Development and internal validation of the Simplified Mortality Score for the Intensive Care Unit (SMS-ICU)

Area of Science:

• Critical Care Medicine
• Prognostic Modeling
• Health Informatics

Background:

• Existing ICU mortality prediction scores are complex and prone to missing data, limiting their clinical utility.
• There is a need for a simpler, more applicable score for predicting short-term mortality in intensive care unit patients.

Purpose of the Study:

• To develop and internally validate a novel, simple score (SMS-ICU) for predicting 90-day mortality in adult ICU patients.
• To create a user-friendly tool with high data availability for ICU clinicians.

Main Methods:

• Utilized data from an international cohort of 2139 ICU patients (2009-2016).
• Employed multiple imputation for missing data and binary logistic regression with backward elimination.
• Developed a point-based score and validated its performance using bootstrapping for optimism correction.

Main Results:

• The SMS-ICU includes seven variables (age, lowest systolic blood pressure, malignancy, surgical admission, vasopressors, respiratory support, renal replacement therapy), with 99.5% data availability.
• Achieved an area under the receiver operating characteristic curve of 0.72 (95% CI: 0.71-0.74) for discrimination.
• Demonstrated good calibration (intercept 0.00, slope 0.99) and stable, optimism-corrected performance.

Conclusions:

• The SMS-ICU demonstrates reasonable and stable performance in predicting 90-day mortality.
• Its simplicity and minimal missing data make it a potentially valuable tool for ICU clinicians and researchers.
• External validation is recommended to confirm its performance in diverse populations.