Intensive care unit (ICU) outcomes in Rwanda and the U.S.: clinical course, morbidity and mortality

Abstract

Background: Little epidemiologic data has been published regarding critical illness presentations, management strategies, and outcomes in low-income countries. Global efforts to provide international perspective on critical illness have had an admittedly poor representation from the developing world. The few studies that have been published from these countries demonstrate different pathology and poor outcomes compared to higher income settings. Methods: Demographic and clinical data were prospectively collected on all adult patients admitted in the intensive care units of national referral hospitals of both Kigali and Butare University Teaching Hospitals from August 19, 2013 and October 6, 2014. ICU admission patterns, treatment and outcomes data were compared to Project IMPACT (PI) data from October 1, 2000 to March 21, 2009. PI patient data were fit to Rwanda Mortality Probability Model(R-MPM) to assess generalizability. Results: Of 504 Rwandan ICU admissions, 422 patients met inclusion criteria. Of 399,205 Project IMPACT (PI) ICU admissions, 384,724 were included in the study. Rwandan data included 2 ICUs from 2 hospitals whereas PI included 186 ICUs from 125 hospitals. Rwandan hospitals were smaller, with fewer ICU beds (5.5 per hospital vs 15) compared to PI hospitals. Rwandan ICU population was younger, with median age 34 years compared with 63 years in PI patients. The most common reason for ICU admission in Rwanda was respiratory failure (72.8% of admissions). Within 24 hours of ICU admission, 79.9% Rwandan patients required mechanical ventilation and 18.7% had a Glasgow Coma score (GCS) between 3 and 5. In PI patients, only 30.2% were mechanically ventilated, and only 1.8% had a GCS between 3 and 5. Forty one per cent (41.9%) of Rwandan patients were diagnosed with sepsis. Twelve per cent (12.8%) of Rwandan ICU admissions developed acute respiratory distress syndrome (ARDS). ICU mortality in Rwanda was 43.8% compared with 8.7% in PI population. PI admission hospital mortality probability (MPM0-III) score was 7.8 (IQR 3.2-16.6) compared to 9.4 (IQR 3.8-23.0) for the Rwandan ICU population. Conclusion: Compared to Project IMPACT, patients admitted to Rwandan ICUs were younger, required more mechanical ventilation, and experienced a higher mortality rate. The findings in this study characterize the ICU patient population in one particular low-income country and suggest a ix need to prioritize resource utilization to syndromes and interventions most associated with higher mortality. Specific to Rwanda, more ICU beds and more ventilators are needed in Rwandan University Teaching Hospitals to alter the course of morbidity and mortality.