Prediction of 28-Day Mortality in Patients With Sepsis: A Prospective Observational Study

Background: Organ-failure scores are used to judge the risk of death in sepsis, but a single set of vital signs is often treated as if it carried the same information. Indian intensive-care series report high hospital mortality, and local performance of the Sequential Organ Failure Assessment (SOFA) score cannot be assumed from those series. Materials and Methods: One hundred adults treated for sepsis and managed in intensive care were enrolled over one year in Hubballi, Karnataka. The recorded endpoint was death by day 28. SOFA score, age, sex, blood-culture result, diabetes, HbA1c, ventilatory mode and one set of mean arterial pressure, heart rate, respiratory rate and temperature were compared between patients who died and those alive at day 28. The primary analysis was logistic regression and the area under the receiver operating characteristic curve (AUROC) for the SOFA score. Results: Twenty-eight patients (28%) had died by day 28. Median SOFA was 11 (interquartile range 7–12) in those who died and 6 (3–11) in survivors (p = 0.001). Each one-point rise in SOFA was associated with an odds ratio for death of 1.22 (95% confidence interval 1.08–1.38; p = 0.002). The AUROC was 0.71 (0.59–0.81). Mortality was 0/10 when SOFA was 0 or 1, 4/27 (14.8%) at 2–5, 7/24 (29.2%) at 6–9 and 17/39 (43.6%) at 10–14. Age, sex, culture status, diabetes, HbA1c, ventilation, blood pressure, heart rate, respiratory rate, temperature and length of ICU stay were not associated with death. Adding age and mechanical ventilation to SOFA did not improve discrimination (apparent AUROC 0.72). Conclusion: In this cohort the SOFA score was the only measured variable that separated 28-day deaths from survivors, and the separation was moderate. A single round of vital signs did not. The estimate is too imprecise, and the sample too small, to support a new prediction rule.