S65

Ebola Hemorrhagic Fever Outbreaks in Gabon, 1994–1997: Epidemiologic and

Health Control Issues
Alain-Jean Georges, Eric M. Leroy, André A. Renaut, Centre International de Recherches Médicales de Franceville,
Carol Tevi Benissan, René J. Nabias, Minh Trinh Ngoc, Franceville, Ministère de la Santé Publique, Faculté de Médecine,
Paul I. Obiang, J. P. M. Lepage,* Eric J. Bertherat,* Université Omar Bongo, Mission Française de Cooperation et d’Action
Culturelle, and Ministère de la Santé Publique, Libreville, Gabon
David D. Bénoni, E. Jean Wickings, Jacques P. Amblard,*
Joseph M. Lansoud-Soukate, J. M. Milleliri, Sylvain Baize,
and Marie-Claude Georges-Courbot*

From the end of 1994 to the beginning of 1995, 49 patients with hemorrhagic symptoms were
hospitalized in the Makokou General Hospital in northeastern Gabon. Yellow fever (YF) virus was
first diagnosed in serum by use of polymerase chain reaction followed by blotting, and a vaccination
campaign was immediately instituted. The epidemic, known as the fall 1994 epidemic, ended 6
weeks later. However, some aspects of this epidemic were atypical of YF infection, so a retrospective
check for other etiologic agents was undertaken. Ebola (EBO) virus was found to be present concomi-
tantly with YF virus in the epidemic. Two other epidemics (spring and fall 1996) occurred in the
same province. GP and L genes of EBO virus isolates from all three epidemics were partially
sequenced, which showed a difference of õ0.1% in the base pairs. Sequencing also showed that all
isolates were very similar to subtype Zaire EBO virus isolates from the Democratic Republic of the
Congo.

Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

Johnson et al. [1] isolated and identified Ebola (EBO) virus
from human cases during a 1976 epidemic of hemorrhagic
fever (HF) in the Democratic Republic of the Congo (DRC).
During the same year, EBO virus was isolated from patients
during an HF epidemic in Sudan [2]. The viruses, which were
closely related to Marburg virus (all members of the Filoviri-
dae), had an 88% and a 53% case fatality rate in DRC [3] and
Sudan [2], respectively. A third outbreak, with a case fatality
of 60%, occurred in Sudan in 1979 [4]. In addition, a death
was registered in Tandala, DRC, in 1979 [5]. In 1994, a new
strain of EBO virus was isolated from a Swiss researcher with
a dengue-like syndrome, who had likely been infected during
the necropsy of a chimpanzee (see Formenty et al., this supple-
ment). The animal had been found dead in the TaıF National
Forest (Côte d’Ivoire) during a 2-year epidemic that killed half
of the population of chimpanzees [6]. A third serious human
epidemic of EBO occurred in 1995 in Kikwit, DRC; it was
associated with a mortality rate similar to that seen during the
DRC and Sudan epidemics [7].
During 1994 and 1995, an outbreak of HF occurred in north-
eastern Gabon. It was first considered to be caused only by
yellow fever (YF) virus on the basis of the clinical symptom-
atology, routine biochemical tests, and specific laboratory re-
sults provided by the Centre National de Référence des Fièvres
Hemorragiques Virales (Institut Pasteur, Paris) [8, 9]. However,
retrospective serologic tests detected concomitant EBO virus
antibodies among some of the patients and the general popula-
tion [10, 11]. Later efforts to isolate the EBO virus from some
of the specimens from the first epidemic (1994) were success-
ful. One year later, in February and in July 1996, two more
HF outbreaks occurred in northeastern Gabon [10 – 13].
Herein, we report on three EBO epidemics that occurred
between late 1994 and early 1997 in northeastern Gabon [10].

The Epidemics

Informed consent was obtained from the patients or their parents or guard-
ians.
Financial support: CIRMF is supported by the Republic of Gabon, the French
Ministry of Foreign Affairs (Coopération et Action Culturelle), and ELF GA-
BON Co. Ltd. (Libreville).
Reprints or correspondence (current affiliation): Dr. Alain-Jean Georges,
Chefferie, Hopital d’Instruction des Armées Desgenettes, 108 Bd Pinel, 69
275 Lyon, Cedex 03, France (ajgeorges@wanadoo.fr).
* Current affiliations: CHA, Lamalou les Bains, France (J.P.M.L.); IMTSSA,
Le Pharo, Marseille Armées, France (E.J.B.); Ministère Affaires Etrangères,
Service de l’Action Humanitaire, Paris (J.P.A.); and CBMS, Institut Pasteur,
Paris (M-C.G.C.).
The Journal of Infectious Diseases 1999;179(Suppl 1):S65–75
q 1999 by the Infectious Diseases Society of America. All rights reserved.
0022–1899/99/79S1–0013$02.00
It is important to note that during this investigation, we
continually faced many difficulties (e.g., logistics problems and
cultural and political constraints) in the collection of data and
management of our research on this disease. Despite our efforts,
the difficulties sometimes led to the loss of important informa-
tion, and at times, forced us to report the scientific data in a
rather unorthodox manner.
First epidemic (fall 1994). The first epidemic in Gabon
had two waves of patients, with the first beginning in early
December 1994 and the second beginning at the end of January
to February 1995. All patients in the first wave came from 3
gold-panning encampments (Mékouka, Andock, and Minkébé)

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

S66 Georges et al.
situated in small clearings of 2000 – 3000 m2 at the edge of the
rain forest. Figure 1 and table 1 show the geographic location
of these and other villages with case-patients. Table 1 also
shows the chronology of illness for the cases. Three hundred
fifty people, mainly of the Bakota ethnic group but also some
Bakwélé, inhabit this area (30 in Mékouka, 20 in Andock, and
300 in Minkébé).
A total of 32 sick persons from the three forest encampments
(23 from Mékouka, 4 from Andock, and 5 from Minkébé)
traveled 100 km south by river to the nearest hospital at Mako-
kou, the main town of the region, for treatment. In addition to
the primary cases in Andock, we were also informed that deaths
had occurred in the local population of great apes (chimpan-
zees, Pan p. troglodytes, and gorillas, Gorilla g. gorilla). How-
ever, despite intensive searches of the forest by teams from
Centre International de Recherches Médicales de Franceville
(CIRMF), no cadavers or skeletons were ever found, even when
a patient told us that he had killed a chimpanzee with abnormal
behavior inside his encampment. This report could be anec-
dotal, but it cannot be totally ignored.
The second wave of patients did not originate from the en-
campments: They were what we should have called (in a survey
following accepted procedures) secondary (or tertiary ?) cases.
Unfortunately, the information that was available from the au-
thorities regarding these patients did not provide a precise gene-
alogy of all cases. The first patient in the second wave was
from Mayela, a small village close to Makokou, far from the
forest encampments. This person, who was probably the first
secondary or tertiary case, lived near a traditional healer (a
‘‘nganga’’) and was probably infected as a result of contact
with a hospitalized patient who, against medical advice, left
the hospital to seek care from the nganga. Sixteen more cases
(table 1) occurred in mid January: 1 case at Makokou Général
Hospital, 12 cases from Mayela, and 3 cases from two villages
(Ekataniabé, 1 patient; Ekobakoba, 2 patients) on the road
running south toward Franceville. None of these patients had
been in the area affected by the first wave of the epidemic
during the previous 3 months; however, all had been either in
direct contact with sick relatives (people hospitalized at Mako-
kou Général Hospital or sleeping in the nganga traditional heal-
er’s home) or with people caring for patients.
The last reported case (infected while caring for a relative
at Makokou’s hospital) occurred at Ekobakoba on 9 February
1995, and the epidemic was declared over by Gabonese health
authorities on 17 February 1995.
Overall, 49 persons were admitted to Makokou’s hospital
with suspected HF. Patients were suspected of having HF if
they had a well-identified contact with an ill person or at least
two of the following clinical signs: fever, diarrhea, vomiting,
melena, conjunctivitis, arthralgia or myalgia, diarrhea, or vom-
iting. Of the first 9 patients, 2 who we examined during the
first wave also had jaundice, which has never been described
as a symptom of EBO infection but is consistent with YF
infection. This case definition of HF, which was proposed by
/ 9d49$$se13 01-12-99 11:06:39 jinfa
JID 1999;179 (Suppl 1)
the physicians in charge of the Makokou hospital (a very poorly
equipped facility), was rather broad, and at least 4 of the 49
patients appeared unlikely to have HF. On 18 December 1994,
during the middle of the first wave, the Gabonese health author-
ities requested that we examine 9 patients in Makokou General
Hospital. Biologic samples from those patients led to the identi-
fication of the etiologic agents.
Between December 1994 and March 1995, CIRMF collected
22 samples in Mayela from contacts (all §15 years old) of
sick persons and 88 samples from the local population of
Ogooué-Ivindo Province, Gabon, where the epidemic occurred.
At that time, many people had fled the area in terror; therefore,
it was difficult to conduct an investigation. A year later (January
1996), we sampled 236 people from the three initially infected
villages and from villages in the Makokou area in order to
further assess exposure to EBO virus in the local population.
Second epidemic (spring 1996). A second epidemic began
during early February 1996 in the village of Mayibout 2, Ga-
bon, which is located on the Ivindo River. Mayibout 2 is 40
km south (6 h by boat) of Mékouka and Andock, where the
first epidemic broke out, and north of Makokou (7 h by boat).
Eighteen people who had skinned and chopped a chimpanzee
Downloaded from jid.oxfordjournals.org by guest on May 8, 2011
cadaver that they found became ill (fever, headache, bloody
diarrhea). They were evacuated from Mayibout 2 to Makokou
on the decision of the village chief, despite governmental in-
structions to the contrary. All patients were admitted to Mako-
kou Général Hospital, where 4 moribund patients died within
48 h.
The bodies of the 4 patients were returned by river to Mayi-
bout 2; a fifth patient, who was moribund when he escaped
from the hospital, died after returning to Mayibout 2. Tradi-
tional burial ceremonies were performed without any special
precautions to avoid disease transmission. Two other ‘‘primary
cases’’ occurred, which appeared not to be connected to the
chimpanzee episode; 1 died. Fifteen serum samples from these
initial 18 primary cases and 6 from secondary and tertiary cases
were obtained. An additional 205 serum samples were obtained
from the population of Mayibout 2 and neighboring villages
(Mayibout 1 and Mvadi, which are 2 km south and 20 km
north, respectively, of Mayibout 2).
Third epidemic (fall 1996). On 5 October 1996, we in-
formed the Gabonese health authorities (who 1 week earlier
had requested our assistance in the investigation) that we had
isolated EBO viruses from 2 of 6 samples from patients hospi-
talized at Booué. Personnel from CIRMF carried out a retro-
spective investigation of this third epidemic, which probably
started as early as 13 July with the death (undeclared) of a 39-
year-old hunter in a logging camp near Booué (0706* S, 11757*
E) between Ovan and Koumameyong, 200 km from Mékouka
and 120 km from Makokou to the southwest (figure 1). The
symptoms of this first case were suggestive of viral HF (VHF;
fever, bleeding, vomiting, diarrhea, and headache). At the be-
ginning of August 1996, information was obtained that several
chimpanzees may have died in the same area, and a field collab-
UC: J Infect
 Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

Figure 1. Geographic location of sites of primary and secondary cases of hemorrhagic fever. A, Gabon, B, Africa, C, sites.

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

S68 Georges et al. JID 1999;179 (Suppl 1)

Table 1. Geographic location (in order of occurrence of illness) of sites of primary and secondary
human Ebola infections during the fall 1994 outbreak in Gabon.

Samples from Samples from Samples from

sick patients patients’ contacts general population
Village Geographic location (first epidemic) (first epidemic) (1996)

Mékouka 1724* N 12759* E 23 — 12

Andock 1729* N 12755* E 4 — 8
Minkébé 1744* N 12749* E 5 — 86
Mayela (December 1994) 0738* N 12756* E — 22 12
Makokou 0733* N 12750* E 1 28 72
Mayela (February 1995) 0738* N 12756* E 13 32 24
Etakaniabé 0732* N 12757* E 1 12 12
Ekobakoba 0736* N 13705* E 2 1 10
Mayibout 2 1707* N 13706* E — — 175
Mayibout 1 1707* N 13706* E — — 10
Mvadi 1746* N 13709* E — — 20

orator was able to obtain skin samples from 1 cadaver for The epidemic was declared over in March 1997 (last offi-
pathologic examination at the Centers for Disease Control and cially declared patient: 18 January), with a total of 60 cases
Prevention (CDC, Atlanta), with which CIRMF was in contact. and 45 deaths. Unfortunately, the information made available
Six weeks later, at the end of August, a second hunter (35 to us from the local authorities did not allow for a very precise

Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

years old) died in the same Gabonese logging camp, but his
death, which was also suggestive of VHF, was not linked to
that of the first hunter. A third hunter (26 years old) became
ill 12 days later and was hospitalized at Booué. He subsequently
evaded the medical authorities at Booué and died in the village
of Balimba, where he was being treated by a nganga. Shortly
after the hunter’s death, both the nganga and his nephew fell
ill in mid-September and were admitted first to Booué Hospital
and then transferred to Makokou General Hospital, where they
died. Three other cases, including another nganga, occurred
among people living in the same area.
After we confirmed that the patients had EBO HF (EHF),
the authorities decided to apply the same clinical and epidemio-
logic definitions that were used during the outbreaks in Yam-
buku, DRC, in 1976 and in Kikwit to diagnose EHF patients
[1, 7]. The disease spread around Booué; 24 cases, including
17 deaths, were officially declared by 13 November 1996. A
Gabonese doctor, who had performed an endoscopy in a private
hospital in Libreville on an EBO-infected patient from Booué,
developed signs and symptoms of EHF. On 27 October, without
first requesting a serologic diagnosis of VHF and without con-
sidering the possible etiology of such a diagnosis, he went to
Johannesburg for treatment. A South African nurse who cared
for the doctor became ill on 2 November and died on 24
November (Swanepoel R, personal communication).
At the end of November 1996, a second wave of this third
epidemic appeared in three locations around Booué: Lolo (3
deaths, 6 cases), SHM, a timber company (4 deaths, 5 cases),
and the logging camp of Balimba (1 death, 1 case). A confirmed
case spread to Lastourville (130 km southeast of Booué) from
Balimba, and again to Libreville, with subsequent transmission
of the disease to the capital (11 deaths, 15 cases).
genealogy of all cases; however, we did obtain accurate data
on 47 of the 60 cases (Milleliri JM, unpublished data).
Materials and Methods
Routine Clinical Chemistry Tests
Routine clinical chemistry tests were done on sera from some
hospitalized cases in each epidemic.
Serologic Assays
IFAs. IFAs were used to detect viruses causing HF (including
Crimean-Congo HF fever virus, Rift Valley fever virus, subtype
Zaire of EBO virus [EBO-Z; Mayinga and Boniface strains], Lassa
fever virus, and Marburg virus) in sera from some patients in each
epidemic. The IFAs were done using CRELM (initials of each
virus) microscope slides [14] provided by the CDC and US Army
Medical Research Institute of Infectious Diseases (Fort Detrick,
Frederick, MD).
IgM anti-EBO EIA. ELISAs were also used to detect EBO
virus infection [15]. For IgM detection, an immunocapture test
was used. Polyvinyl chloride 96-well microtiter plates (no. 1000-3,
Bioblok; Dynatech, Strasbourg, France) were adsorbed overnight
at 47C, using goat anti-human m-chain (Tago, Burlingame, CA)
diluted 1:500 in 0.01 M PBS (pH 7.4). All steps of the assay were
done using volumes of 100 mL/well. Sera from persons who were
infected or were suspected of being infected and sera from their
contacts were added to the wells at a 1:100 dilution in serdil (5%
skim milk [Bacto milk], PBS, 0.1% Tween 20). After the plates
were washed with PBS with 0.1% Tween 20 (Bio-Rad, Richmond,
CA), sera were added (first at a 1:100 dilution and then at 4- to
800-fold dilutions), and the plates were incubated for 60 min at
377C in a humid chamber. For each serum and each dilution, in

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

JID 1999;179 (Suppl 1) Ebola Outbreaks in Gabon, 1994–1997
one well, we used as antigen a cell slurry made in EBO-infected
Vero E6 cells diluted 1:1000 in PBS; in another well, we used a
slurry of uninfected Vero E6 cells. Anti-EBO (subtypes Z, Sudan
[S], and Reston [R]) hyperimmune rabbit serum was diluted 1:2000
in serdil. Plates were incubated for 1 h with anti-rabbit IgG (no.
074-1506; Kirkegaard & Perry, Gaithersburg, MD) diluted
1:10,000 in serdil (as conjugate) and then incubated for 30 min
using an H2O20 ABTS substrate (Kirkegaard & Perry).
In the first assays we used 2 standard serum samples as positive
controls. As soon as we had confirmed the EBO cases, we added
our own 2 positive controls, which were run in standard dilution
series for providing standard curves, which allowed the determina-
tion of the limits of detection of the assay. A panel of 4 negative
sera (from healthy persons) were run in all tests to determine the
background of the assay (the limit at which the positive controls
are positive) and to provide the mean and standard deviation (SD)
of that background. A sample was considered to be positive if its
optical density exceeded the mean / 3 SD of the controls. Plates
were read spectrophotometrically at an optical density of 410 nm
on a microcomputer (LP 200; Sanofi Diagnostics Pasteur, Marnes-
la-Coquette, France).
IgG anti-EBO EIA. For IgG detection, half of a 96-well mi-
crotitration plate was coated with a lysate of EBO-Z–, EBO-S–,
and EBO-R–infected Vero E6 cells, and the other half was coated
with a lysate of uninfected Vero E6 cells to determine the specific
binding of antibody to virus antigens (the background); both ly-
sates were diluted 1:1000 in PBS. Plates were incubated overnight
at 47C. The plates were washed three times with PBS plus 0.1%
(vol/vol) Tween 20, and then sera were added at a 1:100 and 4-
to 800-fold dilutions in serdil and allowed to react with the antigen-
coated wells for 60 min at 377C in a humid chamber. Bound IgG
was detected with mouse anti-human IgG conjugated to horserad-
ish peroxidase. Further identification was done as in the IgM EIA
assay. A panel of specific IgG-positive controls and specific IgG-
negative sera (from healthy persons) was run in all tests in order
to determine the cutoff value. For each assay, the mean and SD
of the adjusted optical density were accumulated and used to calcu-
late a value equal to the mean / 3 SD, which represented the
cutoff value. Plates were read as above.
Anti-YF IgM EIA. At the beginning of the first VHF epidemic,
YF virus was strongly thought to be the probable cause of illness,
on the basis of the clinical and epidemiologic results of the 9
patients examined. Therefore, we tested for YF virus infection,
using IgM capture techniques [17].
Virus Isolation Attempts
Vero E6 cell 6-well plates were inoculated with sera from pa-
tients suspected of harboring a VHF virus. During the first epi-
demic, the biosafety level 4 facilities of the Institut Pasteur were
used for all virologic manipulations of patient-derived material;
further work on the two last epidemics was carried out at CIRMF’s
site or at CDC’s Special Pathogens laboratories or at both sites
for part of the molecular biology testing of EBO virus strains [16].
To identify the EBO virus, EBO reference antisera were tested
against the virus isolates cultured on Vero E6 cells. Tests were
done using polyclonal antibodies raised in mice (IFA test) and a
pool of monoclonal antibodies (antigen-capture test) that were
/ 9d49$$se13 01-12-99 11:06:39 jinfa
S69
known from previous experience to identify EBO antigen for dif-
ferent strains of virus in tissues. Molecular identification of the
EBO virus strains following extraction of RNA from whole blood
clots or from the peripheral blood mononuclear cells separated on
ficoll have been described elsewhere [16].
At Institut Pasteur, attempts to isolate YF virus by inoculation
of suckling mice, mosquitoes, and mosquito cell cultures (AP 61,
C6–36) failed to give positive results. Therefore, RNA was ex-
tracted according to the method of Chomczynski and Sacchi [18],
and the presence of the YF viral genomic RNA was revealed by
reverse transcriptase–polymerase chain reaction (RT-PCR) ampli-
fication [19] using consensus primers for flaviviruses. The ampli-
fied DNA fragment was then transferred onto a nitrocellulose
membrane and hybridized with a YF virus–specific probe.
Antigenemia
The sera of sick people, their contacts, and people suspected of
having EHF were tested by EIA for the presence of EBO antigen,
as were the supernatants of Vero E6 cells seeded with sera from
these persons. Half of a 96-well microtitration plate was coated
(overnight at 47C) with a mouse monoclonal antibody recognizing
EBO virus epitopes; the other half was coated with normal my-
eloma ascites diluted 1:1000 in PBS. Biologic samples to be
Downloaded from jid.oxfordjournals.org by guest on May 8, 2011
checked for the presence of EBO antigen were then added (first
at a dilution of 1:4 and then 4- to 64-fold). After a 1-h incubation
at 377C in a humid chamber, the plates were washed 3 times, and
rabbit anti-EBO (subtypes Z, S, and R) diluted 1:2000 in serdil
was added and incubated for 1 h at 377C in a humid chamber.
Plates were incubated for 1 h with anti-rabbit IgG conjugate (no.
074-1506; Kirkegaard & Perry) diluted 1:10,000 in serdil and then
incubated for 30 min using an H2O20 ABTS substrate (Kirke-
gaard & Perry), and bound rabbit IgG was detected.
Results
Serologic surveys. The occurrence of HF cases in the fall
of 1994 initially led us to suspect an outbreak of sylvatic YF
infection. However, we were aware that another HF agent (in
particular EBO virus) could be circulating, given the possible
fatalities observed in nonhuman primates (apes replicate YF
virus, but they usually do not die from it; Monath TP, personal
communication).
In the first instance, we screened 9 patients (all from Mako-
kou General Hospital) for HF virus. EBO IFA on CRELM
slides was negative for all 9 serum samples, whereas 6 of 9
patients showed evidence of recent YF virus infection (IgM
capture). Of the 49 sick patients, another 14 cases who had
never been vaccinated against YF were also screened. Of the
14 cases, 8 had evidence of recent YF infection as determined
by IgM capture or Southern blot, but none had evidence of
antibodies to the main African viral HFs as determined by use
of IFA on CRELM slides. However, the antigen used in our
IFA test was several years old, so in June and July 1995 we
used ELISAs and fresh antigens to retrospectively test these
same 23 samples (ELISA is generally more sensitive than IFA).
UC: J Infect
S70 Georges et al.
Nineteen patients had signs of a recent infection with EBO as
determined on the basis of at least one of the following: positive
for EBO IgM, positive for antigenemia, positive for virus iso-
lated from cultured Vero E6 cells.
We also used ELISA to test the 88 serum samples obtained
in January 1995 (before YF vaccination) from the population
living around the three encampments (Andock, Minkébé, and
Mékouka) where the first epidemic broke out and the 22 serum
samples obtained from contacts in Mayela. Two of these sera
(1.8%) were positive for anti-EBO IgG but not EBO IgM,
while 9 YF-unvaccinated people (8.1%) were positive by YF
IgM capture, indicating a recent YF virus circulation.
One year later, we obtained samples from 236 persons from
the same area (table 1). For various reasons, only 56 of the
original 110 people were part of this second cohort (e.g., use
of given name in the place of family name in original survey,
fear of being identified and accused of causing the original
epidemic, and migrations of populations). Of 236 subjects, 24
(9.7%) had anti-EBO IgG by ELISA without any anti-EBO
IgM (1 of these IgG-positive cases was the first case of the
first wave in the 1994 epidemic and had survived a typical
EBO-type illness).
During the second epidemic, 205 samples were collected
from people living in 3 villages (Mayibout 1 and 2, where
EBO was known to have occurred, and Mvadi, where 1 uncon-
firmed death of a hunter was said to have occurred). High
frequencies of anti-EBO IgG (14.9% – 30%) and IgM (5.7% –
10.0%) by ELISA in all three villages (table 2) may suggest
the presence of an asymptomatic or mild form of the disease
concomitant with the classical severe form of EBO, or they
may demonstrate the limits, in terms of specificity, of the EBO
IgG and IgM ELISAs that we used.
During the initial phases of the third epidemic, we were
unable to carry out serologic surveys due to local reasons that
were out of our control.
Fatality and descriptive epidemiology. We reviewed files
at Makokou General Hospital for persons who were seen during
the first epidemic in fall 1994; 49 patients were identified with
at least two signs of VHF, 29 of whom died (case fatality rate,
59%). There were 26 male and 23 female patients (sex ratio,
1.1) with a mean age of 37 years.
Table 2. Anti-Ebola antibodies among the general population of
three villages in Ogooué-Ivindo Province, Gabon, during the second
epidemic (spring 1996), as determined by ELISA.
No. (%) of villages
positive for
Geographic No. of
Village location samples IgG IgM
Mayibout 1 1707* N 13706* E 10 3 (30.0) 1 (10.0)
Mayibout 2 1707* N 13706* E 175 26 (14.9) 10 (5.7)
Mvadi 1746* N 13709* E 20 5 (22.5) 2 (10.0)
/ 9d49$$se13 01-12-99 11:06:39 jinfa
JID 1999;179 (Suppl 1)
For the spring 1996 epidemic, the case fatality rate was
67.7% (21 of 31 cases died) as determined using the accepted
(Yambuku and Kikwit) case definition. There were 17 male
and 14 female patients (sex ratio, 1.2) with a mean age of 27.6
years.
In the fall 1996 epidemic, there were 60 cases and 45 deaths
(case fatality rate, 75%). The sex ratio was 2.4. We were unable
to conduct adequate surveys during the epidemic, so complete
age-range data are unavailable.
The development (geographically and chronologically) of the
first two epidemics is shown in table 1 and figures 2 and 3,
respectively. When tracing the computerized (MS Access software;
Microsoft, Redmond, WA) polynomial regression curves obtained
from histograms, the first epidemic shows two waves of morbidity,
whereas in the second outbreak, there is only one peak of morbid-
ity. These observations are related to the notably different types
of spread in the two epidemics. Persons in several villages were
infected during the first epidemic, which lasted longer than the
second, in which only persons in a confined area were affected.
The exact origin(s) of the virus in the first epidemic is uncertain,
and despite reports of the death of great apes, we were unable to
confirm these reports.
Downloaded from jid.oxfordjournals.org by guest on May 8, 2011
Exploitation of the area by gold miners has caused substan-
tial disturbance to the forest canopy. This disturbance may
have caused contact between other species and humans that
otherwise would not have occurred. None of the specimens
(small mammals and insects) collected in the field so far have
yielded any sign of harboring the virus; however, these studies
are ongoing.
In the second outbreak, there is no doubt that the virus came
from one source, a chimpanzee infected by EBO (for which
the chain of infection is unknown). The chimpanzee seems to
have been the index case for infecting 18 primary human cases.
There were 2 other primary cases, for whom we have no infor-
mation on the source of infection (the role of a nganga was
suspected), making a total of 20 primary cases for this epi-
demic. Only those people who had been in contact with the
dead chimpanzee before it was cooked for eating were affected;
nobody was infected by eating the cooked meat. We identified
10 secondary cases involving close contacts of dead or dying
people and 1 tertiary case. It is important to note that no cases
of infection occurred among the professional medical personnel
because of preventive measures that were set in place on the
second day of the epidemic. All equipment and material neces-
sary for barrier nursing and prevention of the spread of disease
were provided. One hundred ninety-one contacts were traced,
with no evidence of infection by EBO virus.
Similar to the case in the first epidemic, the source(s) of the
virus in the third epidemic was not immediately obvious, nor
could it be traced retrospectively. The lack of identification of
cases at the outset and the lack of adequate containment mea-
sures once the diagnosis had been made meant that sick persons
could travel throughout and even outside Gabon. Some contact
cases could not be identified and isolated.
UC: J Infect
JID 1999;179 (Suppl 1) Ebola Outbreaks in Gabon, 1994–1997 S71

Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

Figure 2. Evolution of 1994 – 1995 Ebola virus epidemic in Gabon. A, Evolution by mortality rate. Curve shows biphasic evolution of
epidemic limited to large no. of secondary cases who have spread virus. B, Geographic evolution of epidemic. Hop Å hospital.

Immunostaining [15] done at the CDC on skin biopsies from
a chimpanzee found dead in the forest indicated the presence
of EBO [16]; however, the ape is very unlikely to have had
any contact with humans.
Biologic and clinical features. Clinical signs and symp-
toms observed in 15 serologically or virologically confirmed
patients from the outbreak at Mayibout 2 (spring 1996) are
presented in table 3. Fever was always present, while in Ç75%
of the cases, diarrhea (a dense suspension of black or brown-
red uneven microaggregates õ5 mm in size in liquid or syrupy
stools) and vomiting were seen, and Ç50% had conjunctivitis
or conjunctival bleeding. Taking only the second epidemic
into consideration, since it was the best-documented outbreak,
clinical signs and symptoms were observed between days 6
and 11 after handling the chimpanzee, and deaths occurred
between days 12 and 18 after the appearance of signs or symp-
toms. Despite a small number of observations, it seemed to us
that those who died had a shorter incubation period than those
who survived.
Table 4 presents the biologic and biochemical data for 13
patients infected during the spring 1996 outbreak. Of 13 pa-
tients in the cohort, 6 died. Data were obtained 4 – 7 days after
the onset of disease (16 days after the patients were infected
by a chimpanzee cadaver). The levels of direct and conjugated
bilirubin were normal, but all patients had signs of liver
involvement (i.e., elevations in g-glutamyltransferase, amino-
transferase, and alkaline phosphatase), and 23% had signs of
slightly decreased renal dysfunction (elevated serum creatinine
and urea).
Most of the other blood parameters (e.g., total serum pro-
teins, albumin, uric acid, cholesterol, triglycerides, sodium, po-
tassium, and chloride) that we were able to test were normal.
Characterization of virus strains. EBO viruses were iso-
lated from patients from each of the three epidemics. All the
Gabonese virus strains sequenced showed a high level of ho-
mology with the group of strains from DRC [16]. Viruses
isolated from patients during the same outbreak had identical
base-pair sequences, but base-pair sequences were slightly dif-
ferent for isolates from the individual outbreaks. There was
only a four-nucleotide (nt) difference in the GP gene in each
of the 3 Gabonese strains (i.e., the GP gene from the December
1994 strain differs from that of the February 1996 strain by

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

S72 Georges et al. JID 1999;179 (Suppl 1)

Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

Figure 3. Evolution of spring 1996 Ebola virus epidemic in Gabon. A, Evolution of epidemic by mortality rate. Curve shows monophasic
evolution of epidemic limited to small no. of secondary cases who spread virus. B, Geographic evolution of epidemic. Not included is case
who died on 19 February 1996 and who had different infection not related to Ebola virus since 20 January 1996. Hop Å hospital.

four nt, and the February strain differs from the October 1996
strain by 4 nt).
A conserved region of the L gene was also amplified in virus
strains from fall 1994 and fall 1996, using degenerate primers
selected from sequences of EBO-S and Marburg virus strains;
PCR products were subsequently sequenced [13]. The 2 Gabo-
nese strains showed high homology in the first 156-bp se-
quence, with only a single base change (position 110; A to G)
separating the 2 strains. As stated, during the first epidemic, 6
of 9 patients had anti-YF IgM. At the Institut Pasteur, using
PCR and specific hybridization techniques, YF sequences were
identified in sera from 2 of the 6 patients who were anti-YF
IgM positive and in 1 of the 3 patients who had no anti-
YF IgM. Those 3 patients, who were negative for all EBO
investigations, had clinical signs compatible with YF (e.g.,
severe low back pain, fever, jaundice [2/3 patients], melena,
and melanemesis), and biochemical results showed signs of
liver and renal failure and slightly raised bilirubin levels.
Discussion
An epidemic of HF among humans, some of whom had
clinical and biologic symptoms suggestive of YF infection,
started in December 1994 in Ogooué-Ivindo Province. On the
basis of clinical, biologic, and serologic data, a diagnosis of
YF infection was suspected in 6 hospitalized cases, and a diag-
nosis of YF infection was confirmed for 3 other cases on the
basis of molecular biology data, although the complete virus
could not be isolated, which is not unusual for YF virus. (YF
virus is known to be impossible or difficult to isolate from
clinical material if samples are taken several days after evolu-
tion of the clinical disease; for example, during the 1970 epi-
demic in Okwooga and the 1986 epidemic in Oju region, Nige-
ria, it was impossible and difficult, respectively, to isolate YF
virus [Monath TP, personal communication]). At the time, no
etiologic diagnosis was scientifically established for the 6 pa-
tients who were not confirmed to have YF virus infection;

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

JID 1999;179 (Suppl 1) Ebola Outbreaks in Gabon, 1994–1997 S73

Table 3. Clinical manifestations observed in 15 confirmed cases of of infected case-patients, with barrier nursing, disinfection,
Ebola hemorrhagic fever during the spring 1996 outbreak in Mayibout and insect eradication programs and (2) active reintroduction
1 and Mayibout 2, Gabon.
of a vaccination program against YF virus, which had lapsed
Clinical symptoms % of patients with symptom ú10 years earlier. The fact that the entire YF virus was not
isolated is not a conclusive argument against the fact that YF
Fever 100 occurred during the first VHF epidemic, since samples may
Diarrhea 87 have been collected at a late stage of the disease when the
Vomiting 73
virus is difficult to isolate. However, in fall 1994, there was a
Conjunctivitis 53
Gingival bleeding 40 high prevalence of anti-YF IgM in YF-unvaccinated subjects,
Arthralgia, myalgia 33 which was strongly suggestive of a recent YF outbreak. It is
Cephalgia 27 known that sylvatic YF infection can exist in a mild or symp-
Asthenia 27 tom-free form [17].
Cephalgia 27
A final argument for the presence of YF virus during the
Melena 27
Hiccups 27 first epidemic (fall 1994) comes from Pisano et al. [20], who
Obnubilation 27 identified YF infection in a cohort of 5 patients (different from
Sore throat 20 ours) from the same area by sequencing a cDNA fragment
Cough 20 obtained by RT-PCR and showing 94.4% homology with the
Melanemesis 13
Asibi strain of YF virus [21].
Epistaxis 13
Abdominal pain 7
During the first epidemic, EBO and YF viruses were obvi-
ously coexisting: 9 patients had EBO alone, 11 had EBO and
YF, and 3 had YF alone. Due to the small size of each group,
it is impossible to speculate on the relative lethality of each

Downloaded from jid.oxfordjournals.org by guest on May 8, 2011

nevertheless, we felt, from a public health point of view, that
there was a risk for reemergence of YF (the last epidemic in
Gabon was in 1949 [8]) among the unvaccinated population,
as confirmed later by the epidemiologic investigations of a
World Health Organization expert [9].
Without waiting for a definitive diagnosis, for which inves-
tigations were in progress at the hemorrhagic fever unit at
the Institut Pasteur, and despite the fact that another etiologic
agent could be involved, CIRMF suggested to the Gabonese
health authorities that they rapidly instigate measures to limit
the spread of YF infection. We suggested (1) passive isolation
virus alone or in combination, but a nosocomial amplification
of EBO appears to have occurred inside Makokou Général
Hospital. Such an association of YF and EBO viruses during
the same epidemic has been suspected in the past: In Ethiopia,
where a VHF outbreak occurred in 1961 – 1962, a later sero-
logic retrospective study showed that both EBO and YF viruses
may have been present [22].
Earlier IFA-based studies looking for the presence of EBO
virus in other provinces of Gabon were done by Ivanoff et al.
[23] in 1982 and Meunier et al. in 1986 [24]. Data obtained
by use of the same technique showed that EBO is widespread

Table 4. Biochemical results for 13 patients with Ebola hemorrhagic fever during the spring 1996 outbreak in Gabon.

Conjugated Aspartate Alanine Alkaline

Patient No. of days Total bilirubin bilirubin g-glutyltransferase aminotransferase aminotransferase phosphatase Urea
no. after onset Outcome (õ20 mmol/L) (õ10 mmol/L) (õ45 U/L) (õ45 U/L) (õ45 U/L) (õ125 U/L) (2.5 – 7 mmol)

1 7 R 52.0 35.0 246.0 230.0 666.0 419.0 11.0

2 4 D 42.0 30.0 212.0 220.0 512.0 410.0 10.9
3 4 R 17.0 4.7 16.0 155.0 60.0 126.0 3.1
4 4 D 25.0 6.7 52.0 160.0 75.0 166.0 3.1
5 6 D 16.5 4.4 64.0 314.0 109.0 150.0 3.2
6 4 R 18.0 7.2 38.0 226.0 98.0 72.0 6.7
7 4 D 21.0 6.7 21.0 165.0 70.0 136.0 3.2
8 4 R 31.7 19.8 193.0 185.0 1499.0 637.0 17.9
9 5 R 17.9 7.0 30.0 219.0 88.0 69.0 6.5
10 4 D 14.8 3.5 21.0 135.0 74.0 59.0 2.7
11 4 R 15.0 3.7 26.0 140.0 76.0 72.0 5.2
12 4 R 14.2 9.9 15.0 266.0 50.0 82.0 1.3
13 4 D 19.8 9.0 23.0 56.0 22.0 66.0 3.1
Mean 23.5 11.4 73.6 190.1 261.5 189.5 6.0
SD 11.7 10.3 83.7 65.9 420.5 181.8 4.7

NOTE. Data in parentheses are normal values. R Å recovered, D Å died.

/ 9d49$$se13 01-12-99 11:06:39 jinfa UC: J Infect

S74 Georges et al.
throughout equatorial Africa [25 – 27]. However, these results
must be examined carefully since others have suggested that
IFA is likely to give false-positive results [28]. In addition, in
our hands, this test was insensitive for EBO virus, specifically
during the fall 1994 epidemic, in which all sera collected during
the acute phase of the disease tested negative by IFA, while
some were positive by EIA. Even though ELISA techniques
appear to be very sensitive for assessing EBO infection, we
must be cautious about their specificity.
No serologic data were available from persons in Ogooué-
Ivindo Province before the first epidemic. However, we had
kept 58 untested frozen serum samples from the epidemic pe-
riod. The samples were from asymptomatic adults who were
not associated with gold-panning; 2 of them had ELISA EBO
IgM, suggesting recent EBO infection, and the remaining 56
samples were negative for anti-EBO IgG and IgM. This low
percentage of seropositivity suggests that the circulation of
EBO is a new phenomenon in the area, affecting mainly people
living deep in the forest in temporary encampments.
There are interesting demographic and ecologic differences
between the sites of the first two epidemics. The gold panners
in the fall 1994 outbreak were a transient population and caused
considerable damage to the local forest in which they were
working. Most of those who survived left the area after the
epidemic; a very small population remains. Mayibout 2, a per-
manent, relatively dense settlement whose inhabitants practice
slash-and-burn cultivation, is situated on the river Ivindo and
is surrounded by forest. The disturbance of the forest integrity
is one common feature between the two sites, but the one
important difference is the permanence and the size of the
population around Mayibout 2, which might permit the virus,
once present in the area, to circulate within the human popula-
tion. Both the reservoir and the vector for the EBO virus are
still unknown; however, it is apparent that the great apes, or
at least chimpanzees, are not involved because they are as
susceptible as humans to the disease.
On the basis of a sizeable but incomplete genetic compari-
son, the viruses from our three outbreaks are essentially indis-
tinguishable from the viruses isolated in DRC in 1976 and
1995; however, the viruses isolated in 1976 from Sudan and
DRC were clearly biologically distinct from one another [29].
We cannot rule out the possibility that varied strains of EBO
virus or even different members of the Filoviridae are circulat-
ing in these areas, with some being more virulent than others
yet cross-reacting and, therefore, appearing as asymptomatic
infections in the population. The inability to isolate viruses
from the Gabonese environment and from anyone but ill pa-
tients has not allowed us to address this issue. The fact that
the ELISA test has not been evaluated for specificity against
known human infections also does not permit us to eliminate
this as a possible explanation for the high level of antibody in
the population of Mayibout 2.
Finally, it is very clear that the use of simple barrier nursing
methods in even the most basic of hospital settings is sufficient
/ 9d49$$se13 01-12-99 11:06:39 jinfa
JID 1999;179 (Suppl 1)
to limit the spread of this disease. Nosocomial spread of infection
due to the lack of recognition of the disease and the dearth of
training and material available to establish simple barrier nursing
remains the root cause of most of the disease observed to date.
From the public health point of view, this remains a basic goal for
the prevention of EHF epidemics. Preventing primary infection
certainly requires the identification of risk factors and the reser-
voir, something that continues to elude us all.
Acknowledgments
We thank B. Le Guenno (Institut Pasteur, Paris) for his invalu-
able scientific assistance in affirming the yellow fever diagnosis
in January 1995 and for initiating new retrospective serologic stud-
ies on Ebola virus in June 1995 and new attempts at virus isolation
in samples from the fall 1994 epidemic, which led to the isolation
of the first Ebola isolate from Gabon; P. Tshipamba (CIRMF) for
providing excellent technical assistance; J. B. McCormick (Institut
Pasteur), S. Fisher-Hoch–McCormick (Fondation Mérieux, Lyon,
France), and P. Rollin (CDC, Atlanta) for valuable and pertinent
suggestions regarding the manuscript; and members of the Special
Pathogens Branch of the CDC as well as members of Walter Reid
Army Institute of Research, US Army Medical Research Institute
Downloaded from jid.oxfordjournals.org by guest on May 8, 2011
of Infectious Diseases (Fort Detrick, Frederick, MD), who have
generously provided reagents to CIRMF since 1982.
References
1. Johnson K, Webb P, Lange J, Murphy F. Isolation and partial characterisa-
tion of a new virus causing acute haemorrhagic fever in Zaire. Lancet
1977; 1:569 – 71.
2. WHO/International Study Team. Ebola hemorrhagic fever in Sudan, 1976.
Bull World Health Organ 1978; 56:271 – 93.
3. WHO/International Study Team. Ebola hemorrhagic fever in Zaire, 1976.
Bull World Health Organ 1978; 56:247 – 70.
4. Baron RC, McCormick JB, Zubeir OA. Ebola virus disease in southern
Sudan: hospital dissemination in intrafamilial spread. Bull World Health
Organ 1983; 61:997 – 1003.
5. Heymann D, Weisfeld, Webb P, Johnson K, Cairns T, Berquist H. Ebola
hemorrhagic fever: Tandala, Zaire, 1977 – 1978. J Infect Dis 1980; 142:
372 – 6.
6. Le Guenno B, Formenty P, Wyers M, Gounon P, Walker F, Boesch C.
Isolation and partial characterisation of a new strain of Ebola virus.
Lancet 1995; 345:1271 – 4.
7. WHO/International Study Team. Ebola hemorrhagic fever in Zaire (Kik-
wit). Bull World Health Organ 1995; 19:13.
8. World Health Organization. Yellow fever in Gabon. Bull World Health
Organ 1995:64.
9. Cordelier R. Fièvre jaune au Gabon: enquête sur un épisode amaril dans
la Province d’Ogooué-Ivondo (2 au 23 mars 1995). Rapport OMS.
Brazzaville: OMS, Bureau Régional pour l’Afrique, 1995.
10. Georges AJ, Le Guenno B, Renaut AA, et al. Recent Ebola virus outbreaks
in Gabon from 1994 to 1996: epidemiologic and control issues. Report
of International Colloquium on Ebola virus research (Antwerp, Bel-
gium). Antwerp: Institute of Tropical Medecine, 1996:47.
11. Amblard J, Obiang P, Edzang S, Prehaud C, Bouloy M, Le Guenno B.
Identification of the Ebola virus in Gabon in 1994. Lancet 1997; 1:
181 – 2.
12. World Health Organization. Ebola hemorrhagic fever in Gabon. Bull
World Health Organ, 1996; 9:71.
UC: J Infect
JID 1999;179 (Suppl 1) Ebola Outbreaks in Gabon, 1994–1997
13. Georges-Courbot MC, Lu CY, Lansoud-Soukate J, Leroy E, Baize S.
Isolation and partial molecular characterisation of a strain of Ebola virus
during a recent epidemic of viral haemorrhagic fever in Gabon [letter].
Lancet 1997; 349:181.
14. Wulf H, Lange JV. Indirect immunofluorescence for the diagnosis of Lassa
fever infection. Bull World Health Organ 1975; 52:429 – 36.
15. Ksiazek T, Rollin P, Jarhling P, Johnson E, Dalgard D, Peters JC. Enzyme
immunosorbent assay for Ebola virus antigens in tissues of infected
primates. J Clin Microbiol 1992; 30:947 – 50.
16. Georges-Courbot MC, Sanchez A, Lu CY, et al. Isolation and partial
characterization of Ebola viruses causing different outbreaks in Gabon.
Emerg Infect Dis 1997; 3:59 – 62.
17. Monath TP. Yellow fever. In: Monath JP, ed. The arboviruses: epidemiol-
ogy and ecology. Vol. V. Boca Raton, FL: CRC Press, 1989:139 – 231.
18. Chomczynski P, Sacchi N. Single step method of RNA isolation by acid
guanidium thiocyanate – phenol-chloroform extraction. Anal Biochem
1987; 162:156 – 9.
19. Pierre V, Drouet MT, Deubel V. Identification of mosquito-borne flavivirus
sequences using universal primers and reverse transcription/polymerase
chain reaction. Res Virol 1994; 145:93 – 104.
20. Pisano MR, Durand JP, Tolou H. Partial genomic sequence determination
of yellow fever virus strain associated with recent epidemic in Gabon.
Acta Virol 1996; 40:103 – 5.
21. Hahn CS, Dalrymple JM, Strauss JH, Rice CM. Comparison of the virulent
Asibi strain of yellow fever virus with the 17D vaccine strain derived
from it. Proc Natl Acad Sci USA 1987; 84:2019 – 23.
/ 9d49$$se13 01-12-99 11:06:39 jinfa
S75
22. Tignor GH, Casals J, Shope RE. The yellow fever epidemic in Ethiopia,
1961 – 1962: retrospective serological evidence for concomitant
Ebola or Ebola-like virus infection. Trans R Soc Trop Med Hyg
1993; 87:162.
23. Ivanoff B, Duquesnoy P, Languillat G, et al. Hemorrhagic fever in Gabon.
I. Incidence of Lassa, Ebola and Marburg viruses in Haut-Ogooué. Trans
R Soc Trop Med Hyg 1982; 76:719 – 20.
24. Meunier DMY, Dupont A, Madelon MC, Gonzalez JP, Ivanoff P. Surveil-
lance sérologique des fièvres hémorragiques virales dans le Haut-
Ogooué (Gabon). Ann Inst Pasteur/Virol, 1987; 138:229 – 35.
25. Josse R, Dupont A, Delaporte E, et al. Sero-épidemiologie des affections
virales à haut risque dans la ville de Port Gentil (Republique du Gabon).
Bull Liais document, OCEAC, 1987; 82:63 – 7.
26. Johnson E, Gonzalez JP, Georges AJ. Filovirus activity among selected
ethnic groups inhabiting the tropical forest of equatorial Africa. Trans
R Soc Trop Med Hyg 1993; 87:536 – 8.
27. Gonzalez JP, Josse R, Johnson ED, et al. Antibody prevalence against
hemorrhagic fever viruses in randomized representative Central African
populations. Res Virol 1989; 140:319 – 31.
28. Muhlberger E, Sanchez A, Randolf A, et al. The nucleotide sequence of
the L gene of Marburg virus, a filovirus: homologies with paramyxovi-
ruses and rhabdoviruses. Virology 1992; 187:534 – 47.
29. McCormick J, Bauer S, Elliott L, Webb P, Johnson K. Biologic differences
between strains of Ebola virus from Zaire and Sudan. J Infect Dis 1983;
147:264 – 7.
Downloaded from jid.oxfordjournals.org by guest on May 8, 2011
UC: J Infect