Content uploaded by Oyewale Tomori
Author content
All content in this area was uploaded by Oyewale Tomori on Mar 05, 2016
Content may be subject to copyright.
S76
The Reemergence of Ebola Hemorrhagic Fever, Democratic Republic of the
Congo, 1995
Ali S. Khan, F. Kweteminga Tshioko, David L. Heymann,
Special Pathogens Branch and Infectious Disease Pathology Activity,
Division of Viral and Rickettsial Diseases, National Center for
Bernard Le Guenno, Pierre Nabeth, Barbara Kerstie
¨
ns,
Infectious Diseases, and Epidemiology Program Office, Centers for
Yon Fleerackers, Peter H. Kilmarx, Guenael R. Rodier,
Disease Control and Prevention, Atlanta, Georgia; Ministry of Health,
Okumi Nkuku, Pierre E. Rollin, Anthony Sanchez,
Kinshasa, Democratic Republic of the Congo; World Health
Sherif R. Zaki, Robert Swanepoel, Oyewale Tomori,
Organization, Geneva, Switzerland; Institut Pasteur, and Epicentre,
Stuart T. Nichol, C. J. Peters, J. J. Muyembe-Tamfum,
Paris, France; Me
´
decins sans Frontie
`
res, Brussels, and Institute of
Tropical Medicine, Antwerp, Belgium; Special Pathogens Unit, National
and Thomas G. Ksiazek, for the Commission de Lutte
Institute for Virology, Department of Health, Sandringham, South Africa
contre les Epide
´
mies a
`
Kikwit
In May 1995, an international team characterized and contained an outbreak of Ebola hemorrhagic
fever (EHF) in Kikwit, Democratic Republic of the Congo. Active surveillance was instituted using
several methods, including house-to-house search, review of hospital and dispensary logs, interview
of health care personnel, retrospective contact tracing, and direct follow-up of suspect cases. In the
field, a clinical case was defined as fever and hemorrhagic signs, fever plus contact with a case-
patient, or fever plus at least 3 of 10 symptoms. A total of 315 cases of EHF, with an 81% case
fatality, were identified, excluding 10 clinical cases with negative laboratory results. The earliest
documented case-patient had onset on 6 January, and the last case-patient died on 16 July. Eighty
cases (25%) occurred among health care workers. Two individuals may have been the source of
infection for ú50 cases. The outbreak was terminated by the initiation of barrier-nursing techniques,
health education efforts, and rapid identification of cases.
No pestilence had ever been so fatal, or so hideous.
In 1977, a fatal case of EHF was recognized in a child from
Edgar Allan Poe, The Masque of the Red Death, 1842
Tandala, DRC, 325 km from Yambuku [5]. Investigation of
this case led to the retrospective identification of a case in 1972
Ebola (EBO) hemorrhagic fever (EHF) is a lethal viral hem-
and suggested that EHF was endemic but sporadic in northern
orrhagic fever with person-to-person transmission and the po-
DRC. In 1979, a 34-person cluster of cases with a 65% case
tential to cause major epidemics. It was first recognized during
fatality was noted in southern Sudan; the cases had direct links
the course of parallel outbreaks in Equateur region in northern
to an index patient employed in the same textile factory that
Democratic Republic of the Congo (DRC) and in adjacent
was implicated in the 1976 outbreak [6]. In Co
ˆ
te d’Ivoire in
southern Sudan in 1976 [1, 2]. The outbreak in northern DRC
1994, the isolation of a new strain of EBO from an ethologist
was centered around the town of Yambuku near the banks of
who had necropsied a chimpanzee not only extended the geo-
the Ebola River and was characterized by extensive nosocomial
graphic distribution of known cases and strains but also pro-
transmission in a mission hospital (mainly from needle reuse),
vided another tantalizing clue about a possible role for nonhu-
with 318 documented cases and 280 deaths (88% case fatality)
man primates as a link for these viruses in human outbreaks
over a 2-month period. The southern Sudan outbreak appeared
[7]; imported cynomolgus monkeys (Macaca fascicularis) from
to originate among workers at a cotton factory in Nzara and
the Philippines had been associated with the Reston subtype
subsequently spread to Maridi, where disease was amplified in
of EBO virus in the United States in 1989 and 1990 [8–10].
a large, active hospital. A total of 284 cases, with 151 deaths
Also in 1994, an outbreak of EHF occurred north of Makoukou,
(53% case fatality), were documented over a 5-month period.
the provincial capital of Ogooue
´
-Ivindo, Gabon; there were 44
Despite the geographic and temporal coincidence of these out-
cases and 28 deaths (64% case fatality). This outbreak was
breaks, the implicated viruses were genetically related but dis-
originally misdiagnosed as yellow fever and subsequently was
tinct subtypes [3]. The necessity for the highest level of biologic
ascribed to the Zaire subtype of EBO virus [11]. Although key
containment for these agents was rapidly established after a
epidemiologic and control measures were elucidated during
laboratory infection occurred in England [4].
these investigations, a reservoir for these virus subtypes re-
mains unknown.
In May 1995, the Centers for Disease Control and Prevention
(CDC) was notified of an outbreak of viral hemorrhagic fever
Reprints or correspondence: Dr. Ali S. Khan, Special Pathogens Branch,
in Kikwit, DRC. The outbreak was of unknown magnitude but
CDC, MS A26, 1600 Clif ton Rd., N.E., Atlanta, GA 30333.
had caused the death of at least 2 Italian nuns and other mem-
The Journal of Infectious Diseases 1999;179(Suppl 1):S76 – 86
bers of a surgical team and their subsequent contacts. Speci-
q 1999 by the Infectious Diseases Society of America. All rights reserved.
0022–1899/99/79S1–0014$02.00
mens sent to Belgium were forwarded to CDC for diagnostic
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S77JID 1999;179 (Suppl 1) EHF in Democratic Republic of the Congo, 1995
testing, and the presence of acute or recent EBO virus infection Specimens were obtained on 4 May and sent to the Institute
of Tropical Medicine. The specimens were forwarded and sub-
was confirmed for all 14 persons tested. Within 48 h of diagnos-
sequently arrived at CDC on 9 May, where testing confirmed
tic confirmation, a multinational contingent led by the World
EBO virus as the etiologic agent for the disease in all 14
Health Organization (WHO), composed of physicians, epide-
patients. From that point on, the original ad hoc committee
miologists, sanitarians, health educators, and logistic support
evolved into an international commission to manage the out-
personnel from WHO, CDC, Institut Pasteur, Epicentre, Me
´
de-
break. The various activities of this commission, such as patient
cins sans Frontie
`
res, Institute of Tropical Medicine, and the
management, community education, transportation of cadavers,
International Red Cross were dispatched to assist DRC authori-
and research activities, are described elsewhere in this supple-
ties in controlling and characterizing the outbreak [12 – 15].
ment.
This report describes the surveillance methodologies and key
Not unexpectedly, surveillance activities proved extremely
epidemiologic characteristics of this outbreak.
difficult. Initial case counts were based on nonsystematic re-
ports from various hospitals, clinics, and private individuals.
Background
Reports of cases and deaths often conflicted from day to day.
Because no public health surveillance system existed within
Kikwit (population Ç200,000), the former regional capital
Kikwit or its environs, one had to be constructed de novo by
of Bandundu region, is located on the banks of the Kwilu River,
using a patchwork of paid and unpaid volunteers with various
Ç475 km east of Kinshasa, the capital of DRC. The economy
levels of skill, experience, background, and tenure. The foreign
of the area is based on agriculture, hunting, and fishing, with
members of the international commission were just as diverse,
the majority of the populace commuting to the surrounding
and the problem was compounded by language and cultural
secondary forest for subsistence. As in the rest of DRC, a
differences. Surveillance was hindered by many logistic diffi-
declining public sector has recently had devastating effects on
culties, including the lack of transportation and telephones. It
the community’s health care, education, transportation, and
was also hampered by confounding societal forces that led to
communication facilities. The two major hospitals are Kikwit
concealment and denial of cases to avoid the social stigmata
General Hospital (350 beds) and Kikwit II Maternity Hospital
that rapidly became attached to those with the disease and to
(100 beds). Reportedly, a small nosocomial cluster of EBO
the contradictory reporting of noncases in the belief that this
cases among the nursing staff at Kikwit II Maternity Hospital
could lead to some tangible benefit for the afflicted family or
was first recognized but misdiagnosed as epidemic dysentery
village or reporter. Retrospective surveillance was also initially
in early April 1995.
limited by the early impression that all febrile hemorrhagic
Toward the end of April, a similar nosocomial cluster was
deaths prior to that of the laboratory technician from Kikwit
identified at Kikwit General Hospital among the operating room
II hospital were due to epidemic dysentery.
staff who participated in a surgical procedure on a laboratory
The commission’s health education campaigns proved chal-
technician employed at Kikwit II Maternity Hospital; the pa-
lenging because of the lack of mass media, closure of all
tient had been transferred on 10 April to Kikwit General Hospi-
schools, unpaid health care workers in hospitals and many
tal for a laparotomy to treat a suspected typhoid-associated
health centers, and an estimated literacy rate of 55%–61%.
abdominal perforation. No perforation was noted intraopera-
Education of the populace was conducted using leaflets, post-
tively, and the patient underwent an appendectomy during the
ers, and banners strung across the main roads. The Kikwit
exploration. The next day, the patient had increased abdominal
Diocese of the Catholic Church helped to disseminate accurate
pain and distention, with a diagnostic bloody abdominocente-
educational material at church gatherings. In addition, educa-
sis. A repeat laparotomy was done to exclude bleeding around
tional messages were broadcast by megaphone through the
the appendix suture site; it revealed diffusely bleeding abdomi-
streets of Kikwit and by word of mouth from commission
nal contents. The patient died 2 days later. Within several days
members to the populace. Health care providers at the local
of his death, physicians and nursing personnel, including 2
hospitals received individual instruction on the use of protective
Italian missionary nurses involved in his care, became ill with
clothing and equipment, diagnostic features of the disease, case
a febrile disorder suggestive of a viral hemorrhagic fever. The
definitions, and patient management. Training for Ç60 health
unusual nature of these cases was recognized independently at
care workers from the Kikwit sentinel health clinics and 30
Kikwit II Maternity and Kikwit General Hospitals and two
regional physicians was conducted in 2 separate sessions on
other regional health facilities that managed these patients and
12 and 14 May, respectively. Individual training sessions for
their subsequent contacts.
other local and regional health care workers and facilities were
An ad hoc committee was locally established on 1 May
conducted as needed, and protective equipment was provided.
1995 to examine the presumptive epidemic dysentery deaths.
Methods
In consultation with the physicians at Kikwit General Hospital,
Case Definitions
J. J. Muyembe-Tamfum, Ministry of Health, DRC, who had
participated in the follow-up of the 1976 EHF outbreak, sug-
The field-case definition was modified during the course of the
outbreak investigation. The initial field-case definition required
gested that these deaths were due to viral hemorrhagic fever.
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S78 Khan et al. JID 1999;179 (Suppl 1)
that cases have contact with another case-patient and that they first aid workers) and others who went from door to door in neigh-
borhoods to identify probable cases.meet clinical illness criteria, and the definition categorized sus-
pected cases into either probable or clinical (‘‘clinque’’) cases on Hospital and sentinel health clinic surveillance. A2-or3-
member team was assigned to review the patient logs at Kikwitthe basis of the presence of hemorrhage. Since education of sus-
pected cases and their family members was our major intervention General Hospital and the 11 main functioning mission and public
(i.e., sentinel) health clinics at which staff had initially beenstrategy, this case definition was modified on 13 May to be highly
sensitive and identify individuals who may not have recalled or trained, and they provided protective equipment. Case-report forms
were completed for all patients who met the surveillance casehave had an obtainable history of contact with a case-patient. A
probable case was defined as someone who had (1) an unexplained definition and forwarded to the central registry for completion or
for inclusion as case reports and for follow-up.fever and contact with another case-patient; (2) unexplained fever
plus §3 of 10 symptoms (abdominal pain, anorexia, asthenia, Death registry. By 13 May 1995, most of the corpses for fatal
cases of suspected EHF and those for persons with unexplainedsimple [nonbloody] diarrhea, dysphagia, dyspnea, headache, hic-
cups, myalgia or arthralgias, and nausea or vomiting); or (3) unex- deaths were collected by the secourists, who forwarded case-report
forms to the central registry for completion.plained acute hemorrhagic signs or symptoms, such as melena,
hematemesis, petechiae, or epistaxis. In an effort to completely Active surveillance and family follow-up. A temporally vari-
able group of 22–44 second- and third-year medical students fromascertain the geographic and temporal nature of this outbreak, this
case definition was not restricted to any region of DRC or to a the Bandundu Medical School, Kikwit, were paired into teams to
follow up on all suspected cases to complete case-report formstime frame. Operationally, however, case-finding occurred in the
Bandundu region from 13 May to 1 July 1995 for cases with onset and to determine if the individuals met the surveillance case defini-
tion. If persons who met the case definition were still residing indates after 1 January 1995 (prior to the onset date of the first
identified case). A case-report form was completed on all such the household, they were instructed to seek medical evaluation
and possible hospitalization at Kikwit General Hospital to clini-probable case-patients in addition to all unexplained deaths for
whom additional information was garnered to determine if they cally verify the diagnosis and prevent spread to household mem-
bers. Family members of probable case-patients were educated onotherwise met the case definition. This field-case definition was
the basis for all case counts reported during the course of the how to reduce their risk of infection if the ill individual continued
to reside in the household. These family members were revisitedoutbreak after the arrival of the international team.
For the purpose of this report, with the availability of the final at least 3 times a week for 3 weeks following their last contact
with the probable case-patient prior to death or convalescence tolaboratory results, the case definition was revised as follows: (1)
individuals with negative laboratory results who otherwise met the ascertain if secondary transmission had occurred. Nurses pre-
viously trained in the sentinel clinics also visited the household offield-case definition were excluded and (2) acutely ill individuals
for whom clinical information was insufficient to meet the field- probable case-patients to distribute basic protective material (e.g.,
a pair of gloves, soap, and wash basin) as needed and to reinforcecase definition but who otherwise had laboratory data confirming
an acute or recent EBO virus infection (i.e., presence of detectable the educational messages about risks of transmission and symp-
toms suggesting disease in subsequent family members.antigen or IgM antibodies) were included as case-patients. Healthy
individuals with no recent disease history and from whom blood
specimens were drawn for antibody surveys were not included as
case-patients if EBO antibodies were detected.
Regional Surveillance
Bandundu regional surveillance and case-finding was divided
into two separate components maintained by the regional surveil-
Surveillance and Case-Finding
lance team, which was responsible for appropriate follow-up and
interventions. Efforts were markedly hampered by the scarcity of
Surveillance was conducted separately for the city of Kikwit
motor vehicles and an almost nonexistent communication system.
and the surrounding areas of Bandundu region, but the data were
Rumor surveillance for suspect cases. Similar to the system
coordinated through a central data registry and epidemiologic sub-
within the city of Kikwit, suspected cases from within Bandundu
committee. After 1 July 1995, reports of probable cases were no
region were reported to the commission. Cases from outside Ban-
longer being systematically collected and were not recorded in the
dundu region were reported to the National Committee on Hemor-
central database, although DCR officials continued to monitor the
rhagic Fevers in Kinshasa.
families of some of these individuals.
Mission radio network surveillance. Voice contact by short-
wave radio system (‘‘phonies’’) was made each morning with all
23 Catholic missions functioning in the Diocese of Kikwit, an area
Surveillance in Kikwit and Surrounding Area
comprising several health zones (‘‘zones de sante
´
’’) around the
city, to inquire about possible case-patients and obtain follow-upCity of Kikwit surveillance and case-finding during the acute
phase of the outbreak was divided into four separate components information on villages with known cases within the last 3 weeks.
In addition, individuals from Protestant churches and from theas follows.
Rumor surveillance for suspect cases. Any persons suspected public with radios were contacted as needed.
All suspected case-patients were visited by members of the re-of having viral hemorrhagic fever could be reported by the public
to 2 sites in Kikwit for appropriate follow-up and intervention. gional surveillance team from Kikwit or by local health care work-
ers who had been trained in Kikwit. Probable case-patients wereThese activities were supplemented by the secourists (Red Cross
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S79JID 1999;179 (Suppl 1) EHF in Democratic Republic of the Congo, 1995
confined in their households, instructions for care were given, and
Hospital on 13 January 1995. He directly infected at least 3
basic protective equipment was provided to the primary care giv-
members of his family, all of whom died, and an additional
ers. Family members and other villagers were educated about the
10 secondary cases (all fatal) occurred among members of his
disease and monitored for secondary cases once or twice weekly
extended family over the next 9 weeks in an area encompassing
for 3 weeks after the last contact with the probable case-patient.
Kikwit and 3 surrounding villages. This case-patient had no
known identifiable contact with another EHF case-patient and
is presumed to have been infected from the natural reservoir
Additional Surveillance
at his charcoal pit or farm in Mwembe, 15 km from his resi-
Retrospective case-finding was conducted to identify chains of
dence in Kikwit. Ill persons from this family and secondary
EHF transmission. Several retrospective cases were found by ex-
and tertiary cases in other families that provided nursing care
amining the reported source of infection in addition to interviewing
and participated in the burial rituals of family members of this
health care workers and affected families. Surveillance activities
family were traced through the community and into Kikwit II
were maintained by testing postmortem skin snips [16] and liver
Maternity Hospital, where a small nosocomial outbreak among
biopsy specimens from deceased possible case-patients after 20
9 hospital employees started in the middle of March. The retro-
June 1995, the presumptive date of onset of the presumed last
spective identification of GM and other individuals strongly
identified case. This individual was discharged from isolation on
supports our belief that many ‘‘sporadic’’ cases diagnosed as
14 July 1995, which was the referent date to declare the end of
‘‘diarrhe
´
e rouge’’ (bloody diarrhea or dysentery) before the
the outbreak 6 weeks later [17]. However, the case definition for
recognition of the outbreak were actually EHF and part of a
selecting patients with onset dates after 30 June became highly
variable and erratic without central organization; therefore, these
chain of transmission going back to the presumed primary case
cases, except for a single laboratory-confirmed case, are not in-
for the outbreak.
cluded in the final case counts.
Besides the initial family cluster and other associated cases,
EHF was already well established in Kikwit II maternity and
Kikwit General Hospitals before the large nosocomial outbreak
Data Collection and Analysis
in Kikwit General Hospital. In fact, the main focus of the
nosocomial outbreak in Kikwit General Hospital was patient
Information in our report was obtained from surveillance case-
report forms that were used to solicit basic demographic data and
WB, who was initially thought to have been infected by the
selected clinical and exposure information. Case-report forms were
transferred laboratory worker; however, WB appears to have
completed as soon as EHF was suspected or after the fact from
been infected from a prior independent contact with an obstetric
retrospectively identified case-patients. Although detailed chart re-
patient who had a febrile hemorrhagic disease. Similarly, an
views were available for a minority of patients, this information
obstetric nurse at Kikwit General Hospital was initially thought
was used only to supplement the demographic data from the case-
to have been infected by the transferred laboratory worker;
report forms. The clinical descriptions from these other data
however, she had no contact with the technician and apparently
sources are presented by Bwaka et al. elsewhere in this supplement.
contracted EHF from a nosocomially infected patient who had
Data collection was not uniform for all suspected cases because
undergone a cesarean section at Kikwit II hospital and was
of the evolution of the surveillance system and inherent difficulties
subsequently admitted to Kikwit General Hospital.
in collecting information about deceased persons, but similar vari-
ables were analyzed in this study.
The date of disease onset was unavailable for Ç10% of sus-
Amplification and Time Course of the Epidemic
pected cases, and for the rest, it was often disputable. However,
the date of death was invariably available and rarely in dispute.
At least four generations of cases were traced through the
For certain representation purposes, the month of onset has been
hospital and into the community from the initial nosocomial
calculated as either the difference of the mean onset to hospitaliza-
cases in Kikwit General Hospital (figure 1). This periodicity
tion from the date of hospitalization or the difference of the mean
of deaths has not been previously noted. The epidemic was
onset to death from the date of death. Comparisons between de-
interrupted coincident with the arrival of protective equipment
ceased patients and survivors were made using the
x
2
or Fisher’s
and intensive training and education efforts. The beneficial
exact (2-tailed) tests, as appropriate. The Wilcoxon rank sum test
effect of initiating barrier-nursing precautions in Kikwit Gen-
was used to compare age, phase of the outbreak, and various
eral Hospital, the site designated for management of all suspect
durations [18].
cases, was clearly evident: Only 1 health care worker developed
disease after institution of preventive measures (8 days later;
Results
figure 2). All health care workers, except the 1 who became
sick after institution of the extended personal protective mea-
Index Patient and Initial Cluster of Cases
sures, had previously provided care without these precautions
and had become ill within an incubation period of their lastThe first identified case-patient was GM, a 42-year-old male
charcoal worker and farmer who became ill on 6 January 1995 unprotected contact with case-patients. The final infected health
care provider had laboratory-confirmed infection 15 days afterand died of a febrile hemorrhagic disease at Kikwit General
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S80 Khan et al. JID 1999;179 (Suppl 1)
Figure 1. Fatal Ebola hemorrhagic fever cases by date of death, Bandundu region, Democratic Republic of the Congo, 1 March to 21 July
1995. Mar Å March, Apr Å April, Jun Å June, Jul Å July.
institution of these precautions. This provider tended 2 case- subsequent disease as part of the follow-up surveillance for
suspect deaths. EBO antigens were detected in this patient’spatients in a separate 2-room building and wore a disposable
HEPA filter mask and eye protection in addition to gloves, tissue specimen by immunohistochemistry on 8 September
1995.gowns, and boots; however, she claimed to have inadvertently
rubbed her eyes with soiled gloves. In addition, we could not
exclude the possibility of an occult needle stick or otherwise
Case-Finding and Characteristics of Case-Patients
assess adherence to all other infection-control precautions.
The last identified case was a 27-year-old housewife from A total of 315 case-patients met the revised case definition
for EHF; 314 had onset dates between 6 January and 20 JuneNzinda, Kikwit. She died at home on 16 July 1995, shortly
after discharge from Kikwit II hospital, where she was hospital- 1995, and the case-patient who died on 16 July (mentioned
above) was retrospectively added to the final case count (figureized for management of a septic abortion. Her cadaver was
brought to Kikwit II hospital on the day she died and was 3). All case-patients for whom information was available re-
sided within 3 administrative subregions of Bandundu region:transferred to the morgue, where a postmortem biopsy speci-
men was obtained the next day and transported to CDC with Kikwit (262 cases), Kwilu (45 cases), and Kwango (1 case)
(figure 4). This area represented 10 of 39 health zones encom-other such samples on 27 August 1995. Her surveillance case-
report form indicated that she had contact with a case-patient, passing Ç30 affected villages, although 268 (87%) of the cases
came from Kikwit, and 17 (6%) came from Mosango. Theseand she had fever, headache, vomiting, anorexia, diarrhea, as-
thenia, abdominal pain, dyspnea, and melena. Additional his- two towns were the principal foci of infection and source of
secondary case-patients, some of whom went back to theirtory was unavailable at that time, but the diagnosis of EHF or
septic shock was indicated on the form by the medical director villages and died without subsequent disease transmission. A
major hospital located in Mosango (80 km west of Kikwit)of Kikwit II hospital. The postmortem sample was not flagged
in any way to indicate that it was obtained from a patient with admitted patients and was a site for subsequent nosocomial
transmission of disease. All affected villages were within a 150-a highly suspicious clinical illness and was collected as part
of the nonsystematic process by which access to corpses was km radius of Kikwit. A single case-patient, who was infected in
Kikwit, was hospitalized in Kinshasa, but no secondary trans-provided after the presumptive last case. However, 13 family
members of this individual were monitored for 21 days without mission occurred despite the initial lack of barrier-nursing pre-
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S81JID 1999;179 (Suppl 1) EHF in Democratic Republic of the Congo, 1995
Figure 2. Ebola hemorrhagic fever cases by date of symptom onset and occupation, Bandundu region, Democratic Republic of the Congo,
1 March to 21 July 1995. Arrow indicates date of initiation of upgraded infection control practices. Mar Å March, Apr Å April, Jun Å June,
Jul Å July.
cautions. No case-patient associated with this outbreak was two case-patients (7.5%) were children or adolescents (£16
years). Eighty (25%) of the 315 case-patients were health careidentified outside DRC.
Of 310 case-patients for whom information was recorded, workers. The overall case fatality was 81% (250/310). A linear
decrease in mortality was noted from January (100%) to June165 (53%) were female. The average age of case-patients was
37.4 years (median, 37; range, 2 months to 71 years). Twenty- (62%) (
x
2
for linear trend by month, P õ .0001). A significant
Figure 3. Ebola hemorrhagic fever cases by month of
onset, Bandundu region, Democratic Republic of the
Congo, January to July 1995. Month of onset could not
be calculated for 9 case-patients. Jan Å January, Feb Å
February, Mar Å March, Apr Å April, Jun Å June, Jul
Å July.
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S82 Khan et al. JID 1999;179 (Suppl 1)
Figure 4. Geographic distribution of Ebola hemorrhagic fever cases, Bandundu region, Democratic Republic of the Congo (DRC), 1995.
EHF Å Ebola hemorrhagic fever.
difference in age was noted among survivors compared with with maternal fatality; however, these data were not collected
systematically.nonsurvivors (32.8 vs. 38.5 years, respectively; P £ .01). A
crude onset-to-onset period based on an estimation in the mean The field case definition was met by 318 persons, 92 of
whom had samples collected for diagnostic testing. Laboratorydifference in onset dates between a case-patient and the impli-
cated source of infection was 14 days. testing was positive for EBO virus infection for 82 case-patients
and negative for 10. Only 315 cases met the revised case defi-Hospitalization was noted for 80% (247/310) of case-pa-
tients. The mean number of days from onset to hospitalization nition because the 10 individuals with negative laboratory re-
sults were replaced by 7 case-patients, as indicated in the Meth-for 219 patients was 5 days (median, 4; range, 0–19). There
was a mean of 4.6 days (median, 4; range, 0–20) from hospital- ods section, with positive laboratory results and for whom
clinical information was not available. The positive predictiveization to death for 185 nonsurvivors for whom data were
available. The mean number of days from hospitalization to value of the field-case definition was 89%. The high positive
predictive value of the case definition used in the field is baseddischarge for 34 survivors for whom data were available was
17 (median, 14; range, 0 – 56), and the mean number of days mainly on laboratory confirmation during the second half of
the outbreak because only 1 person with disease onset prior tofrom symptom onset to death for 224 patients was 9.6 (median,
9; range, 0–34). Reported signs and symptoms up to the time 14 May 1995 had laboratory-confirmed infection. Moreover,
because specimens were obtained mainly from survivors, wewhen the case-report form was completed (often on admission
to the hospital but frequently covering the whole course of cannot exclude that this estimate may include a selection bias
because of the large number of missing laboratory confirma-illness for retrospectively identified case-patients) are listed in
table 1. Some form of frank hemorrhage (gingivorrhagia, me- tions for case-patients who died; all the case-patients with labo-
ratory-negative results were alive.lena, hematemesis, or petechiae) was reported for 37% (82/
219) of case-patients; subconjunctival hemorrhage was ex- Prior contact with a suspected case-patient was reported by
93.5% (159/170) of case-patients for whom data were available.cluded because this sign was often confused with conjunctivitis.
Eight abortions were reported, 4 of which were concomitant Eleven case-patients reported receiving an injection within 3
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S83JID 1999;179 (Suppl 1) EHF in Democratic Republic of the Congo, 1995
Table 1. Signs and symptoms recorded for 219 Ebola hemorrhagic fever case-patients, surveillance
data, Democratic Republic of the Congo, 1995.
Sign or symptom No. (%) with No. without Not recorded Comment
Headache 160 (73) 57 2
Nausea/vomiting 154 (70) 60 5 Not differentiated
Anorexia 159 (73) 57 3
Diarrhea 162 (74) 53 4
Asthenia 170 (78) 46 3 Severe weakness
Abdominal pain 123 (56) 89 7
Myalgia/arthralgias 111 (51) 95 13
Dysphagia 89 (41) 120 10
Dyspnea 55 (25) 154 10
Hiccups 32 (14) 95 92
Gingival hemorrhage 46 (21) 162 11
Conjunctival inflammation/hemorrhage 75 (34) 135 9 Not differentiated
Petechiae 33 (15) 175 11 Difficult to assess
Melena 30 (14) 178 11
Hematemesis 28 (13) 182 9
weeks of onset of symptoms (mean, 3 days; range, 1 – 6). An presence of classic disease with clear chains of transmission,
unexpected observation during this outbreak was the identifi-
multiple hospitalizations, and a very high case fatality rate is
cation of 2 persons as a potential source of infection for Ç20%
disturbing. Equally disturbing was the initial misdiagnosis of
of all the case-patients (figure 5). The first individual, desig-
dysentery, resulting in an additional 4- to 6-week delay before
nated as identification number (IDNUM) 3 (patient WB), was
the correct diagnosis was considered and confirmed. This out-
a 29-year-old man associated with 38 secondary cases, although
break demonstrated once again the propensity for this disease
10 of these also reported other potential sources of infection.
to affect health care workers when proper barrier-nursing pro-
He was a popular anesthetist at Kikwit General Hospital who
cedures are absent (25% of all cases were among health care
died with fulminant hemorrhagic disease at the end of April
providers), and it demonstrated the potential for this disease to
and had many visitors during his 8-day hospitalization. The
be amplified in health care settings, even with little or no needle
second person, IDNUM 2260, was a 45-year-old woman who
reuse. Moreover, we reaffirmed that education and the use of
died of hemorrhagic disease at Kikwit II Maternity Hospital at
personal protective equipment can rapidly interrupt ongoing
the end of May with 21 case-patients among her contacts, none
disease transmission. These features emphasize the necessity
of whom reported other sources of infection. Conflicting details
of rudimentary public health surveillance coupled with adher-
of her case history suggest that she was misdiagnosed with
ence to barrier-nursing precautions and infection-control prac-
dysentery before her cadaver was turned over to the family.
tices, such as elimination of needle and syringe reuse or proper
Although the disease was well recognized in the community,
sterilization of these items between uses.
details to explain the large number of cases among her contacts
Experimental studies of nonhuman primates have been help-
are obscure; however, it appeared that most contacts occurred
ful in assessing the potential routes of transmission of EBO
postmortem during her traditional funeral, which included
virus. Late in infection in monkeys, there are large amounts of
washing and touching of the cadaver.
virus in saliva, feces, and urine, and ultrastructural and immune
electron microscopy examination of tissue from the respiratory
tract of monkeys has shown virus replication in type 1 pneumo-
Discussion
cytes, with abundant, apparently infectious, virions in alveoli
and bronchi [19 – 21]. EBO virus requires only 1– 2 pfu of EBO
The magnitude of this EHF outbreak in DRC after an 18-
virus must be retained in the alveoli to initiate a lethal infection
year hiatus dramatically confirms the potential of EBO virus
in the African green monkey (Cercopithecus aethiops), and
to cause large outbreaks associated with high fatality. This was
filoviruses are moderately stable in aerosol [19, 22]. Thus, it
the first outbreak in a large population center in proximate
is not surprising that intercage transmission, suggesting media-
location to the even larger population of Kinshasa, with its
tion by small-particle aerosols, has been reported [23]. How-
established intra- and intercontinental transportation links. The
ever, direct application of EBO virus onto conjunctiva and
newly described waves of death in the Kikwit community may
ingestion has resulted in fatal monkey disease; therefore, large-
be a reflection of the density of this large population. The 3-
droplet dissemination may well be another mechanism for this
month latency between occurrence of the first case in Kikwit,
in January 1995, and recognition of the outbreak despite the observed intercage transmission [24].
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S84 Khan et al. JID 1999;179 (Suppl 1)
Figure 5. Ebola hemorrhagic fever cases by date of symptom onset and source of infection (patient IDNUM3 [identification no. 3], ID-
NUM2260, or other), Bandundu region, Democratic Republic of the Congo, 1 March to 30 June 1995. Shaded and black columns indicate
period of illness, from onset to death, for IDNUM3 and IDNUM2260 and those infected by them. Mar Å March, Apr Å April, Jun Å June.
The observed environmental stability of EBO virus (Swanep- membranes and indirectly via fomites and body fluids. The
periodicity of deaths implies that direct contact due to burialoel R, unpublished data, 1995) raises the possibility of fomites
as a source of infection [22]. Recent evidence of a copious rituals, which included washing and touching the cadaver, may
have played a role in disease transmission. The pattern of trans-amount of viral antigen in skin and sometimes within the lumen
of sweat glands, although its viability is unknown, suggests a mission of EHF from the infected person to the primary care
provider and from that person to their primary care provider,potential mechanism for direct disease transmission through
abraded skin or inoculation onto mucous membranes [16]. without affecting household members who did not share nurs-
ing duties but who may have slept in the same room, furtherHowever, the barrier presented by intact skin is unknown. Di-
rect observations in deceased EHF patients show the same suggests that airborne transmission was an unlikely route of
transmission and is consistent with similar observations duringabundant alveolar particles [17], but we have no way of know-
ing if this material is aerosolized, is stable in the aqueous milieu the 1976 EHF outbreaks [1, 2, 25].
A large number of case-patients identified 2 case-patients asin which it is suspended [22], or is an equivalent infection for
humans; thus, we must rely on field epidemiology to assess the their contacts, both of whom had gastrointestinal hemorrhag-
ing. Although the number of contacts may have been biasedinfluence of these multiple potential sources of transmission.
Field epidemiologic studies have found no human correlate by the reporting of an individual who was known to be a
suspected case-patient and some reported secondary contacts,for aerosol infection, and airborne transmission has not ap-
peared to play a role in human-to-human transmission, except the sheer magnitude of the effect suggests that it is not arti-
factual. The concept of a ‘‘super-spreader’’ or ‘‘high-frequencyfor possibly in a rare subset of patients. The preponderance of
evidence during this outbreak suggests that EBO virus was transmitter’’ is novel for this viral hemorrhagic fever, and the
mechanism for this high-frequency transmission is unknown.once again transmitted between humans via direct contact or
direct projection (droplet spread) of droplets onto mucous The large number of cases may simply have been a matter of
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S85JID 1999;179 (Suppl 1) EHF in Democratic Republic of the Congo, 1995
a larger number of contacts of these case-patients or some vulnerability of human populations to these infections, and they
highlight the potential for further large outbreaks and the lackinherent differences in the virus-host relationship. These differ-
ences could include a more virulent virus strain or high levels of information regarding the natural reservoir.
Although investigation of previous cases has helped to refineof viral shedding; unfortunately, no samples are available from
these case-patients. Enhanced transmission via a respiratory control measures, public health authorities remain unprepared
to anticipate and prevent future large outbreaks. Preventionroute (either small-particle aerosols or large droplets) is a re-
mote possibility that is supported by the experimental data and requires better definition of the epidemiologic features of these
diseases by clarifying their geographic distribution, the spec-evidence that a portion of experimentally inoculated animals
are more likely to have alveolar disease. However, the onset trum of illness, and risk factors for both disease acquisition
and transmission. Targeted serologic surveys using newly de-dates of these secondary cases suggest that most of the contacts
occurred during the course of a traditional funeral, when aero- veloped diagnostic assays that circumvent the specificity prob-
lems of older assays should also be done to define these issues.sols are unlikely to be generated. A similar increased risk for
transmission of viral hemorrhagic fevers in a hospital setting While recognizing that the general lack of facilities and infra-
structure remains the paramount problem in dealing with thishas been reported for Lassa fever and Bolivian hemorrhagic
fever [26, 27]. An analogous super-spreader concept has been disease, we also need to provide ‘‘field-friendly’’ mechanisms
prepositioned at regional sites for rapid identification of in-recognized for rubella and group A streptococci in addition to
others [28, 29]. fected individuals to anticipate outbreaks. However, even with
simpler diagnostic technology, it will never be possible to makeKey determinants for secondary transmission of infection
remain unknown. Although the paucity of children with this a rapid field diagnosis in all locations, given the challenges
imposed by a huge geographic area with spotty communicationdisease may be explained by shielding from contact with an
ill family member, the lack of such secondary transmission in and transportation links to urban centers.
A feasible approach would be to continue uncoupling the usecertain settings is inexplicable [30].
A large number of case-patients returned to their villages, of local prevention activities from the occurrence of laboratory-
confirmed diagnoses. All patients with suspected viral hemor-yet no subsequent transmission was recorded in these settings.
Similarly, many of the initial case-patients, including the pri- rhagic fever should be considered to be sentinel events and
managed according to generic accepted guidelines [41]. A kitmary case-patient for the outbreak, were hospitalized without
evident nosocomial transmission. This was also true for the comprising personal protective equipment assigned specifically
for such patients and materials to collect specimens, with acase-patient hospitalized in Kinshasa who had active mucosal
bleeding and was admitted with little or no barrier-nursing set of guidelines on preventing disease transmission, can be
distributed to remote locations [41]. Integration of this kit intoprecautions. This unexplained lack of secondary transmission
is in contrast to the many family clusters of disease, in which national health care surveillance protocols (governmental and
nongovernmental) should help prevent additional disease asso-2 – 3 generations of secondary cases were documented. How-
ever, it further establishes that sporadic cases of unrecognized ciated with suspected EBO patients, allow for laboratory con-
firmation of remote cases, and help anticipate ‘‘hot spots’’ onEHF may not be uncommon.
Since this outbreak, EHF has continued to be identified in the basis of geographic or temporal clustering of confirmed
cases.Africa, including a case from Liberia that was not confirmed
by virus isolation [31, 32]. A recent outbreak of EHF in the
village of Mayibout 2, Makokou Health District, Ogooue
´
-
Acknowledgments
Ivindo region, near the site of the 1994 outbreak [33, 34],
was linked to the butchering, transport, and preparation for
This work is not a reflection of the relatively few individuals
consumption of a chimpanzee found dead in the forest on 22
chosen to write the report nor the many other members of Commis-
sion de Lutte contre les Epide
´
mies a
`
Kikwit who were actively
January 1996. The total number of cases was 37 (20 males, 17
involved in characterizing and controlling this epidemic. It is a
females), the mean age was 27 years (range, 7 months to 70
reflection of the hard work and dedication of the medical students,
years), and the case fatality rate was 57%. Two generations of
local officials, military personnel, Red Cross, Catholic Relief Ser-
secondary cases were documented, and the outbreak terminated
vices, Kikwit Diocese, and other concerned citizenry who educated
over a 7-week period. Most recently, there was a 7-month
the populace and helped gather the information that has been pre-
outbreak due to EBO (subtype Zaire) in Booue
´
, Ogooue
´
-Ivindo
sented. We thank Ramses Sadek for assistance with statistics, Kent
Province, Gabon, involving 60 cases (case fatality rate, 75%)
Wagoner for assistance with graphics, and John O’Connor and
associated with 8 generations of secondary cases and with sec-
Claudia Chesley for editorial review of this manuscript.
ondary cases in the capital of Libreville [35–40]. In Johannes-
burg, South Africa, the death of a nurse who was nosocomially
References
infected by a physician who was transferred for medical treat-
ment from Libreville with unsuspected EHF was also linked
1. International Commission. Ebola haemorrhagic fever in Zaire, 1976. Bull
World Health Organ 1978;56:271 – 93.
to this outbreak [36]. These outbreaks continue to highlight the
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
S86 Khan et al. JID 1999;179 (Suppl 1)
2. WHO/International Study Team. Ebola haemorrhagic fever in Sudan, filoviruses from the 1989– 1990 US epizootic. Arch Virol 1996;
11(suppl):115– 34.
1976. Bull World Health Organ 1978;56:247– 70.
21. Johnson E, Jaax N, White J, Jahrling P. Lethal experimental infections of
3. Sanchez A, Trappier SG, Mahy BWJ, Peters CJ, Nichol ST. The virion
rhesus monkeys by aerosolized Ebola virus. Int J Exp Pathol 1995;76:
glycoproteins of Ebola viruses are encoded in two reading frames and
227– 36.
are expressed through transcriptional editing. Proc Natl Acad Sci USA
22. Cheprunov AA, Chuev YP, P’yankov OV, Efimova IV. Effects of some
1996;93:3602 – 7.
physical and chemical factors of inactivation of Ebola virus. Vopr Vi-
4. Emond RTD, Evans B, Bowen ETW, Lloyd G. A case of Ebola virus
rusol 1995;2:74–6.
infection. Br Med J 1977; 2:541 –4.
23. Jaax N, Jahrling P, Geisbert T, et al. Transmission of Ebola virus (Zaire
5. Heymann DL, Weisfeld JS, Webb PA, et al. Ebola hemorrhagic fever:
strain) to uninfected control monkeys in a biocontainment laboratory.
Tandala, Zaire, 1977 –1978. J Infect Dis 1980;142:372– 6.
Lancet 1995; 346:1669 –71.
6. Baron RC, McCormick JB, Zubeir OA. Ebola virus disease in southern
24. Jaax NK, Davis KJ, Geisbert TJ, et al. Lethal experimental infection of
Sudan: hospital dissemination and intrafamilial spread. Bull World
rhesus monkeys with Ebola-Zaire (Mayinga) virus by the oral and con-
Health Organ 1983; 61:997– 1003.
junctival route of exposure. Arch Pathol Lab Med 1996;120:140– 55.
7. Le Guenno B, Formenty P, Wyers M, Gounon P, Walker F, Boesch C.
25. Dowell SF, Mukunu R, Ksiazek TG, et al. Transmission of Ebola hemor-
Isolation and partial characterization of a new strain of Ebola virus.
rhagic fever: a study of risk factors in family members, Kikwit, Demo-
Lancet 1995; 345:1271 –4.
cratic Republic of the Congo, 1995. J Infect Dis 1999;179(suppl 1)S87–
8. CDC. Ebola virus infection in imported primates— Virginia, 1989.
91.
MMWR Morb Mortal Wkly Rep 1989;38:831– 2, 837– 8.
26. Carey DE, Kemp GE, White HA, et al. Lassa fever. Epidemiologic aspects
9. Dalgard DW, Hardy RJ, Pearson SL, et al. Combined simian hemorrhagic
of the 1970 epidemic, Jos, Nigeria. Trans R Soc Trop Med Hyg 1972;
fever and Ebola virus infection in cynomolgus monkeys. Lab Animal
66:402– 8.
Sci 1992; 42:152 –7.
27. Peters CJ, Kuehne RW, Mercado RR, Le Bow RH, Spertzel RO, Webb
10. Jahrling PB, Geisbert TW, Dalgard DW, et al. Preliminary report: isolation
PA. Hemorrhagic fever in Cochabamba, Bolivia, 1971. Am J Epidemiol
of Ebola virus from monkeys imported to USA. Lancet 1990; 335:
1974;99:425 – 33.
502–5.
28. Hattis RP, Halstead SB, Hermann KL, Witte JJ. Rubella in an immunized
11. Amblard J, Obiang P, Edzang S, Prehaud C, Bouloy M, LeGuenno B.
island population. JAMA 1973;223:1010 – 21.
Identification of the Ebola virus in Gabon in 1994 [letter]. Lancet 1997;
29. Hamburger M Jr, Green MJ, Hamburger VG. The problem of the ‘‘danger-
ous carrier’’ of hemolytic streptococci. II. Spread of infection by indi-
349:181– 2.
viduals with strongly positive nose cultures who expelled large numbers
12. CDC. Outbreak of Ebola hemorrhagic fever —Zaire, 1995. MMWR Morb
of hemolytic streptococci. J Infect Dis 1945;77:96 –108.
Mortal Wkly Rep 1995;44:381 – 2.
30. Dowell S. Ebola hemorrhagic fever: why were children spared? Pediatr
13. CDC. Update: outbreak of Ebola hemorrhagic fever — Zaire, 1995.
Infect Dis J 1996;15:189 – 91.
MMWR Morb Mortal Wkly Rep 1995;44:399.
31. WHO. Ebola haemorrhagic fever— confirmed case in Co
ˆ
te d’Ivoire and
14. CDC. Update: outbreak of Ebola hemorrhagic fever — Zaire, 1995.
suspect cases in Liberia. Wkly Epidemiol Rec 1995; 70:359.
MMWR Morb Mortal Wkly Rep 1995;44:468– 9, 475.
32. Georges-Courbot MC, Sanchez A, Lu CY, et al. Isolation and phylogenetic
15. Muyembe T, Kipasa M, International Scientific and Technical Committee,
characterization of Ebola viruses causing different outbreaks in Gabon.
WHO Collaborating Centre for Haemorrhagic Fevers. Ebola haemo-
Emerg Infect Dis 1997;1:59 – 62.
rrhagic fever in Kikwit, Zaire [letter]. Lancet 1995; 345:1448.
33. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1996; 71:71.
16. Zaki SR, Shieh WJ, Greer P, et. al. A novel immunohistochemical assay
34. WHO. Outbreak of haemorrhagic fever in Gabon officially declared over.
for detection of Ebola virus in skin: implications for diagnosis, spread,
Wkly Epidemiol Rec 1996;71:125 – 6.
and surveillance of Ebola hemorrhagic fever. J Infect Dis 1999;
35. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1996; 71:320.
179(suppl 1):S36 – 47.
36. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1996; 71:359.
17. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1995; 70:241 –2.
37. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1996; 71:366.
18. Lehmann EL. Nonparametrics: statistical methods based on ranks. San
38. WHO. Ebola haemorrhagic fever —a summary of the outbreak in Gabon.
Francisco: Holden-Day, 1975.
Wkly Epidemiol Rec 1997;72:7–8.
19. Bazhutin NB, Belanov EF, Spiridonov VA, et al. The effects of the meth-
39. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1997; 72:23 –4.
ods for producing an experimental Marburg virus infection on the char-
40. WHO. Ebola haemorrhagic fever. Wkly Epidemiol Rec 1997; 72:71.
acteristics of the course of illness in green monkeys. Vopr Virusol 1992;
41. Lloyd ES, Zaki SR, Rollin PE, et al. Long-term disease surveillance in
37:153– 6.
Bandundu region, Democratic Republic of the Congo: a model for early
20. Jahrling PB, Geisbert TW, Jaax NK, Hanes MA, Ksiazek TG, Peters CJ.
detection and prevention of Ebola hemorrhagic fever. J Infect Dis 1999;
179(suppl 1)S275 – 80.Experimental infection of cynomolgus macaques with Ebola-Reston
/ 9d49$$se29 01-05-99 13:05:22 jinfa UC: J Infect
by guest on December 28, 2015http://jid.oxfordjournals.org/Downloaded from
