Original article Severity score for predicting in-facility Ebola treatment outcome Jia Bainga Kangbai, MPH a, b, * , Christian Heumann c , Michael Hoelscher, MD a, d , Foday Sahr, MD, PhD e, f , Guenter Froeschl, MD a, d a Center for International Health, University of Munich (LMU), Munich, Germany b Department of Environmental Health Sciences, Njala University, Bo, Sierra Leone c Department of Statistics, University of Munich (LMU), Munich, Germany d Division of Infectious Diseases and Tropical Medicine, University Hospital, LMU Munich, Munich, Germany e Department of Microbiology, College of Medicine and Allied Health Sciences, University of Sierra Leone, Freetown, Sierra Leone f The 34 Military Hospital, Wilberforce, Sierra Leone article info Article history: Received 14 January 2020 Accepted 24 July 2020 Available online 5 August 2020 Keywords: Ebola Ebola treatment center Adult Pediatric Treatment outcomes Predictive score Case fatality rate Sierra Leone abstract Purpose: Sierra Leone recorded the highest incidence rate for the 2013e2016 West African Ebola outbreak. In this investigation, we used the medical records of Ebola patients with different socio- demographic and clinical features to determine the factors that are associated with Ebola treatment outcome during the 2013e2016 West African Ebola outbreak in Sierra Leone and constructed a predictive in-facility mortality score. Methods: We used the anonymized medical records of 1077 laboratory-confirmed pediatric and adult patients with EVD who received treatment at the 34 Military Hospital and the Police Training School Ebola Treatment Centers in Sierra Leone between the period of June 2014 and April 2015. We later determined the in-facility case fatality rates for Ebola, the odds of dying during Ebola treatment, and later constructed a predictive in-facility mortality score for these patients based on their clinical and socio- demographic characteristics. Results: We constructed a model that partitioned the study population into three mortality risk groups of equal patient numbers, based on risk scoring: low (score  e5), medium (score e4 to 1), and high-risk group (score  2). The CFR of patients with EVD belonging to the low- (e5), medium (e4 to 1), and high- (2) risk groups were 0.56%, 9.75%, and 67.41%, respectively. Conclusions: We succeeded in designing an in-facility mortality risk score that reflects EVD clinical severity and can assist in the clinical prioritization of patients with EVD. © 2020 Elsevier Inc. All rights reserved. Introduction Ebola virus disease (EVD) is a severe infection by a member of the filovirus family which causes various symptoms such as fever, hemorrhage, myalgia, and diarrhea [1,2]. The West African EVD outbreak in 2013e2016 affected more than 28,000 individuals and resulted in over 11,000 deaths [3]. Before the 2013e2016 EVD outbreak, there were just over 2300 EVD cases and just over 1500 EVD-related deaths documented globally [4]. Sierra Leone was among the hardest-hit countries, and the country recorded more than 10,000 EVD cases and over 4000 EVD-related deaths during the 2013e2016 EVD outbreak [5]. Several EVD treatment outcome studies [6,7e10] have demonstrated variability (37%e74%) in EVD case fatality rates (CFRs). Such variability has prompted calls for further investigation to understand the reasons for these differ- ences in CFRs and hence offer differentiated EVD treatment and management options. Symptoms of EVD are similar to many trop- ical infections and hamper, therefore, specificity in predictive al- gorithms. Even though EVD disease onset is nonspecific, it is often characterized by symptoms such as fever, myalgia, chills, vomiting, and diarrhea. These symptoms evolve within an incubation period of 2e21 days from the time of infection; mostly within 4e10 days [1,2]. A maculopapular rash, erythema, and desquamation are often visible by the fiftheseventh day of EVD infection and can serve as a valuable differential diagnostic feature for the infection [11]. Pa- tients with EVD may also present with other symptoms including Funding: No part of this study received funding or compensation whatsoever during its conception, execution, or publication. * Corresponding author. Center for International Health, University of Munich (LMU), Munich, Germany. E-mail address: Jia.Kangbai@lrz.uni-muenchen.de (J.B. Kangbai). Contents lists available at ScienceDirect Annals of Epidemiology https://doi.org/10.1016/j.annepidem.2020.07.017 1047-2797/© 2020 Elsevier Inc. All rights reserved. Annals of Epidemiology 49 (2020) 68e74