is defined as it can be
diagnosed on clinical examination or by bronchoscopy.
Parameters which remain contentious are those of gender,
mechanism of injury and the physiological variables which
come in when burn factors are combined with any of the
established ITU prognostic models.
The principal disagreement between many of the proposed
scoring systems is the relative weighting of the basic
components. The continued findings of each unit that a
new model best predicts performance in their patient group
may simply reflect the heterogeneity of burns patient groups
and practice and be evidence for the futility of attempts to
develop an all-encompassing prognostic tool.

against the current accepted standard. Moreover, however, it

demonstrates a lack of need for it at the bedside. If the rigmarole
of collating scores offers no meaningful benefit over the opinion
of an experienced team, they have no clinical merit.
The exception to this may been where experience is limited
and a useful mortality prediction provides something of an
alternative. In Western medicine in the modern era with
centralized burn services and considerable specialist expertise, this is rarely a requirement. However, in parts of the
world where access to specialist burns services are more
limited, this kind of prognostic evaluation may be invaluable.
For this to work, however, the scoring system in use must been
one validated on an equivalent population with a similar lack
of access to facilities. Such a situation may be best served by
the simplest Baux calculation, but the work is yet to be done to
demonstrate this.
Medicine in the developed world is under increasing
financial pressure and here there are further warnings for
the implementation of mortality prediction. In uncertain
economic times there is a concern that scoring patients may
provide opportunities for bureaucrats to limit therapeutic
options [2].

3.1.3.
3.1.1.

Mortality prediction in practice

The margin for error of any scoring system depends entirely

on its intended use. Of the potential uses covered earlier some
are beyond the scope of any of the existing scoring measures.
Developing new measures or selecting from existing scoring
systems must be done in the context of the intended
application. All of the published scoring systems show validity
for the patient groups on which they were developed, but care
must be taken when planning to broaden their scope.

3.1.2.

Predicting death at the bedside

It has been described as a nightmare scenario that bedside

decisions may be made on the basis of a score [2]. Ryan et al.
found that in 37 similar patients with an intermediate risk for
death according to his scoring system, 11 had do no attempt
to resuscitate (DNAR) orders and 26 did not. Only 5 of the 26
patients who were expected to die went on to do so. There
were no differences between the resuscitated or DNAR groups
in terms of Ryan score or any of the other factors examined in
the paper [12]. The scoring system did not provide useful
information during the decision-making process and the
eventual outcomes show that had it have been used for
decision-making, more of the 26 for whom DNAR orders were
not issued may have died unnecessarily. Such a retrospective
analysis, albeit at risk of selection and observer bias would be a
worthy test of any score being put forward for clinical
application. Ryans findings are a salutary warning against
the clinical application of such measures.
The alternative to an objective scoring system is the
subjective view of the clinician, in combination with those of
the whole clinical team. Sensitivity and specificity for APACHE II
versus clinical opinion were compared and subjective appraisal
was found to perform better in the mixed intensive care
population studied [52]. Clinical sense has subsequently been
shown not to differ in accuracy from the APACHE III score [53].
Such similarity can be argued to validate the scoring system

1293

Predicting outcome for research purposes

There is a clear requirement to risk-stratify patient groups in

order to properly assess outcomes following the introduction
of novel therapies. With so many scoring systems to choose
from, how does the researcher make a selection to best
analyse their units outcomes or the efficacy of a new
intervention? One approach may be to retrospectively
compare the existing systems within their unit and ascertain
which proves to be the most applicable to use prospectively in
their research population.
The most rigorous approach may be to use more than one
or as many as possible. Since most of the burns systems are
based on similar information, scores could be included from
numerous models with little extra effort. This would have the
double benefit of adding rigor to the findings of an individual
study, but will also provide much-needed information for a
later reviewer to conclude which is the most reliable of the
existing prognostic prediction measures.

3.1.4.

Internal audit

Sound clinical practice and the ongoing assessment of its

quality includes the need for audit. Internal audit of mortality
is the most basic means of assessing trends in a units
performance. Doseresponse data such as burn area to
mortality require probit analysis to determine meaningful
statistics. This renders doseresponse data in this case e.g.
TBSA and mortality into a linear relationship from which an
LD50 can be calculated. That is, in this example, the percentage
burn at which that unit has a mortality rate of 50%. It can also
be used independently for age and other continuous variables.
It is useful as an audit tool for its simplicity, but does not
attempt to provide multivariate information.
A more complex scoring system provides a basis for local
audit of mortality with a correction for case-mix. Internal
audit is the most robust circumstance in which scoring
systems are useful as many of the criticisms applied thus far
apply when their remit is thus reduced.

1294

burns 37 (2011) 12881295

In the evolving climate of transparency there may be merit

in the compulsory publication of a units figures for agreed
outcome parameters including mortality. This could be
corrected for case-mix using one of the above systems. While
comparisons between units may be unfair or problematic,
trends with time within a unit will be a good guide to its
ongoing performance and may allow timely intervention
should standards show evidence of falling.
Cumulative SUMmation (CUSUM) charts have been described as a method of continuous monitoring and have been
extensively used in industry for quality control purposes.
CUSUM charts for prospective surveillance of mortality have
been proposed and utilized in many clinical fields including
Cardiac Surgery, Anaesthesia and General Practice [54,55] to
enable early detection of substandard outcomes. We are in the
process of assessing this methodology to continuously
monitor our own burn service to see if this is a valid tool in
the burn setting.
To perform a CUSUM analysis the data for each patient is
continuously plotted as a cumulative sum of the observed (Xo)
minus the expected (Xe) outcome or Xn = Xn(1) + (Xo  Xe).
Expected outcomes can be generated by using the scoring
systems described. CUSUM charts can be generated for burn
service providers to continuously monitor outcome data.
Deviation below the expected outcome could be used to
trigger review of clinical practice within a burn service. This
system could be used within one service or across many
burn care providers as a method of quality control and
improvement.

4.

Conclusion

Numerous burns prognostic scoring systems exist of which no

one can claim to be the most accurate across the entire burn
population. An ideal prognostic system should include
information about quality of life.
There is no evidence to support their use at the bedside for
decision-making. The subjective view of the clinical team is as
valid and is more likely to take into account the gamut of
factors which underpin the decision-making required in the
severely injured patient.
As a research tool they are, however, invaluable. It is
necessary to be able to compare outcomes to an expected or
control value and a single value from an established scoring
system would facilitate this greatly. Intuition may suggest that
a scoring system which comprises more components would
prove more accurate. However, for mortality in burns, TBSA,
age and inhalation injury appear to outweigh other factors
significantly enough to be the only components used to give a
prediction. It would be very interesting to see further efforts to
combine APACHE or SOFA-based system with these three burn
parameters to provide a complex, but robust burn scoring
system for research purposes.
As it stands the achievements so far in this area have
provided a choice of schemes to control for case-mix in a burn
population, but their value in prediction for a given patient is
limited. As such, they are best used for internal audit and
research though not yet for the planning of services and less
still bedside decision-making.

Conflict of interest statement

None declared.

references

[1] Knaus WA, Wagner DP, Lynn J. Short-term mortality

predictions for critically ill hospitalized adults: science and
ethics. Science 1991;254(5030):38994.
[2] Andel, Kamolz. Scoring in burned patients. Our opinion.
Burns 2003;29(4):2978.
[3] Lutken. Mortality from burns. Ugesk Laeger 1937;99:409.
[4] Farmer. Experience with burns at the hospital for sick
children. Am J Surg 1943;59:195.
[5] Bull, Fisher. A study of mortality in a burns unit: a revised
estimate. Ann Surg 1954;139(3):26974.
[6] Baux. Contribution a letude du traitement local des
brulures thermiques etendues. These, Paris; 1961.
[7] Nakae, Wada. Characteristics of burn patients transported
by ambulance to treatment facilities in Akita Prefecture.
Jpn Burns 2002;28(1):739.
[8] Barnes BA, Constable JD, Burke AF. Mortality of burns at the
Massachusetts General Hospital 19551969: transactions of
the 3rd International Congress on Research in Burns. Res
Burns 1971;430.
[9] Clark CJ, Reid WH, Gilmour WH, Campbell D. Mortality
probability in victims of fire trauma: revised equation to
include inhalation injury. Br Med J (Clin Res Ed)
1986;292(6531):13035.
[10] Tobiasen J, Hiebert JM, Edlich RF. The abbreviated burn
severity index. Ann Emerg Med 1982;11(5):2602.
[11] Godwin, Wood. Major burns in Cape Town: a modified
burns score for patient triage. Burns 1998;24(1):5863.
[12] Ryan CM, Schoenfeld DA, Thorpe WP, Sheridan RL, Cassem
EH, Tompkins RG. Objective estimates of the probability of
death from burn injuries. N Engl J Med 1998;338(6):3626.
[13] OKeefe GE, Hunt JL, Purdue GF. An evaluation of risk
factors for mortality after burn trauma and the
identification of gender-dependent differences in
outcomes. J Am Coll Surg 2001;192(2):15360.
[14] Sheppard NN, Hemington-Gorse S, Ghanem A, Philp B,
Dziewulski P, Shelley OP. The Belgian severity prediction
model compared to other scoring systems in a burn
intensive care population. Burns 2010;36:13201.
[15] Ho WS, Ying SY, Burd A. Outcome analysis of 286 severely
burned patients: retrospective study. Hong Kong Med J
2002;8(4):2359.
[16] Belgian Outcome in Burn Injury Study Group. Development
and validation of a model for prediction of mortality in
patients with acute burn injury. Br J Surg 2009;96(1):
1117.
[17] Brusselaers N, Juhasz I, Erdei I, Monstrey S, Blot S.
Evaluation of mortality following severe burns injury in
Hungary: external validation of a prediction model
developed on Belgian burn data. Burns 2009;35(7):100914.
[18] Kudlackova M, Andel M, Hajkova H, Novakova J. Acute
phase proteins and prognostic inflammatory and
nutritional index (PINI) in moderately burned children aged
up to 3 years. Burns 1990;16(1):536.
[19] Gottschlich MM, Baumer T, Jenkins M, Khoury J, Warden
GD. The prognostic value of nutritional and inflammatory
indices in patients with burns. J Burn Care Rehabil
1992;13(1):10513.
[20] Steinvall I, Bak Z, Sjoberg F. Acute kidney injury is
common, parallels organ dysfunction or failure, and carries

burns 37 (2011) 12881295

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

appreciable mortality in patients with major burns: a

prospective exploratory cohort study. Crit Care
2008;12(5):R124.
Andel D, Kamolz L-P, Roka J, Schramm W, Zimpfer M, Frey
M, et al. Base deficit and lactate: early predictors of
morbidity and mortality in patients with burns. Burns
2007;33(8):9738.
Marano MA, Fong Y, Moldawer LL, Wei H, Calvano SE,
Tracey KJ, et al. Serum cachectin/tumor necrosis factor in
critically ill patients with burns correlates with infection
and mortality. Surg Gynecol Obstet 1990;170(1):328.
Endo S, Inada K, Yamada Y, Kasai T, Takakuwa T, Nakae H,
et al. Plasma tumour necrosis factor-alpha (TNF-alpha)
levels in patients with burns. Burns 1993;19(2):1247.
Yeh FL, Shen HD, Fang RH. Deficient transforming growth
factor beta and interleukin-10 responses contribute to the
septic death of burn patients. Burns 2002;28:6317.
Csontos C, Foldi V, Palinkas L, Bogar L, Roth E, Weber G,
et al. Time course of pro- and anti-inflammatory cytokine
levels in patients with burns prognostic value of
interleukin-10. Burns 2010;36(4):48394.
Knaus WA, Zimmerman JE, Wagner DP, Draper EA,
Lawrence DE. APACHE acute physiology and chronic
health evaluation: a physiologically based classification
system. Crit Care Med 1981;9(8):5917.
Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE
II: a severity of disease classification system. Crit Care Med
1985;13(10):81829.
Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner
M, Bastos PG, et al. The APACHE III prognostic system. Risk
prediction of hospital mortality for critically ill hospitalized
adults. Chest 1991;100(6):161936.
Wong Ti H, Tan BH, Ling ML, Song C. Multi-resistant
Acinetobacter baumannii on a burns unit clinical risk factors
and prognosis. Burns 2002;28(4):34957.
Tanaka Y, Shimizu M, Hirabayashi H. Acute physiology,
age, and chronic health evaluation (APACHE) III score is an
alternative efficient predictor of mortality in burn patients.
Burns 2007;33(3):31620.
Gomez M, Wong DT, Stewart TE, Redelmeier DA, Fish JS.
The FLAMES score accurately predicts mortality risk in
burn patients. J Trauma 2008;65(3):63645.
Moore EC, Pilcher DV, Bailey MJ, Cleland H, McNamee J. A
simple tool for mortality prediction in burns patients:
APACHE III score and FTSA. Burns 2010;36(7):108691.
Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute
Physiology and Chronic Health Evaluation (APACHE) IV:
hospital mortality assessment for todays critically ill
patients. Crit Care Med 2006;34(5):1297310.
Vincent JL, Moreno R, Takala J, Willatts S, de Mendonca A,
Bruining H, et al. The SOFA (Sepsis-related Organ Failure
Assessment) score to describe organ dysfunction/failure.
On behalf of the Working Group on Sepsis-Related
Problems of the European Society of Intensive Care
Medicine. Intensive Care Med 1996;22(7):70710.
Lorente JA, Vallejo A, Galeiras R, Tomicic V, Zamora J, Cerda
E. Organ dysfunction as estimated by the sequential organ
failure assessment score is related to outcome in critically
ill burn patients. Shock 2009;31(2):12531.
Saffle JR. Multiple organ failure in patients with thermal
injury. Crit Care Med 1993;21(11):167383.

1295

[37] Le Gall JR, Lemeshow S, Saulnier F. A new Simplified

Acute Physiology Score (SAPS II) based on a European/
North American multicenter study. JAMA 1993;270(24):
295763.
[38] Pavoni V, Gianesello L, Paparella L, Buoninsegni LT, Barboni
E. Outcome predictors and quality of life of severe burn
patients admitted to intensive care unit. Scand J Trauma
Resusc Emerg Med 2010;18:24.
[39] Pollack MM, Ruttimann UE, Getson PR. Pediatric risk of
mortality (PRISM) score. Crit Care Med 1988;16(11):11106.
[40] Pollack MM, Patel KM, Ruttimann UE. PRISM III: an updated
Pediatric Risk of Mortality score. Crit Care Med
1996;24(5):74352.
[41] Pereira C, Murphy K, Herndon D. Outcome measures in
burn care. Is mortality dead? Burns 2004;30(8):76171.
[42] Mossey, Shapiro. Self-rated health: a prediction of
mortality among the elderly. Am J Public Health
1982;72:8008.
[43] Anzarut A, Chen M, Shankowsky H, Tredget EE. Quality-oflife and outcome predictors following massive burn injury.
Plast Reconstr Surg 2005;116(3):7917.
[44] Munster. Measurements of quality of life: then and now.
Burns 1999;25(1):258.
[45] Yoder LH, Nayback AM, Gaylord K. The evolution and utility
of the burn specific health scale: a systematic review. Burns
2010;36:114356.
[46] Baker RA, Jones S, Sanders C, Sadinski C, Martin-Duffy K,
Berchin H, et al. Degree of burn, location of burn, and
length of hospital stay as predictors of psychosocial status
and physical functioning. J Burn Care Rehabil
1996;17(4):32733.
[47] Kimmo T, Jyrki V, Sirpa AS. Health status after recovery
from burn injury. Burns 1998;24(4):2938.
[48] Pruzinsky T, Rice L, Himel H, Morgan R, Edlich R.
Psychometric assessment of psychologic factors
influencing adult burn rehabilitation. J Burn Care Rehabil
1992;13:7988.
[49] Willebrand M, Andersson G, Ekselius L. Prediction of
psychological health after an accidental burn. J Trauma
2004;57(2):36774.
[50] Smith DL, Cairns BA, Ramadan F, Dalston JS, Fakhry SM,
Rutledge R, et al. Effect of inhalation injury, burn size, and
age on mortality: a study of 1447 consecutive burn patients.
J Trauma 1994;37(4):6559.
[51] Galeiras R, Lorente JA, Pertega S, Vallejo A, Tomicic V, de La
Cal MA, et al. A model for predicting mortality among
critically ill burn victims. Burns 2009;35(2):2019.
[52] Marks RJ, Simons RS, Blizzard RA, Browne DR. Predicting
outcome in intensive therapy units a comparison of
Apache II with subjective assessments. Intensive Care Med
1991;17(3):15963.
[53] Atiyat B, Kloub A, Abu-Ali H, Massad I. Clinical Sense in the
Prediction of Surgical/Trauma Intensive Care Mortality. Eur
J Sci Res 2009;30(2):26571.
[54] Noyez. Control charts Cusum techniques and funnel plots.
A review of methods for monitoring performance in
healthcare. Interact Cardiovasc Thorac Surg 2009;9(3):p494.
[55] Guthrie B, Love T, Kaye R, MacLeod M, Chalmers J. Routine
mortality monitoring for detecting mass murder in UK
general practice: test of effectiveness using modelling. Br J
Gen Pract 2008;58(550):3117.