Vous êtes sur la page 1sur 3

Journal of Critical Care 41 (2017) 191–193

Contents lists available at ScienceDirect

Journal of Critical Care

journal homepage: www.jccjournal.org

A comparison of pre ICU admission SIRS, EWS and q SOFA scores for
predicting mortality and length of stay in ICU
Shahla Siddiqui, MBBS, DABA, FCCM ⁎, Maureen Chua, MBBs, MMed,
Venkatesan Kumaresh, MBBS, FRCA, EDIC, Robin Choo, MS
a
Anaesthesia and Intensive Care, Khoo Teck Puat Hospital, Singapore
b
MOhh, Singapore
c
KTPH, Singapore
d
Yishun Comm. Hospital, Singapore.

a r t i c l e i n f o a b s t r a c t

Available online xxxx Introduction: The 2015sepsis definitions suggest using the quick SOFA score for risk stratification of sepsis pa-
Keywords: tients among other changes in sepsis definition. Our aim was to validate the q sofa score for diagnosing sepsis
Sepsis 3 and comparing it to traditional scores of pre ICU admission sepsis outcome prediction such as EWS and SIRS in
q SOFA our setting in order to predict mortality and length of stay.
Risk stratification Methods: This was a retrospective cohort study. We retrospectively calculated the q sofa, SIRS and EWS scores of
all ICU patients admitted with the diagnosis of sepsis at our center in 2015. This was analysed using STATA 12.
Logistic regression and ROC curves were used for analysis in addition to descriptive analysis.
Results: 58 patients were included in the study. Based on our one year results we have shown that although q
SOFA is more sensitive in predicting LOS in ICU of sepsis patients, the EWS score is more sensitive and specific
in predicting mortality in the ICU of such patients when compared to q SOFA and SIRS scores.
Conclusion: In conclusion, we find that in our setting, EWS is better than SIRS and q SOFA for predicting mortality
and perhaps length of stay as well. The q Sofa score remains validated for diagnosis of sepsis.
© 2017 Elsevier Inc. All rights reserved.

1. Introduction threshold for ICU admission but is not specific for sepsis. The q SOFA
score is a composite of clinical signs (hypotension b 100 SBP, altered
Since the advent of the new sepsis definitions by The Third Interna- consciousness, GCS b 15, and a respiratory rate N 22 bpm) which has a
tional Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) [1] high internal validity in the Sepsis 3 study for outcome prediction. The
in early 2016 many publications urge the validation of the suggested q score ranges from 0 to 3 points. The presence of 2 or more q SOFA points
SOFA score in local contexts. The authors of the new definitions them- at the onset of infection was associated with a greater risk of death or
selves urge further epidemiological research in this arena. Sepsis is de- prolonged intensive care unit stay [6]. These are outcomes that are
fined as “a life threatening syndrome which is due to a dysregulated more common in infected patients who may be septic than those with
host response and causes organ dysfunction” [2]. Patients admitted uncomplicated infection. Based upon these findings, the Third Interna-
with sepsis to the ICU were evaluated for severity using the SIRS (sys- tional Consensus Definitions for Sepsis recommends q SOFA as a simple
temic inflammatory response score) or EWS (early warning score), prompt to identify infected patients outside the ICU. The internal valid-
however after the new definition has been published we are urged to ity of the q SOFA score for predicting poor outcomes (LOS and mortality)
use q SOFA score in the preadmission period for prediction of adverse was high. The authors concluded that the initial, retrospective analysis
outcomes [3]. SIRS has been the oldest definition of mild sepsis but indicated that q SOFA could be a useful clinical tool, especially to physi-
has been criticized for having a poor specificity [4]. The presence of N 2 cians and other practitioners working outside the ICU to promptly iden-
SIRS criteria in non-infectious cases has led to its use being obsolete in tify infected patients likely to fare poorly [7]. However, because most of
sepsis outcomes prediction. The early warning score based on 6 physio- the data were extracted from only US databases, the task force strongly
logic scores have been widely adopted in UK hospitals and is the norm encourages prospective validation in non-US health care settings to
for pre ICU assessment in sepsis patients [5]. A score of four is a confirm its robustness and potential for incorporation into future clini-
cal management [8].
⁎ Corresponding author. Our aim was to compare the q SOFA score with SIRS and EWS score
E-mail address: Siddiqui.shahla@alexandrahealth.com.sg (S. Siddiqui). traditionally used in our setting for pre ICU risk prediction in sepsis

http://dx.doi.org/10.1016/j.jcrc.2017.05.017
0883-9441/© 2017 Elsevier Inc. All rights reserved.
192 S. Siddiqui et al. / Journal of Critical Care 41 (2017) 191–193

patients. We wished not only to validate q SOFA for diagnosis of sepsis Table 2
but also to study its efficacy in predicting mortality and length of stay. The association between different sepsis assessment tools and mortality.

Model Unadjusted p-Value Adjusteda p-Value


OR (95% CI) OR (95% CI)
1.1. Subjects and methods
q SOFA 2.08 (0.99–4.37) 0.053 2.53 (1.10–5.82) 0.029
SIRS 2.45 (1.01–5.92) 0.047 2.82 (1.09–7.25) 0.032
1.1.1. Study sample
EWS 1.62 (1.22–2.16) 0.001 1.92 (1.24–2.96) 0.003
After obtaining an IRB waiver, we conducted a retrospective cohort
a
study on all adult ICU or High dependency unit admissions in the ICUs Adjusted with age and ethnicity.

of Khoo Teck Puat Hospital, with the presumed diagnosis of ‘sepsis’.


Medical, surgical and cardiac ICUs were included during Jan–December
2015. Exclusion criteria included patients admitted without a diagnosis
of sepsis. Data collected included demographics, SIRS; EWS and q SOFA
scores calculated for each patient from their pre ICU admission comput- with a range of 0–12. The overall mortality was 17.2%. The ICU length
erized vitals chart as well as our outcomes: mortality and ICU Length of of stay (LOS) was 6.3 ± 11.6 days with a range of 1 to 87 days.
stay. Data was recovered using the Philips IntelliSpace Critical Care and All three sepsis assessment tools were found to have associations
anesthesia system used in our ICU as well as the Sunrise Clinical Manag- with mortality in ICU patient subjects (Table 2). We found that a higher
er system for ward and ICU data. All scores were manually calculated by q SOFA score had an unadjusted Odds Ratio (OR) of 2.08 (p-value =
the authors and analysed. Our hospital is a 600 bedded regional hospital 0.053, 95% Confidence Interval (CI) 0.99–4.37) for in patient mortality.
with 36 ICU and High dependency beds. Higher SIRS score had an unadjusted OR of 2.45 (p-value = 0.047, 95%
CI 1.01–5.92) and a higher EWS score had an unadjusted OR of 1.62
(p-value = 0.001, 95% CI 1.22–2.16) for in patient mortality. With
1.1.2. Statistical methods model adjustments for patient's age and ethnicity, a higher q SOFA
Descriptive statistics were carried out. Logistic regression models score had an adjusted Odds Ratio (OR) of 2.53 (p-value = 0.029, 95%
were used to examine the associations between mortality and each of Confidence Interval (CI) 1.10–5.82) for in patient mortality. A higher
the sepsis assessment tools: q SOFA, SIRS and EWS. We assessed each SIRS score had an adjusted OR of 2.82 (p-value = 0.032, 95% CI 1.09–
model's discrimination using the area under the receiver operator char- 7.25) and a higher EWS score had an adjusted OR of 1.92 (p-value =
acteristics (ROC) curve, the C-statistics. Length of stay was naturally a 0.003, 95% CI 1.24–2.96) for in patient mortality.
skew distribution with no zero (and outpatient visits were counted as When we compare the ROC curves of each sepsis assessment tools
1 day of LOS). Zero-truncated Poisson regression models were used to with patient mortality (Fig. 1), EWS has the highest C-statistic, 0.8781,
examine the associations between length of stay and each of sepsis as- followed by SIRS, 0.7073, and q SOFA, 0.6875. The C-statistics is the
sessment tools. All analyses were carried out using Stata 12. area under the ROC curve where the true positive rate (sensitivity) is
plotted in the function of the false positive rate (1-specificity) for differ-
2. Results ent threshold points of a parameter. With adjustments for age and eth-
nicity, the adjusted C-statistics are 0.7432, 0.7739 and 0.9068 for q
58 adult patients were admitted to the ICUs of Khoo Teck Puat Hos- SOFA, SIRS and EWS respectively. Therefore, EWS had the highest unad-
pital with a diagnosis of ‘sepsis’. Our mean age was 64.4 ± 12.9 years justed (univariate) C-statistic to predict mortality. When we take the
(Table 1). Of these 63.8% were Chinese, 17.2% Malay, 12.1% Indian and EWS threshold score of 6, the true positive rate is 0.90 and the false pos-
6.9% others, representing the usual distribution of ethnicities in Singa- itive rate is 0.23. This means that EWS has a stronger power to discrim-
pore. 44.8% were admitted to the surgical ICU (SICU), 32.8% to MICU inate mortality than SIRS and q SOFA because of EWS's wider range
and 22.4% to CCU. 44.8% of the patients had a q SOFA score of ≥ 2, (0−12) then SIRS's (0–4) and q SOFA's (0–3).
62.1% had a SIRS score N 2 and the mean EWS score was 4.5 ± 3.4 The length of stay may have associations with the sepsis assessment
tools as well (Table 3). For every unit increase of q SOFA score, there was
Table 1 an unadjusted log rate increase of 0.32 days in the length of stay (p-
Characteristics of patients. value = 0.001, 95% CI 0.13–0.51). SIRS had an increased unadjusted
log rate of 0.18 (p-value = 0.061, 95% CI –0.01–0.36) and EWS had
Range
0.12 (p-value = 0.008, 95% CI 0.03–0.21). Adjusting the model for the
Age, mean ± SD y 64.4 ± 12.9 24–87
Ethnicity, n (%)
Chinese 37 (63.8)
Malay 10 (17.2)
Indian 7 (12.1)
Others 4 (6.9)
Ward, n (%)
W36 SICU 26 (44.8)
W26 CCU 13 (22.4)
W26 MICU 19 (32.8)
q SOFA, mean ± SD 1.5 ± 1.0 0–3
q SOFA, n (%)
q SOFA b 2 32 (55.2)
q SOFA ≥ 2 26 (44.8)
SIRS, mean ± SD 2.6 ± 1.2 0–4
SIRS, n (%)
SIRS ≤ 2 22 (37.9)
SIRS N 2 36 (62.1)
EWS, mean ± SD 4.5 ± 3.4 0–12
Mortality, n (%)
No 48 (82.8)
Yes 10 (17.2)
ICU length of stay, mean ± SD day 6.3 ± 11.6 1–87

Notes: SD = Standard Deviation. Fig. 1. Combined ROC curves of q SOFA, SIRS and EWS with patient mortality outcome.
S. Siddiqui et al. / Journal of Critical Care 41 (2017) 191–193 193

Table 3 definitions and scores is required before international acceptance can


The association between different sepsis assessment tools and length of stay. occur. Our limitations were the small sample size of sepsis patients
Model Unadjusted p-Value Adjusteda p-Value over one year. In conclusion, we find that in our setting, EWS is better
Coef (95% CI) Coef (95% CI) than SIRS and q SOFA for predicting mortality and perhaps length of
q SOFA 0.32 (0.13–0.51) 0.001 0.36 (0.10–0.63) 0.007 stay as well.
SIRS 0.18 (−0.01–0.36) 0.061 0.16 (−0.04–0.36) 0.109
EWS 0.12 (0.03–0.21) 0.008 0.15 (0.03–0.27) 0.015
a
Contributions
Adjusted with age and ethnicity.

SS - conception, data collection, analysis, manuscript writing


MC - data collection, editing
patient's age and ethnicity, q SOFA had an increase adjusted log rate of
VK - concept, data collection
0.36 days in the length of stay (p-value = 0.007, 95% CI 0.10–0.63).
RC - analysis
SIRS had an increased adjusted log rate of 0.16 days (p-value = 0.109,
95% CI –0.04–0.36) and EWS had 0.15 days (p-value = 0.015, 95% CI
0.03–0.27). All unadjusted and adjusted scores are significant but only References
SIRS adjusted score is non-significant for predicting an increased length
[1] Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al.
of stay. The third international consensus definitions for sepsis and septic shock (sepsis-3).
EWS had the highest pseudo R-square of the zero truncated Poisson JAMA 2016 Feb 23;315(8):801–10.
model of predicting length of stay (unadjusted = 0.1028; adjusted = [2] Giamarellos-Bourboulis EJ, Tsaganos T, Tsangaris I, Lada M, Routsi C, Sinapidis D, et al.
Validation of the new sepsis-3 definitions: proposal for improvement in early risk
0.1912) as compared to SIRS's pseudo R-square (unadjusted = 0.0205; identification. Clin Microbiol Infect 2016 Nov 14;23(2):104–9.
adjusted = 0.0780) and q SOFA's pseudo R-square (unadjusted = [3] Finkelsztein Eli J, et al. Comparison of qSOFA and SIRS for predicting adverse out-
0.0556; adjusted = 0.1257) [9]. EWS therefore could be a better predic- comes of patients with suspicion of sepsis outside the intensive care unit. Crit Care
2017;21(1):73.
tor of LOS compared to SIRS and q SOFA score as well. [4] Balk RA. Systemic inflammatory response syndrome (SIRS): where did it come from
and is it still relevant today? Virulence 2014;5(1):20–6. http://dx.doi.org/10.4161/
3. Conclusions viru.27135.
[5] Bhattacharjee P, Edelson DP, Churpek MM. Identifying patients with sepsis on the
hospital wards. Chest 2016 Jun 30 (Epub 2016 Jun 30).
Based on our one year results we have shown that although q SOFA [6] Shankar-Hari M, Deutschman CS, Singer M. Do we need a new definition of sepsis?
is sensitive in diagnosis of sepsis, it remains deficient in predicting LOS Intensive Care Med 2015 May;41(5):909–11.
[7] Seymour CW, Liu V, Iwashyna TJ, et al. Assessment of clinical criteria for sepsis: For
and mortality in ICU of sepsis patients. The EWS score is more sensitive
the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-
and specific in predicting mortality in the ICU of such patients when 3). JAMA 2016;315(8):762–77.
compared to q SOFA and SIRS scores. Although the q SOFA score was de- [8] Vincent Jean-Louis, Martin Greg S, Mitchell M. Levy q SOFA does not replace SIRS in
signed to accurately predict sepsis diagnosis, it has been shown to have the definition of sepsis. Crit Care 2016;20:210.
[9] IWM Verburg, de Keizer NF, de Jonge E, Peek N. Comparison of regression methods for
good internal validity for outcomes prediction in other initial studies. modeling intensive care length of stay. Salluh JIF, ed. PLoS ONE 2014;9(10):e109684.
However, as urged by the Consensus group local validation for such

Vous aimerez peut-être aussi