The institutional review board at Beijing 302 Hospital and the Sierra Leone Ethics and Scientific Review Committee approved this project. These committees waived the requirement to obtain informed consent during the West African Ebola outbreak.

From Oct. 1st, 2014 to Jan. 18th, 2015, the Chinese Medical Team in the Jui Government Hospital, Sierra Leone admitted 661 patients and 269 of them were diagnosed EVD. Noticeably, Jui Government Hospital was positioned as a holding center from Oct 1st, 2014 to Dec 31st, 2014, and upgraded as an Ebola Treatment Center (ETC) on Jan 1^st, 2015. The admission of patients was coordinated by the National Emergency Response Center (NERC). First of all, venous blood of the patients was immediately taken for sampling. They were then hospitalized and the samples were examined in the Chinese portable biosafety level 3 laboratories within the Jui Government Hospital. While waiting for the test results, patients were treated based on their clinical presentations. All the patients were given oral rehydration salts as a routine treatment, and the dose was dependent on the severity of dehydration. Intravenous administration of supplements was given under certain conditions. Patients with headache and/or muscle pain were given Acetaminophen or Ibuprofen. Patients with fever were given Cefixime or anti-infective Ciprofloxacin and anti-malaria Compound Naphthoquine Phosphate Tablets. Patients with upper abdominal pain or burning sensation were given antacid drugs such as Ranitidine or Omeprazole. Patients feeling fretful or insomnia were given Diazepam. A few patients were offered intravenous lactated Ringer’s solution. All the treatments were performed in compliance with ‘Clinical management of patients with viral haemorrhagic fever A pocket guide for the front-line health worker’ (http://www.who.int/csr/resources/publications/clinical-management-patients/en/) and ‘Manual for the care and management of patients in Ebola Care Units/ Community Care Centres Interim emergency guidance’ (http://www.who.int/csr/resources/publications/ebola/patient-care-CCUs/en/). Patients confirmed with Ebola infection were reported to NERC, and were transported to other treatment centers before December 31^st, 2014, or stayed in the special treatment zone of Jui Government Hospital after the hospital was upgraded to ETC. Ebola-negative patients were arranged for corresponding treatments or discharge.

Sierra Leone is one the most severely affected countries. However, due to its weak health system and lack of human resources, it was extremely difficult to obtain complete patient data during the outbreak. Holding center faced even worse situation because EVD confirmed patients were transferred to different ETCs after the diagnosis. Of the 269 EVD cases, 201 subjects were diagnosed when Jui Government Hospital functioned as a holding center. Only 7 patients were obtained relatively complete information. Sixty eight patients were confirmed EVD between January 1^st and January 18th, 2015 after the hospital was upgraded to ETC, in which 56 cases provided relatively complete information, and the other 12 patients died soon after the admission and did not give enough information for the study. Patient data included demographic information such as gender, age and job; epidemiological history of attending traditional funerals; history of contacting with EVD patients; symptoms; clinical signs, etc. Patients were sent to the hospital ward after the healthcare workers evaluated the general situations and filled out the Ebola case investigation form or viral hemorrhagic fever case investigation form formulated by NERC. The EVD Patient Observation Sheet was completed daily after routine check, and all the sheets were transported either via closed-circuit surveillance system or by WiFi after being photographed. In order to keep the results accurate, all the data was recorded and compared separately in the Excel database by two healthcare professionals. Discharged patients were followed by calling randomly to patients themselves or their families or related treatment centers and the information of their conditions (whether they were cured/ discharged or died, and the date of cure or death) were being accessed.

Viral load was examined upon the admission to the hospital. Detection of virus was performed by the Chinese-CDC portable biosafety level 3 laboratories based on previously published method [14]. These laboratories accepted specimens from the Jui Government Hospital as well as specimens transported by NERC from other ETC and holding centers. After recording the sign-in information in the biosafety level 3 laboratory, technicians inactivated the specimens by incubating in the water bath at 60 degree for 1 hour. Then 50 ul of inactivated specimen was used to isolate RNA by Viral RNA Isolation Kit (Life Technologies, Grand Island, NY, USA) and DNA by Automatic DNA Extractor (Life Technologies, Grand Island, NY, USA). The isolated RNA and DNA were tested in biosafety level 2 laboratory. Thermal cycling parameters of the real-time reverse transcription-polymerase chain reaction were 30 min at 42°C followed by 10 min at 95°C and a 40 cycles of amplification. The range of cycle threshold was between 18.34 and 35.81 and the range of viral load was between 10³ and 3.9x10⁸ copies/ml in our study population.

Data were analyzed by PASW statistics 18.0 software. Pearson's chi-squared test was applied to compare dichotomous variables. The Yates’s correction was applied when the expected (theoretical) frequency was <5 and ≥ 1. The survival curve was estimated by Kaplan-Meier method. The Cox proportional hazard model was used for multi-factor analysis.

In our study population, businessman (17 cases) and students (13 cases) were the two major professions. Of the 63 patients, 27 subjects were non-survivors and 36 subjects were survivors; 35 subjects had clear contact history, in which 11 cases were through funerals and 10 cases were through family members. As shown in Table 1, common symptoms included fever (80.95%), diarrhea (49.21%), vomiting (50.79%), fatigue (90.48%), anorexia (87.30%), abdominal pain (49.21%), chest pain (55.56%), muscle pain (74.60%), joint pain (77.78%), headache (65.08%), cough (39.68%), difficulty in breathing (41.27%), difficulty in swallowing (42.86%), sore throat (38.10%), jaundice (26.98%), redeye (28.57%), skin rash (6.35%), hiccups (19.05%), pain behind eye (12.70%), coma (7.94%), confusion (14.29%), and bleeding (9.52%).

We first investigated patient’s age and the prognosis of EVD. Based on WHO and other literatures [9], patients were divided into two groups (≥40 years old and <40 years old). Log-rank test showed that the ≥40 years old group had moderately shorter survival time than the <40 years old group (P = 0.087). We further examined the viral load and the survival of EVD patients. Since 10⁶ is the closest integer to the median value of the viral load of our study population, we divided the cases into two subsets according to the value. Data showed that patients with viral titer higher than 10⁶ copies/ml presented significantly shorter survival time than those whose viral titer were lower than 10⁶ copies/ml (P = 0.005, Fig 1).

We also examined the correlation between clinical symptoms and the survival of EVD. Using Pearson chi-square test, we found that fever, diarrhea, vomiting, fatigue, anorexia, abdominal pain, chest pain, muscle pain, joint pain, headache, cough, difficulty in breathing, difficulty in swallowing, sore throat, jaundice, redeye, skin rash, hiccups, pain behind eye, coma, confusion, and bleeding were not significantly associated with EVD death, whereas symptoms such as chest pain, coma and confusion showed significantly association with EVD mortality (P = 0.040, P = 0.007, and P = 0.022, Table 1). However, value of confusion lost statistical significance after corrected for continuity (P = 0.055, Table 1). Furthermore, we analyzed the positive predictive values (PV+) and negative predictive values (PV-) of chest pain, coma and confusion. Data showed that chest pain presented 54.3% of PV+ and 71.4% of PV-; coma presented 100% of PV+ and 62.1% of PV-; confusion presented 60.6% of PV+ and 76.7% of PV-. In addition, we investigated whether coma, chest pain, confusion, and viral load were associated with each other using Pearson's chi-squared test. No significant difference was identified (coma vs. chest pain, p = 0.371; coma vs. confusion, p = 0.704; chest pain vs. confusion, p = 0.469, coma vs. viral load, p = 0.286; chest pain vs. viral load, p = 0.269; confusion vs. viral load, p = 0.847).

One-way ANOVA analysis sometimes cannot reflect the combined effect of multiple variables on EVD. Since it is difficult to obtain complete patient information during the outbreak of EVD, whereas it is biased and wasting of data to rule out patients with relatively non-detailed information, applying COX's proportional hazard model to deal with this kind of data set is necessary. We found that that patient’s age, the symptom of confusion, and viral load were the significantly associated with the survival of EVD cases (P = 0.017, P = 0.002, and P = 0.027, respectively, Table 2).