You are on page 1of 4

Intensive Care Med (1996) 22:707-710

9 Springer-Verlag 1996

J.-L. Vincent The SOFA (Sepsis.related Organ Failure


R. Moreno
J. Takala Assessment) score to describe organ
S. Willatts
A. De Mendon~a dysfunction/failure
H. Bruining
C.K. Reinhart On behalf of the Working Group on Sepsis.Related
P.M. Suter
L.G. Thijs
Problems of the European Society of Intensive Care
Medicine (see contributors to the project in the appendix)

during a complex clinical course, what is sometimes re-


Received: 18 January 1996
Accepted: 19 April 1996 ferred to as a "multiple-hit" scenario. (c) Time evaluation
allows a greater understanding of the disease process as
a natural process or under the influence of therapeutic in-
J.-L. Vincent (~) 9 R. Moreno terventions. The collection of data on a daily basis seems
J. Takala 9 S. Willatts 9 A. De Mendonga adequate.
H. Bruining 9 C.K. Reinhart
RM. Suter 9 L.G. Thijs 3. The evaluation of organ dysfunction/failure should be
Erasme University Hospital, based on a limited number of simple but objective vari-
808 route de Lennik, ables that are easily and routinely measured in every insti-
B-1070 Brussels, Belgium tution. The collection of this information should not im-
pose any intervention beyond what is routinely performed
Multiple organ failure (MOF) is a major cause of mor- in every ICU. The variables used should as much as possi-
bidity and mortality in the critically ill patient. Emerging ble be independent of therapy, since therapeutic manage-
in the 1970s, the concept of M O F was linked to modern ment may vary from one institution to another and even
developments in intensive care medicine [1]. Although an from one patient to another (Table 1).
uncontrolled infection can lead to M O F [2], such a phe- Until recently, none of the existing systems describing
nomenon is not always found. A number of mediators organ failure met these criteria, since they were based on
and the persistence of tissue hypoxia have been incrimi- categorial definitions or described organ failure as pre-
nated in the development of M O F [3]. The gut has been sent or absent [5-7].
cited as a possible " m o t o r " of M O F [4]. Nevertheless, our The E S I C M organized a consensus meeting in Paris in
knowledge regarding the pathophysiology of M O F re- October 1994 to create a so-called sepsis-related organ
mains limited. Furthermore, the development of new failure assessment (SOFA) score, to describe quantitative-
therapeutic interventions aiming at a reduction of the in- ly and as objectively as possible the degree of organ dys-
cidence and severity of organ failure calls for a better def- function/failure over time in groups of patients or even in
inition of the severity of organ dysfunction/failure to individual patients (Fig. 1). There are two major applica-
quantify the severity of illness. Accordingly, it is impor- tions of such a SOFA score:
tant to set some simple but objective criteria to define the 1. To improve our Understanding of the natural history of
degree of organ dysfunction/failure. organ dysfunction/failure and the interrelation between
The evolution of our knowledge of organ dysfunc- the failure of the various organs.
tion/failure led us to establish several principles:
1. Organ dysfunction/failure is a process rather than an
event. Hence, it should be seen as a continuum and Table 1 Ideal variables for describing organ dysfunction/failure
should not be described simply as "present" or "absent~'
- Objective
Hence, the assessment should be based on a scale. - Simple, easily available, but reliable
2. The time factor is fundamental for several reasons: - Obtained routinely and regularly in every institution
(a) Development and similarly resolution of organ failure - Specific for the function of the organ considered
may take some time. Patients dying early may not have - Continuous variable
time to develop organ dysfunction/failure. (b) The time - Independent of the type of patients
- Independent of the therapeutic interventions
course of organ dysfunction/failure can be multimodal
708

TOTAL 9 10 10 12 14 16 Table 2 Differences between commonly used scoring systems and


the SOFA score
4. . . . ..... ~.. _...)~ respiratory
*
Scoring systems SOFA score
. ,,.' ! ; coagulation Evaluate risk of mortality
O 3 t- ...... ~ ,*'"7""~'"'": "~* ..... ~-..... Evaluate morbidity
(.9
oo : 9' ,~'"o~ ! ;a liver Aim = prediction Aim = description
< 9 / ~,_/ - ; ...... -fit= . . . . . . Often complex Simple, easily calculated
LL
O 2 I E~ i .' 7 " ' " " " L. t - - '~ cardiovascular__,._renal
Does not individualize the Does individualize the degree
GO degree of dysfunction/failure of dysfunction/failure of each
of each organ usually ob- organ obtained daily
1. i ,* . . . . -.-,.-.
s
tained early after admission

0 i ! i ! ! !
1 2 3 4 5 6
TIME, d a y s
indexes, b u t c o m p l e m e n t s t h e m (Table 2). Severity indices
have been designed p r i m a r i l y to evaluate a risk o f d e a t h
Fig. 1 Time course of the SOFA score in a 61-year-old patient who
presented with severe sepsis due to extensive bronchopneumonia. f r o m an initial e v a l u a t i o n [8], even t h o u g h there has been
Improvement of the respiratory failure was associated with worsen- a recent t e n d e n c y to evaluate severity indexes r e p e a t e d l y
ing of the coagulation, cardiovascular, hepatic and eventually renal to evaluate the t i m e course o f the disease [9]. M o s t i m p o r -
systems before the patient died. tantly, the existing severity indices do n o t allow evaluation
o f the i n d i v i d u a l f u n c t i o n o f each organ separately.
T h e p a r t i c i p a n t s decided: (1) to limit the n u m b e r o f
2. To assess the effects o f new therapies on the course o f organs s t u d i e d to 6. A s an example, a t t e m p t i n g to include
o r g a n d y s f u n c t i o n / f a i l u r e . This c o u l d be used to charac- d y s f u n c t i o n / f a i l u r e o f the gut was felt to be very i m p o r -
terize p a t i e n t s at e n t r y (and even serve within the e n t r y tant, b u t also t o o complex a n d was therefore a b a n d o n e d .
criteria) or to evaluate the effects o f t r e a t m e n t . (2) To use a score from 0 ( n o r m a l ) to 4 (most a b n o r m a l )
It is i m p o r t a n t to realize t h a t the S O F A score is design- for each organ. (3) To record the worst values on each day.
ed n o t to p r e d i c t o u t c o m e b u t to describe a sequence o f The S O F A score is presented in Table 3.
c o m p l i c a t i o n s in the critically ill. A l t h o u g h any assess- Since the m o r t a l i t y rate is directly related to the degree
m e n t o f m o r b i d i t y m u s t be related to m o r t a l i t y to s o m e o f o r g a n d y s f u n c t i o n , it is evident t h a t it m u s t also be re-
degree, the S O F A is n o t designed j u s t to describe o r g a n lated to the S O F A score for each o r g a n system. Neverthe-
d y s f u n c t i o n / f a i l u r e a c c o r d i n g to mortality. Hence, the less, the relation between the score a n d the m o r t a l i t y rate
S O F A score does n o t c o m p e t e with the existing severity o f critically ill patients needs to be d o c u m e n t e d . Such an

Table3 The SOFA score

SOFA score 1 2 3 4

Respiration
PaO 2/FiO2, mmHg < 400 < 300 < 200 < 100
- - with respiratory support - -
Coagulation
Platelets x 103/mm 3 < 150 < 100 <50 <20
Liver
Bilirubin, mg/dl 1 . 2 - 1.9 2 . 0 - 5.9 6.0-11.9 >12.0
(lxmol/1) (20 - 32) (33 - 101) (102 - 204) ( < 204)
Cardiovascular
Hypotension MAP < 70 mmHg Dopamine _<5 Dopamine > 5 Dopamine > 15
or dobutamine (any dose) a or epinephrine _<0.1 or epinephrine > 0.1
or norepinephrine _<0.1 or norepinephrine > 0.i
Central nervous system
Glasgow Coma Score 13 - 14 1 0 - 12 6-9 <6
Renal
Creatinine, mg/dl 1 . 2 - 1.9 2.0 - 3.4 3.5-4.9 >5.0
(gmol/1) or urine (110 - 170) (171 - 299) (300 - 440) ( > 440)
output or < 500 ml/day or <200 ml/day

Adrenergic agents administered for at least 1 h (doses given are in gg/kg-min)


709

Fig. 2 Mortality rate ver- '~


sus SOFA score in 1643
septic patients included in
the European/North
American Study of Severi-
ty Systems.
,oI IEIZI i Esi ii oi v[22EiiiiiilEiiill ..... :; -i.o ,o,ov,,oo R.-1 ...........................
,o IIZIIIIIIIIIIIIZI2 .............................. ZI
,o .................
.....
iiiii
~lO~ll.i ~ i i i ~ i i i i ~ ....
iiii iiii ii'
0 1 2,3,4
0 1 2 3 4 N,= 688 245 710
N= 200 146 606 441 250
SO
+tCQAQUON I . . . . . . . . . . . . . . .

20 . . . . . . . . . . . . . . . . . . . . . . .

10" . . . . . . . . . . . . . . . . . . . . . . .
I
0,1,2 3,4. 0 1 2 3 4
N= 1463 180 N= 783 278 241 196 145

60 .............................................................. ,.,

0
0 1 2 3 4 0 1 2 3 4
N= 443 730 364 63 43 N=, 677 480 210 118 188

SOFA score (0-4)

analysis may also result in revision of the limits of the pa- At least two similar scores have been proposed recent-
rameters used to score each organ. The relation between ly. A "Multiple Organ Dysfunction Score" was developed
the SOFA score on ICU admission and the mortality rate by J. Marshall et al. [10] and a so-called "Brussels Score"
was studied in 1643 patients with sepsis by the Europe- [l 1] was developed by G. Bernard et al. at the time of the
a n / N o r t h American Study of Severity System (ENAS). round table conference on clinical trials in sepsis [12]. A
Such a retrospective analysis has several problems. First, major difference between the three scores lies in the defi-
the ENAS data base was not created to study sepsis and nition of cardiovascular dysfunction/failure. In the "Mul-
septic shock specifically, so that the identification of sep- tiple Organ Dysfunction Score~' it is based on the complex
sis was accomplished retrospectively. Second, it was not calculation of the pressure adjusted heart rate, defined as
always possible to separate the patients in the ENAS data the product of heart rate times the right atrial (central ve-
base for all value limits used in the SOFA. This was true nous) pressure divided by the mean arterial pressure. Such
for the cardiovascular status (only three groups) and for a score, calculated a number of times over any 24 h peri-
the coagulation system (only two groups). Finally, patient od, can only be computed, so that it removes the simplici-
prognosis was only related to the SOFA on ICU admis- ty of the score. In the "Brussels Score;' it is based on
sion. Nevertheless two aspects of the data are encourag- hypotension and acidemia, but acidemia can be caused by
ing. First, they generally show an increasing mortality rate factors other than circulatory failure, including renal fail-
with a greater SOFA score for each organ. Second, they ure or (permissive) hypercapnia. Thus, even if it is signifi-
show a good distribution of patient numbers among the cantly related to mortality, it does not reflect the degree
different scores. of cardiovascular dysfunction. In the SOFA score, cardio-
In addition, a prospective collection of data was also vascular dysfunction/failure is based on the requirements
performed on all patients admitted to the ICU through- for adrenergic support. Even though it is preferable to
out the month of May 1995, except for those staying for avoid treatment-related criteria, the participants found no
less than 48 h for elective surgery (routine postoperative better way to describe cardiovascular dysfunction/failure.
surveillance). Although the SOFA score is primarily de- Although the type of adrenergic support may differ from
signed for use in the septic patient, it was felt that the se- one institution to another, the categories were broad
ries should not be limited to those patients. However, the enough to avoid a major impact of local protocols on this
presence or absence of infection was noted. These pa- assessment.
tients were monitored throughout their ICU stay. A re- The neurological evaluation is complicated by the fre-
port on this analysis will follow. quent use of sedative agents in critically ill patients. A1-
710

though the Glasgow C o m a Score is considered to be most FA score represents a valuable approach. The criteria used
useful in this assessment, it is not clear whether the actual and especially the individual values for each of the pa-
or the assumed (in the absence of sedative/relaxant drugs) rameters used in the SOFA score should not be considered
should be used, so that it was decided to include both, at as definitive, but can be altered when sufficient data are
least initially. Importantly, any given score is not estab- collected.
lished indefinitely. This is a continuing process, requiring
regular re-evaluation. Acknowledgements The authors acknowledge the expertise of
The assessment of organ dysfunction/failure remains Critical Care Analytics (Amherst, Mass.) in the analysis of the data
difficult, but we believe that the development of the SO- from the ENAS data base.

References
1. Baue AE (1975) Multiple, progressive, 7. Fagon JY, Chastre J, Novara A, Me- 10. Marshall JC, Cook DJ, Christou NV,
or sequential systems failure. A syn- dioni P, Gibert C (1993) Characteriza- Bernard GR, Sprung CL, Sibbald WJ
drome of the 1970s. Arch Surg 110: tion of intensive care unit patients using (1995) The multiple organ dysfunction
779-781 a model based on the presence or ab- (MOD) score: a reliable descriptor of a
2. Fry DE, Pearlstein L, Fulton RL, sence of organ dysfunction and/or in- complex clinical outcome. Crit Care
Hiram CP (1980) Multiple system or- fection: the ODIN model. Intensive Med 23:1638-1652
gan failure: the role of uncontrolled in- Care Med i9:137- 144 11. Bernard GR, Doig BG, Hudson Get al
fection. Arch Surg 115:136-140 8. Le Gall JR, Lemeshow S, Leleu G et al (I995) Quantification of organ failure
3. Beal AL, Cerra FB (1994) Multiple or- (1995) Customized probability models for clinical trials and clinical practice.
gan failure in the 1990s. JAMA 271: for early severe sepsis in adult intensive Am J Respir Crit Care Med 151:A323
226- 233 care patients. JAMA 237:644-650 (abstract)
4. Deitch EA (1992) Multiple organ fail- 9. Chang RWS, Jacobs S, Lee B (1988) 12. Sibbald WJ, Vincent JL (1995) Round
ure: pathophysiology and potential fu- Predicting outcome among intensive table conference: clinical trials in sepsis.
ture therapy. Ann Surg 216:117-134 care unit patients using computerized Intensive Care Med 21:184-189
5. Goris RJA, Boekhorst TPA (1985) Mul- trend analysis of daily APACHE II
tiple-organ failure. Arch Surg 120: scores corrected for organ system fail-
1109-1115 ure. Intensive Care Med 14:558-566
6. Knaus WA, Draper EA, Wagner DP,
Zimmerman JE (1985) Prognosis in
acute organ-system failure. Ann Surg
202:685 - 693

lands), J. Groeneveld (Amsterdam, The Netherlands), A.


Appendix Gullo (Trieste, Italy), J. Hazelzet (Rotterdam, The Neth-
Contributors to the SOFA project were: J.L. Vincent erlands), J. Ibafiez-Nolla (Barcelona, Spain), J. Labrousse
(Brussels, Belgium, chairman); G. Bernard (Nashville, (Paris, France), M. Lamy (Liege, Belgium), J.R. Le Gall
USA), D. Bihari (Sydney, Australia), C. Brun-Buisson (Paris, France), M. Levi (Amsterdam, The Netherlands),
(Paris, France), H. Burchardi (GOttingen, Germany), M. C. Martin (Marseille, France), D. Miller (Edinburgh,
Langer (Milan, Italy), J. Marshall (Toronto, Canada), United Kingdom), R. Moreno (Lisboa, Portugal), R
C.K. Reinhart (Jena, Germany), P. Surer (Geneva, Swit- Nightingale (Manchester, United Kingdom), M. Nolla-
zerland), L.G. Thijs (Amsterdam, The Netherlands), R. Salas (Barcelona, Spain), G. Nowak (Jena, Germany), M.
Abizanda (Castellon, Spain), M. Antonelli (Rome, Italy), Palazzo (London, United Kingdom), A. Pesenti (Monza,
A. Artigas (Sabadell, Spain), R. Beale (London, United Italy), H. Prange (G6ttingen, Germany), P. Radermacher
Kingdom), G. Berlot (Trieste, Italy), H. Bruining (Rotter- (Ulm, Germany), M. Rogy (Vienna, Austria), C. Ronco
dam, The Netherlands), J. Carlet (Paris, France), V. Cer- (Vicenza, Italy), A. Salgado (Barcelona, Spain), M. Sing-
ny (Hradec Kralove, Czech Republic), E Colardyn er (London, United Kingdom), C. Sprung (Jerusalem, Is-
(Ghent, Belgium), N.M. Dearden (Leeds, United King- rael), J. Strom (Glostrup, Denmark), J. Takala (Kuopio,
dom), D.J. Edwards (Manchester, United Kingdom), J. Finland), D. Teres (Springfield, USA), D. Villers (France),
Fagon (Paris, France), J. Goris (Nijmegen, The Nether- S. Willats (United Kingdom).

You might also like