EBOLA FEVER IN THE SUDAN, 1976 : EPIDEMIOLOGICAL ASPECTS OF THE DISEASE
D.P. FRANCIS (1,2), D.H. SMITH (2,3), R.B. HIGHTON (2,3), D.I.H. SIMPSON (2,4), PACIFICO LOLIK (5), ISIAIH MAYOM DENG (5), ANTHONY LAGU GILLO (5), ALI AHMED IDRIS (6), BABIKER EL TAHIR (6)
1. Department of Microbiology, Harvard School of Public Health, 665 Huntingdon Avenue, Boston, Mass. 02115, U.S.A.
2. WHO Investigation Team, Sudan.
3. Division of Communicable Diseases, Ministry of Health, Nairobi.
4. London School of Hygiene and Tropical Medicine.
5. Regional Ministry of Health, Juba, Sudan.
6. Ministry of Health, Khartoum, Sudan.
A large outbreak of haemorrhagic fever (subsequently named Ebola Fever) occurred in the Southern Sudan between June and November 1976. There was a total of 284 cases; 67 in the source town of Nzara, 213 in Maridi, three in Tembura, and one in Juba. The outbreak in Nzara appears to have originated in the workers of a cotton factory. The disease in Maridi was amplified by transmission in a large, active hospital in that town. The overall mortality rate was 53". Transmission of the disease required close contact with an acute case and was usually associated with the act of nursing a patient. The incubation period was between 7 and 14 days. Although the link was not well established, it appears that Nzara could have been the source of infection for a similar outbreak in the Bumba zone of Zaire.
In June and July 1976, a severe viral haemorrhagic fever erupted in Nzara in Western Equatoria Province of Southern Sudan. From Nzara, the disease spread to Tembura, Maridi, Juba, and Khartoum in the Sudan and also involved the Bumba zone of northern Zaire where a large outbreak followed.
Investigation of the outbreak in Sudan involved finding all of the cases and recording pertinent information. Several sources were used to find as many cases as Possible: 1) Hospital records, 2) visiting homes of patients and ferreting out other cases or their contacts, 3) searching house-to-house in the infected areas for additional cases, and 4) contacting local chiefs for information.
A case was defined as: 1) anyone having the symptoms of fever and headache lasting for at least two days with the addition of gastrointestinal symptoms (diarrhea or vomiting) or chest pain; or 2) diagnosed by a physician in a hospital. Using this definition, a total of 284 cases occurred in Sudan between June and November, 1976 -- 67 in Nzara, 213 in Maridi, three in Tembura, and one in Juba.
THE ORIGIN OF THE OUTBREAK - NZARA
The origin of the outbreak was Nzara, a small town with clusters of houses scattered in the dense woodlands bordering the African rain forest zone. The total population of the area within 10 miles is estimated to be about 20,000 most of whom live in mud-walled thatchedroofed houses surrounding the town proper. The main employer in Nzara is a large agricultural corporation which has 2,000 employees, half of whom work in a large cotton factory in the town. The corporation has excellent records of employee absenteeism which facilitated the epidemic investigation. A small hospital is operative in Nzara, but at the time of the outbreak the facilities were limited and few patients were admitted.
Discussions with local people and a review of the factory records for the previous two years did not reveal any fatal haemorrhagic disease in Nzara until late June or early July, 1976. At that time, one or two factory workers per week started dying of haemorrhagic disease and subsequently their families or friends who cared for them would manifest the same symptoms. By the first week in September, six factory workers and 25 of their contacts had developed the same syndrome and 21 had died. Of the six factory employees, five worked in one specific end of the cotton factory. Extensive discussions with friends and families of these workers did not reveal any possible link between them except the factory. None had cared for any pre-existing cases of the disease nor were they ill prior to onset when they might have received an injection with a contaminated needle, nor did they have any known contact with monkeys or any other wild animals. Their houses were widely scattered over the area and their social circles were very different.
With their only link being the cotton factory, the investigation for an animal reservoir of infection was concentrated in Nzara and specifically in the cotton factory itself. The results of this investigation will be reported elsewhere.
The outbreak in Nzara continued on until late October infecting a total of 67 people of whom 31 (46%) died. Before the outbreak spontaneously died out, cases were exported to two neighbouring areas. One was Tembura, a small town 160 km to the north where an ill woman went to be nursed by her family. Subsequent to her death, the three women caring for her died of the same haemorrhagic disease. No subsequent cases were discovered in the area despite active searching.
MARIDI, JUBA, AND KHARTOUM
The other exportation was to Maridi with a very different outcome. Maridi, a town with an estimated population of 10,000 with another 5,000 people in its environs, is located about 180 km east of Nzara (between Nzara and the regional capital of Juba). The hospital in Maridi, in contrast to the one in Nzara, is an actively practising hospital with a large staff. It serves as a teaching hospital where student nurses are taught patient care. With its large staff, (total 230), it served as an ideal centre for the large hospital associated outbreak that followed the introduction of the disease from Nzara. After Maridi was infected, four cases (one from Nzara and three from Maridi) were transferred to the Regional Hospital in Juba. In late September, three of these cases were flown to Khartoum (1,200 km north) where two died. Fortunately, only one secondary case in a nurse from Juba occurred as a result of these exportations.
In total, 284 cases occurred with 151 deaths (mortality rate -53ol). All but four of the cases were in Nzara and Maridi. In these two areas, the clinical disease appeared identical and the mortality rate (46% in Nzara and 54% in Maridi) very similar. Moreover, the mortality rate stayed relatively constant through the five months of the outbreak, during which approximately 15 generations of person to person disease occurred. Monthly mortality ranged from 63% in October to 40% in November with the previous month's rates falling somewhere between. The mortality rate by age and sex showed similar, but insignificant, variations. For children, teenagers and adults it was 44%, 39%, and 56% respectively while the overall mortality rate for males was 56% and females, 48%.
CONTRAST OF NZARA AND MARIDI
The similarity of the clinical disease and its mortality in Nzara and Maridi is contrasted with the type of outbreaks in the two locations. The Nzara outbreak was centred in factory workers and spread to their families. In Maridi, the hospital served both as the focus and amplifier of the infection. After two separate importations from Nzara, (the major one during the last week of July and another during the first week of September), Maridi hospital workers became ill and they were in turn hospitalized. Those caring for them were then infected and the cycle repeated itself. The difference between Nzara and Maridi is best exemplified by examining where patients most probably became infected. Few patients (26%) were even hospitalized in Nzara and they seldom stayed more than a few days which decreased the chance of infection within the hospital. But in Maridi, almost three quarts of patients were hospitalized and often for over two weeks. As a result, Maridi hospital was a common source of infection (46% of cases) whereas the Nzara hospital was not (3% of cases).
The outbreaks that followed in both instances reflected the initial seeding of infection in each locale. The outbreak in Maridi was larger (213 cases) and accelerated rapidly after the initial importation within the hospital, whereas the Nzara outbreak was smaller (67 cases) and more sustained. The age specific attack rates reflect these differences. The Maridi attack rates were higher (overall 14.2 cases per 1,000 population compared to 3.4 cases per 1,000 in Nzara) and the cases were primarily adult males reflecting the predominance of male staff in the hospital (75%). In contrast, Nzara had equal attack rates in adults and had a moderate number of teenage male cases. These teenage cases were all associated with a single chain of transmission that links Nzara to Maridi and Juba.
The transmission of viral haemorrhagic fever in both outbreaks was similar, requiring intimate contact with a previous case. There was seldom any problem, in determining the source of infection for a case. Usually, the possible source (or sources) was well known to the patient being investigated. This is well documented in Maridi where sources were determined for all but five of 203 investigated patients, This contrasts and reemphasizes the uniqueness of the Nzara outbreak where 14 of 67 (21%) cases had no contact. Nine of these were employees of the cotton factory and possibly represent the initial introduction of the disease into the human population of the area.
Once in humans, transmissions from one person to the next were not rapid. It required close and usually prolonged contact with an acutely ill patient. Transmission did not seem to occur via the airborne route. To better define the pattern of transmission, we took a small sample of 17 highly-infected households and studied the type of contact that resulted in spread of infection. In this group, all secondary cases had slept in the same room and all secondary cases had touched the primary case during his/her acute disease. However, touching and sleeping in the same room had a relatively low attack rate (23%), whereas actually nursing of the case greatly increased the chances of becoming infected (81%). This requirement for close contact explains the lack of cases in children who, although sleeping in the same room, did not become infected. No casual infections in passersby or in contacts of primary contacts were observed.
From these observations. it is evident that nursing of a patient was almost a requirement for becoming infected (39 of 44 instances). Therefore, a hospital should be, and was, an ideal environment in which to transmit disease. At least one third of the staff of Maridi hospital had disease and 41 staff members died. All of the 6 medical assistants were infected and 41% of the student nurses. Before the disease was recognized, most wards of the hospital had haemorrhagic patients in them and at the height of the epidemic, the hospital was in chaos. In total, 93 of Maridi's 213 patients acquired their disease in the hospital. Most of these (72) were hospital staff infected during their duties. At least six others were patients who were infected by contact or injection with infectious material from nearby acutely ill patients. Fifteen additional people, probably received their infection as visitors of infectious patients in hospital. All of them were involved in the care of an acutely ill patient during their visit to the hospital.
SECONDARY ATTACK RATES IN THE COUNTRY
Once out of the hospital and into the community, the disease spread in a similar manner, but did not have the large substrate on which to feed that the hospital provided. We studied thirty-six families with 38 primary cases and listed contacts that resided in the same house (Table 1). As can be seen from the table, the original 38 cases had 232 contacts of whom 30 (13%) developed subsequent disease. Similar rates (14% and 9%) were observed in the subsequent generations giving an overall secondary attack rate of 12%. These results document the relatively slow rate of spread of this disease.
TABLE 1 SECONDARY ATTACK RATES IN FAMILY CONTACTS OF CASES, MARIDI - 36 FAMILIES.
Secondary Attack Rate
Since most cases had had prolonged contact with their source case, it was difficult to make accurate measurements of incubation periods. However, 11 cases were discovered in whom relatively short exposure times were documented and subsequent onset of disease accurately recalled. From these cases, and the lack of any documented extremely short or long generation times, we concluded that the incubation period was usually between 7 and 14 days.
The control of this outbreak in the Sudan relied on the classic public health Principles of identification and isolation. The outbreak in Nzara died out spontaneously. However, the outbreak in Maridi required intervention. Strict barrier nursing was established initially in early October, 1976 and reinforced with addition al disposable isolation equipment in mid-October. In early November, surveillance teams were established to search house-to-house in Maridi and any patients discovered were placed in a specially constructed isolation ward. With time, the surveillance was expanded to include most of Western Equatoria Province. The last known case in Sudan occurred on November 20, 1976.
LINK BETWEEN SUDAN AND ZAIRE
Despite extensive efforts, the exact link between viral haemorrhagic fever in Nzara, Sudan and Bumba Zone, Zaire, remains undetermined. There is no doubt that there is extensive traffic bringing commercial goods to and from the two areas. In fact, we interviewed one truck driver in Nzara who personally escorted an early case in Nzara to hospital on July 24, 1976 with the help of two other friends. The two other friends both developed disease respectively on the 1st and 3rd of August,1976 and died 10 days later. The driver left the day after aiding his friend and arrived in Bumba 4 days later. He stated he had no disease and that he knew of no one who travelled with him who had become ill. It is very possible that several other people who had had similar close contact with sick patients travelled from Nzara to Zaire during the course of the outbreak in Nzara. One of them may have become ill and set up a secondary chain of infection in Zaire.
Ebola fever is a newly recognized disease with many similarities to two other haemorrhagic diseases (Lassa and Marburg fevers) that are known to spread from person to person 1-4. All three diseases, although caused by different agents, are clinically similar, have high mortality rates, and are all spread among humans by close contact with infected effluvia from acutely ill patients. The most likely sources of infectious virus responsible for this person to person transmission are blood (or blood containing excreta) or urine. Respiratory spread has been postulated in one outbreak of Lassa 2 but since large numbers of secondary cases are unusual for all of these diseases, the frequency of aerosolized virus is probably low. Additional late spread has been documented via semen in Marburg disease (3).
Because of the relatively insufficient mode of spread, most outbreaks of these diseases have undergone few generations and have been short lived. However, the outbreak in Maridi is an example of the potential of these diseases to proliferate in fertile soil. The obvious means of preventing such a disaster is to recognize cases early and establish barrier nursing. Considering that 30% to 71% of patients with these diseases have haemorrhagic manifestations (3,5), it is not difficult to recognize a cluster of epidemiologically related cases. Once recognized, the diseases are not difficult to contain.
Any hospital in the tropics or any hospital receiving patients from the tropics should be on constant alert for haemorrhagic signs in febrile patients. Once cases are recognized, the following procedures should be followed :
1) Collect specimens (blood or tissue) for diagnosis. (The World Health Organization will arrange for proper shipment of specimens to high security laboratories).
2) Establish a surveillance system to identify other cases of influenza like disease with or without haemorrhagic manifestations.
3) Establish barrier nursing with gowns, gloves, and masks (or better, respirators).
a. Disinfect patients' excreta with a proper agent like formaldehyde,
b. Laboratory specimens, if taken at all , must be extremely cautiously dealt with.
4) Identify contacts and perform daily temperature surveillance. Once temperature rises, patients should be isolated.
5) Once the diagnosis is made, convalescent plasma should be dispatched to the area.
With such a simple system, these diseases can be identified, isolated, and patients properly treated. From the Lassa story, it is evident that Lassa fever is far more prevalent than 6 previously thought . The same could be equally true for Marburg and Ebola fevers. As we become more aware we will, no doubt, identify more outbreaks of these diseases and as this happens we will see that large outbreaks are indeed rare and can be prevented with simple precautions. However, if such precautions are not taken early, as was the case with the first outbreak of viral haemorrhagic fever in Sudan, these diseases can spread far and wide across international borders. The hospital, especially the referral hospital, is the site where such outbreaks can either be recognized and halted, or unrecognized and disseminated. With them rests the responsibility for stopping the spread of these dangerous diseases.
1. Carey, D.E., Kemp, G.E., White, H.A., Pinneo, L., Addy, R.F. (1972) Lassa fever-epidemiological aspects of the 1970 epidemic, Jos, Nigeria, Trans. R. Soc. Trop. Med. Hyg., 66: 402-408.
2. Monath, T.P., Mertens, P.E., Patton, R., Moser, C.R., Baum, J.J., Pinneo, L., Gary, G.W., Kissling, R.E. (1973) A hospital epidemic of Lassa fever in Zorzor, Liberia, March-April, 1972, Am. J. Trop. Med. Hyg., 22: 773-779.
3. Martini, G.A. (1969) Marburg agent disease in man, Trans. R. Soc. Med. Hyg., 63: 295-302.
4. Gear, J.S.S., Cassel, G.A., Gear, A.J., Trapper, B., Clausen, L., Meyers, A.M., Kew, M.C., Bothwell, T.H., Sher, R., Miller, G.B., Schneider, J., Koornhof, H.J., Gomperts, E.D., Isaacson, M., Gear, J.H.S. (1975) Outbreak of Marburg virus disease in Johannesburg, Brit. Med. J., 4: 489-493.
5. Mertens, P.E., Patton, R., Baum, J.J., Monath, T.P. (1973) Clinical presentation of Lassa fever during the hospital epidemic at Zorzor, Liberia, March-April, 1972, Am. J. Trop. Vied. Hyg., 22: 780-784.
6. McCormick, J.B. (1977) Lassa fever in Sierra Leone, preliminary report given at the Epidemic Intelligence Service Conference, Atlanta.