Table 2. Performance Metrics for Classification of Admission (All Sepsis vs Nonsepsis).
All sepsis classification (n = 1663) | Performance metric (95% CI) | ||||||
---|---|---|---|---|---|---|---|
Accuracy | Sensitivity | Specificity | False negativea | False positiveb | Balanced accuracyc | Diagnostic odds ratiod | |
PSS ≥5 | 0.54 (0.54-0.55) | 0.95 (0.93-0.96) | 0.53 (0.53-0.54) | 0.05 (0.04-0.07) | 0.47 (0.46-0.47) | 0.74 (0.73-0.74) | 19.82 (16.01-24.52) |
PSS ≥6 | 0.63 (0.62-0.63) | 0.92 (0.91-0.93) | 0.62 (0.61-0.62) | 0.08 (0.07-0.09) | 0.38 (0.38-0.39) | 0.77 (0.76-0.78) | 18.63 (15.59-22.26) |
PSS ≥7 | 0.69 (0.69-0.69) | 0.89 (0.87-0.90) | 0.68 (0.68-0.69) | 0.11 (0.10-0.13) | 0.32 (0.31-0.32) | 0.79 (0.78-0.79) | 17.44 (14.94-20.34) |
PSS ≥8 | 0.74 (0.73-0.74) | 0.85 (0.83-0.87) | 0.73 (0.73-0.74) | 0.15 (0.13-0.17) | 0.27 (0.26-0.27) | 0.79 (0.78-0.80) | 15.96 (13.92-18.30) |
PSS ≥9 | 0.78 (0.77-0.78) | 0.82 (0.80-0.84) | 0.77 (0.77-0.78) | 0.18 (0.16-0.20) | 0.23 (0.22-0.23) | 0.80 (0.79-0.81) | 15.78 (13.90-17.91) |
PSS ≥10 | 0.81 (0.80-0.81) | 0.78 (0.76-0.80) | 0.81 (0.80-0.81) | 0.22 (0.20-0.24) | 0.19 (0.19-0.20) | 0.80 (0.79-0.81) | 15.10 (13.41-16.99) |
SIRS | 0.43 (0.43-0.44) | 0.95 (0.94-0.96) | 0.42 (0.41-0.42) | 0.05 (0.04-0.06) | 0.58 (0.58-0.59) | 0.68 (0.68-0.69) | 13.49 (10.82-16.81) |
qSOFA | 0.70 (0.69-0.70) | 0.83 (0.81-0.85) | 0.69 (0.69-0.70) | 0.17 (0.15-0.19) | 0.31 (0.30-0.31) | 0.76 (0.75-0.77) | 11.08 (9.73-12.61) |
SOFA | 0.45 (0.44-0.45) | 0.97 (0.97-0.98) | 0.43 (0.43-0.44) | 0.03 (0.02-0.03) | 0.57 (0.56-0.57) | 0.70 (0.70-0.71) | 28.60 (21.12-38.73) |
Abbreviations: PSS, Predicting Sepsis Score; qSOFA, quick Sepsis-Related Organ Failure Assessment; SIRS, Systemic Inflammatory Response Syndrome; SOFA, Sequential Organ Failure Assessment.
False negative indicates the proportion of missed true sepsis cases.
False positive indicates the proportion of nonsepsis cases falsely classified as sepsis.
Balanced accuracy is calculated as sensitivity plus specificity divided by 2.
Diagnostic odds ratio is calculated as the positive likelihood ratio divided by the negative likelihood ratio.