Baseline measure
|
Cut-offa
|
Number of patients at increased risk using cut-off (n [%])
|
Odds ratio (95% CI)
|
AUCb
|
---|
Protein C (%)
|
<40
|
243 (31.4%)
|
2.12 (1.55–2.89)
|
58.9%
|
Protein S (%)
|
<46
|
239 (31.5%)
|
1.91 (1.38–2.64)
|
57.7%
|
Antithrombin III (%)
|
<53
|
240 (31.4%)
|
2.32 (1.70–3.18)
|
60.1%
|
interleukin-6 (pg/ml)
|
≥704.6
|
252 (31.2%)
|
2.21 (1.63–2.99)
|
59.7%
|
Prothrombin time (seconds)
|
≥18.4
|
240 (31.5%)
|
1.89 (1.38–2.58)
|
57.4%
|
D-dimer (μg/ml)
|
≥4.45
|
241 (31.8%)
|
1.51 (1.11–2.05)
|
55.1%
|
Cardiovascular SOFA
|
≥4
|
259 (30.8%)
|
1.63 (1.21–2.18)
|
56.0%
|
Respiratory SOFA
|
≥4
|
257 (31.2%)
|
1.76 (1.27–2.44)
|
55.5%
|
Renal SOFA
|
≥1
|
258 (30.8%)
|
2.14 (1.55–2.95)
|
58.6%
|
Hematologic SOFA
|
≥2
|
259 (30.8%)
|
1.69 (1.20–2.38)
|
54.6%
|
Hepatic SOFA
|
≥2
|
239 (31.3%)
|
1.31 (0.89–1.93)
|
52.1%
|
- aCut-off based on maximum sensitivity and specificity when both were ≥ 40% for predicting 28-day mortality. Using a cut-off for each measure allowed comparison of odds ratios and treatment interactions on a consistent binary scale across variables. bArea under the receiver operating characteristic curve (AUC) based on 28-day mortality outcome in logistic regression models with the cut-off as the univariate independent variable; this is a combined measure of sensitivity and specificity. CI, confidence interal; PROWESS, Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis; SOFA, Sequential Organ Failure Assessment.