Predictive Validity of Sepsis-3 Definitions and Sepsis Outcomes in Critically Ill Patients: A Cohort Study in 49 ICUs in Argentina

Abstract

Objectives: The new Sepsis-3 definitions have been scarcely assessed in low- and middle-income countries; besides, regional information of sepsis outcomes is sparse. Our objective was to evaluate Sepsis-3 definition performance in Argentina.

Design: Cohort study of 3-month duration beginning on July 1, 2016.

Settings: Forty-nine ICUs.

Patients: Consecutive patients admitted to the ICU with suspected infection that triggered blood cultures and antibiotic administration.

Interventions: None.

Measurements and main results: Patients were classified as having infection, sepsis (infection + change in Sequential Organ Failure Assessment ≥ 2 points), and septic shock (vasopressors + lactate > 2 mmol/L). Patients on vasopressors and lactate less than or equal to 2 mmol/L (cardiovascular dysfunction) were analyzed separately, as those on vasopressors without serum lactate measurement. Systemic inflammatory response syndrome was also recorded. Main outcome was hospital mortality. Of 809 patients, 6% had infection, 29% sepsis, 20% cardiovascular dysfunction, 40% septic shock, and 3% received vasopressors with lactate unmeasured. Hospital mortality was 13%, 20%, 39%, 51%, and 41%, respectively (p = 0.000). Independent predictors of outcome were lactate, Sequential Organ Failure Assessment score, comorbidities, prior duration of symptoms (hr), mechanical ventilation requirement, and infection by highly resistant microorganisms. Area under the receiver operating characteristic curves for mortality for systemic inflammatory response syndrome and Sequential Organ Failure Assessment were 0.53 (0.48-0.55) and 0.74 (0.69-0.77), respectively (p = 0.000).

Conclusions: Increasing severity of Sepsis-3 categories adequately tracks mortality; cardiovascular dysfunction subgroup, not included in Sepsis-3, has distinct characteristics. Sequential Organ Failure Assessment score shows adequate prognosis accuracy-contrary to systemic inflammatory response syndrome. This study supports the predictive validity of Sepsis-3 definitions.