Table 3.
Model | O.R. | 95% C.I. |
---|---|---|
Baseline (24-hour) model* | ||
Delayed sepsis (onset ≥2 days after admission) | 12.06 | (4.64, 31.36) |
Inter-facility transfer | 3.41 | (1.39, 8.37) |
Septic shock | 3.49 | (1.30, 9.36) |
SOFA score ≥6 in 1st 24 hours | 2.12 | (0.83, 5.43) |
Post resuscitation (72-hour) model** | ||
Delayed sepsis (onset ≥2 days after admission) | 11.37 | (4.27, 30.28) |
Inter-facility transfer | 2.75 | (1.11, 6.83) |
Septic shock | 2.80 | (1.04, 7.52) |
SOFA score ≥5 at 72 hours | 5.03 | (2.00, 12.62) |
Definition of abbreviations: O.R., odds ratio; C.I., confidence interval; SOFA, sequential organ failure assessment score.
Area under receiver operator curve = 0.84 (0.77, 0.91); Hosmer-Lemeshow goodness of fit = 0.076
Area under receiver operator curve = 0.86 (0.80, 0.92); Hosmer-Lemeshow goodness of fit = 0.734
Adjusted odds ratios were derived using the multivariable logistic regression that included all listed variables in the model simultaneously. Baseline (24 hour) model included age, gender, Charlson comorbidity score, delayed sepsis, inter-facility transfer, septic shock status, and dichotomized SOFA scores where cut-off was calculated by maximizing sensitivity and specificity along the receiver operating characteristic curves. Final reduced set of variables were selected by a stepwise model selection at a significance level of 0.10.