Presence and Persistence of Ebola or Marburg Virus in Patients and Survivors: A Rapid Systematic Review

Abstract

Background: The 2013-15 Ebola outbreak was unprecedented due to sustained transmission within urban environments and thousands of survivors. In 2014 the World Health Organization stated that there was insufficient evidence to give definitive guidance about which body fluids are infectious and when they pose a risk to humans. We report a rapid systematic review of published evidence on the presence of filoviruses in body fluids of infected people and survivors.

Methods: Scientific articles were screened for information about filovirus in human body fluids. The aim was to find primary data that suggested high likelihood of actively infectious filovirus in human body fluids (viral RNA). Eligible infections were from Marburg virus (MARV or RAVV) and Zaire, Sudan, Taï Forest and Bundibugyo species of Ebola. Cause of infection had to be laboratory confirmed (in practice either tissue culture or RT-PCR tests), or evidenced by compatible clinical history with subsequent positivity for filovirus antibodies or inflammatory factors. Data were extracted and summarized narratively.

Results: 6831 unique articles were found, and after screening, 33 studies were eligible. For most body fluid types there were insufficient patients to draw strong conclusions, and prevalence of positivity was highly variable. Body fluids taken >16 days after onset were usually negative. In the six studies that used both assay methods RT-PCR tests for filovirus RNA gave positive results about 4 times more often than tissue culture.

Conclusions: Filovirus was reported in most types of body fluid, but not in every sample from every otherwise confirmed patient. Apart from semen, most non-blood, RT-PCR positive samples are likely to be culture negative and so possibly of low infectious risk. Nevertheless, it is not apparent how relatively infectious many body fluids are during or after illness, even when culture-positive, not least because most test results come from more severe cases. Contact with blood and blood-stained body fluids remains the major risk for disease transmission because of the known high viral loads in blood.

Fig 1. Study selection procedure.

Fig 2. Viral load from blood samples in filovirus patients.

Results are for blood or blood products (serum or plasma) until day 24 of illness. Units are as stated in cited articles. Leftside panels a-c: culture only detection methods. Right-side panels d-f: RT-PCR detection only. Bottom chart value = stated limit of detection ([18,25,26,38], all data in panel d) or implied detection threshold (all other sources). Panel source data: a. 1 patient in [20]; b. many patients in [7]; c. averages from many patients in [14]; d. one patient from each of [32,34,39], two patients in [28], four patients in [19] (including the two patients in [28]), many patients in [24]; e. four patients in [18] who are same four patients as in [19] (part of panel d, also duplicated two patients in [28]); f. one patient from each of [25,26,38], six patients in [45] and many patients in [15].

Fig 3. Probability of positivity for all samples tested by culture through day 110 post illness onset.

The numbers in parenthesis after each fluid type indicate the number of patients who provided samples for each body fluid. The numbers on the right side are the mean probability for positivity (also shown as a cross mark) with 95% confidence interval in parentheses (shown as lines both sides of the cross).

Fig 4. Probability of positivity for all samples tested by RT-PCR through day 110 of illness onset.

The numbers in parenthesis after each fluid type indicate the number of patients who provided samples for each body fluid. The numbers on the right side are the mean probability for positivity (also shown as a cross mark) with 95% confidence interval in parentheses (shown as lines both sides of the cross).

Fig 5. Probability of positivity for all samples tested by RT-PCR days 1–16 of illness onset.

The numbers in parenthesis after each fluid type indicate the number of patients who provided samples for each body fluid. The numbers on the right side are the mean probability for positivity (also shown as a cross mark) with 95% confidence interval in parentheses (shown as lines both sides of the cross).

Fig 6. Probability of positivity for all samples tested by RT-PCR days 17–110 of illness onset.

The numbers in parenthesis after each fluid type indicate the number of patients who provided samples for each body fluid. The numbers on the right side are the mean probability for positivity (also shown as a cross mark) with 95% confidence interval in parentheses (shown as lines both sides of the cross).