Predictors of mortality in a paediatric intensive care unit in Kigali, Rwanda

Abstract

Background: The enormous burden of critical illness in resource-limited settings has led to a growing interest in paediatric critical care in these regions. However, published data on the practice of critical care and patient outcomes in these settings are scant.

Objective: This study sought to identify risk factors associated with mortality in the newly established Paediatric Intensive Care Unit (PICU) at Kigali University Teaching Hospital (KUTH) in Rwanda and test the predictive ability of a newly devised mortality risk score, the modified PRISM (MP) score.

Methods: All admissions to the PICU at KUTH from October 2012 to October 2014 were included. Demographic and physiological data on each patient were gathered and each was assigned a MP score. This prospective cross-sectional study examined the association between the characteristics and physiological status of these patients and mortality. Using logistic regression, factors associated with mortality in the PICU were analysed.

Results: A total of 213 children were admitted to the PICU during the study period. Three patients were excluded because of missing data. Of this total, 59% were male, 25% were neonates and nearly 60% were moderately to severely malnourished. The overall mortality rate was 50%. On bivariate analysis, factors associated with increased mortality were male sex, use of vasoactive medications, a MP score ≥ 5, a discharge diagnosis of septic shock, and malnutrition on admission. On multivariate analysis, only the use of vasoactive drugs [odds ratio (OR) 12.24, 95% confidence interval (CI) 4.4–35.4, p < 0.001] and MP score ≥ 5 (OR 16.1, CI 6.3–40.8, p < 0.001) were associated with mortality.

Conclusion: The observed mortality rate was in the range reported in other resource-limited settings. The initial attempt to create and implement a risk of mortality tool for this setting determined a score that could identify those patients at higher risk of mortality. In PICUs in resource-limited settings, the gathering of data and use of severity of illness tools could improve care in a number of ways.

Keywords:

• Critical care
• Resource-limited settings
• Sepsis
• Mechanical ventilation

Abbreviations:

• PRISM, paediatric risk of mortality
• SOI, severity of illness
• PCC, paediatric critical care
• PICU, paediatric critical care unit
• MP, modified PRISM