Relationship between mortality and first-day events index from routinely gathered physiological variables in ICU...

Area of Science:

• Critical Care Medicine
• Health Informatics
• Biostatistics

Background:

• Traditional mortality prediction systems in intensive care units (ICUs) often rely on complex data.
• There is a need for simpler, yet effective, methods to predict patient outcomes in the ICU.

Purpose of the Study:

• To evaluate if routine ICU monitoring data, collected during the first day of admission, can predict mortality similarly to traditional systems.
• To develop a prognostic index using readily available ICU data.

Main Methods:

• A prospective, observational, multicenter study (EURICUS II) involving 17,598 patients across 55 European ICUs.
• Hourly data on systolic blood pressure, heart rate, and oxygen saturation events were collected.
• An events index was constructed, along with mortality prediction models incorporating patient characteristics like age and Glasgow Coma Score.

Main Results:

• The first-day events index was significantly associated with hospital mortality (AUC 0.666).
• A comprehensive prognostic index, including the events index, age, pre-admission location, and Glasgow Coma Score, achieved an AUC of 0.818 for mortality prediction.
• The model demonstrated good validation using the Jackknife method.

Conclusions:

• A prognostic index derived from routine ICU data and basic patient characteristics performs comparably to traditional systems.
• This approach may facilitate the automated construction of ICU prognostic indexes.
• Routine data offers a valuable resource for predicting patient outcomes in critical care settings.