A Predictive Rule of 30-Day Mortality of Critically Ill Patients after ICU Discharge. A Prospective Observational Study

Résumé

Background: Intensive care unit (ICU) survivors have a high and ongoing risk of death after discharge. Aim: To determine the rate and predictors of 30-day post-ICU mortality in critical care survivors to develop a predictive triage model for ICU discharge.

Methods: A prospective observational cohort study included all consecutive survivors admitted to the ICU. Data were collected between January 2014 and December 2015. Outcomes were assessed by telephone interviews at 30 days after ICU discharge. Univariate and multivariate analyses were performed to identify independent factors associated with 30-day post-ICU mortality used to develop a predictive triage model.

Results: Among 573 ICU-admitted patients, 215 discharged survivors were included. Thirty-four (16%) died within the first month. At ICU admission, patients had, WHO performance status > 3, 72(33.5%), severe acute respiratory failure 130(60.5%), invasive mechanical ventilation, 111(51.6). Mean length of stay, 8.5±9.7days. Multivariate regression analysis identified, (OR, 95%CI, p): SAPS II ≥30, (3.258 [1.1-9.6], <0.032), tachycardia at discharge (heart rate≥90b/mn) (3.024 [1.01- 9.11], <0.049), decline in functional handicap status (15.868 [15.18- 48.56], <0.000), and WHO performance status ≥3 (6.57 [2.03- 21.25], <0.002), as independent risk factors of 30-day post-ICU mortality. AUC/ ROC curve of the predictive triage model, 0.914 (95%CI, [0.86-0.96]).

Conclusion: The present study revealed a high rate of mortality among ICU survivors at 30-day post-ICU discharge. A predictive triage model including the severity at ICU admission, performance status at ICU discharge, decline of physiological reserve, and persistent tachycardia demonstrated good discriminative properties to identify patients at risk of 30-day post-ICU mortality.