Professional Documents
Culture Documents
Vincent2010 Sepsis
Vincent2010 Sepsis
Objective: Severity scores such as Acute Physiology and difficult to use them at the bedside and there is considerable inter-
Chronic Health Evaluation II have been advocated as entry criteria observer variability in score calculation. Inclusion of chronic health
for clinical trials and in clinical decision-making. We present ten and age points in severity scores may prevent younger, previously
reasons why we believe this approach is not appropriate and may healthy patients, with similar acute physiological dysfunction and
even be detrimental. therefore total lower severity scores, from trial inclusion or from
Data Sources: Available relevant literature from authors’ per- receiving therapies that may be beneficial.
sonal databases and personal knowledge of past and future Conclusions: We believe severity of illness scores are poor
clinical trial development. surrogates for risk stratification and should not be used as a
Data Synthesis: Severity scores were not designed for use in criterion for patient enrollment into clinical trials or as the basis
individual patients or for therapeutic decision-making for specific for individual treatment decisions. (Crit Care Med 2010; 38:
interventions. Difficulties with the time window needed to calculate 283–287)
these scores and the need to administer therapies early further limit KEY WORDS: APACHE II; risk prediction; clinical trial; sepsis
their use in this context. The complex nature of the scores makes it studies; organ failure scores; outcomes; critical illness
W hether or not we should context is shown in the package insert for may be more effective in less severely ill
use severity scores as an drotrecogin alfa (activated), which indi- patients (3) and in the Transfusion Re-
entry criterion for clinical cates that the drug should be restricted to quirements in Critical Care trial, the ben-
trials or in deciding the sickest patients, that is, those with an efits of the restrictive transfusion strategy
whether to start a specific therapeutic Acute Physiology and Chronic Health were more marked in the group with low
intervention is a timely question. The ra- Evaluation (APACHE) II score ⬎25. This APACHE II scores (4). One could, there-
tionale for using severity scores in this recommendation followed a subgroup fore, consider the relationship between
way is the seemingly logical proposition analysis of the data from the initial ran- disease severity and potential to respond
that the intervention should be limited to domized, controlled clinical trial (Protein to an intervention to be bell-shaped
the sickest group of patients, who pre- C Worldwide Evaluation in Severe Sepsis rather than linear (Fig. 1B). At both ex-
sumably would be the most likely to ben- [PROWESS]) (1), which showed that pa- tremes of disease severity, the chances of
efit. This approach has been applied to tients with an APACHE II score ⬎25 were detecting a survival benefit from any new
drugs already available and is currently more likely to benefit from the drug. Fu- therapy are limited with patients highly
being used in ongoing trials of new ther- ture clinical trials of new therapeutic in- likely to survive (low severity) or die
apeutic agents. One key example of how terventions may limit enrollment to pa- (high severity) regardless of the thera-
severity scores have been used in this tients with a severity score above a peutic intervention. In this case, study
threshold value to maximize the chances enrollment should be limited to a
of success. midrange of severity scores to focus on
*See also p. 334.
From the Department of Intensive Care (J-LV), In an analysis of clinical trials of anti- patients who are sick enough to benefit
Erasme University Hospital, Free University of Brussels, inflammatory agents in patients with sep- but not so sick that they are about to die.
Belgium; Infectious Disease Division (SMO), Memorial sis, Eichacker et al reported that treatment An approach based on this suggested bell-
Hospital of Rhode Island, The Warren Alpert Medical efficacy was dependent on the risk of death shaped relationship has been proposed in
School of Brown University, Providence, RI; and De-
partment of Surgery and the Interdepartmental Division (2). These authors propose that the patho- an ongoing sepsis trial and has already
of Critical Care Medicine (JCM), University of Toronto, physiological events found in the sickest been used in some studies. As one exam-
St. Michael’s Hospital, Toronto, Ontario, Canada. patients closely reflect the pathology found ple, Gutierrez et al (5), in a study of
Dr. Vincent has consulted for Eli Lilly, GSK, Artisan, in preclinical animal models of sepsis, in gastric tonometry, selected patients with
AstraZeneca, Eisai, Wyeth, and NovoNordisk; has re-
ceived honoraria/speaking fees from GSK, Eli Lilly, and which placebo mortality rates approach- an APACHE II score between 15 and 25;
Roche; and has received grant support from Esai, ing 100% are the norm. However, the they showed that prevention of a decrease
Artisan, and Eli Lilly. The remaining authors have not assumption that there is a linear relation- in intramucosal pH was associated with
disclosed any potential conflicts of interest. ship between severity and potential to reduced mortality rates.
For information regarding this article, E-mail:
jlvincen@ulb.ac.be respond to therapy may be too simplistic. In view of this, we believe that a cutoff
Copyright © 2009 by the Society of Critical Care Indeed, some interventions may have severity score threshold should not be
Medicine and Lippincott Williams & Wilkins greater benefit in patients with moderate used as a criterion for patient inclusion in
DOI: 10.1097/CCM.0b013e3181b785a2 disease severity. As examples, antibiotics clinical trials or to direct therapy and will