Performance of a new Risk Score in a Patient Monitoring System in Low-Resource Settings in a Hospital in Zomba, Malawi

Abstract

In Low- and Middle-Income countries (LMIC), it is hard to provide adequate healthcare to critically ill children due to limited resources in staff and equipment. Within these Low-Resource settings (LRS), pediatric mortality and the occurrence of Critical Illness Events (CIE) are high. Continuous patient monitoring of vital signs can be particularly helpful in LRS, especially for critically ill children. GOAL 3 designed a Continuous Patient Monitoring System (CPMS), service, and training fit for LRS, IMPALA, and implemented it in the Pediatric High-Dependency Unit of Zomba Central Hospital in Malawi, Africa. To this IMPALA CPMS, a newly developed Risk Score was added. This Risk Score, adjusted from an earlier developed physiological score, can potentially assign a health risk level with a score between 0-100, show the progress of the patient over time, and function as an early warning system. In this study, the performance of the Risk Score is assessed, based on its distinctive ability to assign a high-risk level to patients who eventually passed away and to patients who had one or more CIE. The study consisted of 192 participants under 5 years of age, whose clinical and score data were analyzed. With a mortality rate of 12% and 140 CIE in total, there were many critically ill children. That group showed the highest average Risk Score of 64 in the death subset, and 37 in all patients with one or more CIE, significantly different from an average score of 23 in participants who had no CIE and survived. With an AUROC in the range of comparable literature, it shows that the Risk Score can distinguish and potentially predict patients who pass away moderately to well, especially in the last 6-8 hours before discharge. Observations and data analysis revealed challenges with the Risk Score availability, which results in an average score coverage of 82%, mainly due to problems with the oxygen saturation probe and nurses not implementing the score system in their routines at the beginning of the study. Optimization and more research have to be done to improve these challenges and validate the Risk Score.