Predicting Mortality in Low-Income Country ICUs: The Rwanda Mortality Probability Model (R-MPM)

Area of Science:

• Critical Care Medicine
• Health Services Research
• Epidemiology

Background:

• Intensive Care Unit (ICU) risk prediction models are crucial for quality improvement and research in high-income countries.
• Existing models are often not validated for low-income countries due to differing patient demographics and limited laboratory data.
• This study addresses the need for validated risk prediction tools in resource-limited settings.

Purpose of the Study:

• To validate the Mortality Probability Admission Model, version III (MPM0-III) in Rwandan ICUs.
• To develop a new, locally relevant risk prediction model for low-income countries.
• To improve critical care research and quality improvement initiatives in low-income countries.

Main Methods:

• Prospective data collection of adult ICU patients in two Rwandan public ICUs.
• Assessment of the discrimination and calibration of the MPM0-III model.
• Development of the Rwanda Mortality Probability Model (R-MPM) using stepwise logistic regression and bootstrapping.

Main Results:

• The MPM0-III demonstrated modest predictive ability (AUC 0.72) with poor calibration (p=0.024).
• The R-MPM was developed using five variables: age, infection, hypotension/shock, Glasgow Coma Scale, and heart rate.
• The R-MPM showed significantly improved predictive power (AUC 0.81, p=0.154 for calibration).

Conclusions:

• The MPM0-III has limited utility for predicting mortality in Rwandan ICU patients.
• The R-MPM offers a more accurate and parsimonious alternative for risk prediction in low-income countries.
• The R-MPM has the potential to enhance the reliability of critical care research and quality improvement in resource-limited settings.