Abstract
Prevention of nosocomial Ebola virus (EBOV) infection among patients admitted to an Ebola management centre (EMC) is paramount. Current Médecins Sans Frontières (MSF) guidelines recommend classifying admitted patients at triage into suspect and highly-suspect categories pending laboratory confirmation. We investigated the performance of the MSF triage system to separate patients with subsequent EBOV-positive laboratory test (true-positive admissions) from patients who were initially admitted on clinical grounds but subsequently tested EBOV-negative (false-positive admissions). We calculated standard diagnostic test statistics for triage allocation into suspect or highly-suspect wards (index test) and subsequent positive or negative laboratory results (reference test) among 433 patients admitted into the MSF EMC Kailahun, Sierra Leone, between 1 July and 30 September 2014. 254 (59%) of admissions were classified as highly-suspect, the remaining 179 (41%) as suspect. 276 (64%) were true-positive admissions, leaving 157 (36.3%) false-positive admissions exposed to the risk of nosocomial EBOV infection. The positive predictive value for receiving a positive laboratory result after being allocated to the highly-suspect ward was 76%. The corresponding negative predictive value was 54%. Sensitivity and specificity were 70% and 61%, respectively. Results for accurate patient classification were unconvincing. The current triage system should be changed. Whenever possible, patients should be accommodated in single compartments pending laboratory confirmation. Furthermore, the initial triage step on whether or not to admit a patient in the first place must be improved. What is ultimately needed is a point-of-care EBOV diagnostic test that is reliable, accurate, robust, mobile, affordable, easy to use outside strict biosafety protocols, providing results with quick turnaround time.
