The size of the west African Ebola virus disease outbreak led to the urgent establishment of Ebola holding unit facilities for isolation and diagnostic testing of patients with suspected Ebola virus disease. Following the onset of the outbreak in Sierra Leone, patients presenting to Connaught Hospital in Freetown were screened for suspected Ebola virus disease on arrival and, if necessary, were admitted to the on-site Ebola holding unit. Since demand for beds in this unit greatly exceeded capacity, we aimed to improve the selection of patients with suspected Ebola virus disease for admission by identifying presenting clinical characteristics that were predictive of a confirmed diagnosis.

In this retrospective cohort study, we recorded the presenting clinical characteristics of suspected Ebola virus disease cases admitted to Connaught Hospital's Ebola holding unit. Patients were subsequently classified as confirmed Ebola virus disease cases or non-cases according to the result of Ebola virus reverse-transcriptase PCR (EBOV RT-PCR) testing. The sensitivity, specificity, positive predictive value, negative predictive value, and likelihood ratio of every clinical characteristic were calculated, to estimate the diagnostic accuracy and predictive value of each clinical characteristic for confirmed Ebola virus disease.

Between May 29, 2014, and Dec 8, 2014, 850 patients with suspected Ebola virus disease were admitted to the holding unit, of whom 724 had an EBOV RT-PCR result recorded and were included in the analysis. In 464 (64%) of these patients, a diagnosis of Ebola virus disease was confirmed. Fever or history of fever (n=599, 83%), intense fatigue or weakness (n=495, 68%), vomiting or nausea (n=365, 50%), and diarrhoea (n=294, 41%) were the most common presenting symptoms in suspected cases. Presentation with intense fatigue, confusion, conjunctivitis, hiccups, diarrhea, or vomiting was associated with increased likelihood of confirmed Ebola virus disease. Three or more of these symptoms in combination increased the probability of Ebola virus disease by 3·2-fold (95% CI 2·3–4·4), but the sensitivity of this strategy for Ebola virus disease diagnosis was low. In a subgroup analysis, 15 (9%) of 161 confirmed Ebola virus disease cases reported neither a history of fever nor a risk factor for Ebola virus disease exposure.

Discrimination of Ebola virus disease cases from patients without the disease is a major challenge in an outbreak and needs rapid diagnostic testing. Suspected Ebola virus disease case definitions that rely on history of fever and risk factors for Ebola virus disease exposure do not have sufficient sensitivity to identify all cases of the disease.

None.