36 Rift Valley fever

Rift Valley fever (RVF) is an acute disease of domestic ruminants in mainland Africa and Madagascar, caused by a mosquito–borne virus and characterized by necrotic hepatitis and a haemorrhagic state. Large outbreaks of the disease in sheep, cattle and goats occur at irregular intervals of several years when exceptionally heavy rains favour the breeding of the mosquito vectors, and are distinguished by heavy mortality among newborn animals and abortion in pregnant animals. Humans become infected from contact with tissues of infected animals or from mosquito bite, and usually develop mild to moderately severe febrile illness, but severe complications, which occur in a small proportion of patients, include ocular sequelae, encephalitis and fatal haemorrhagic disease. Despite the occurrence of low case fatality rates, substantial numbers of humans may succumb to the disease during large outbreaks. Modified live and inactivated vaccines are available for use in livestock, and an inactivated vaccine was used on a limited scale in humans with occupational exposure to infection. The literature on the disease has been the subject of several extensive reviews from which the information presented here is drawn, except where indicated otherwise (Henning 1956; Weiss 1957; Easterday 1965; Peters and Meegan 1981; Shimshony and Barzilai 1983; Meegan and Bailey 1989; Swanepoel and Coetzer 2004; Flick and Bouloy 2005). In September 2000, the disease appeared in south west Saudi Arabia and adjacent Yemen, and the outbreak lasted until early 2001 (Al Hazmi et al. 2003; Madani et al. 2003; Abdo-Salem et al. 2006). The virus was probably introduced with infected livestock from the Horn of Africa, and it remains to be determined whether it has become endemic on the Arabian Peninsula.

The disease was first recognized in sheep in the Rift Valley in Kenya at the turn of the twentieth century, but the causative agent was not isolated until 1930. Over the next four decades, epizootics were recorded only in eastern and southern Africa, where they tended to occur in association with population explosions of floodwater-breeding aedine mosquitoes following heavy rains. Large outbreaks affecting sheep and cattle occurred in Kenya in 1930–31, 1968 and 1978–79, and lesser outbreaks at irregular intervals in the intervening years. A major epizootic, which caused an estimated 500,000 abortions and 100,000 deaths of sheep, occurred in South Africa in 1950–51; a second major and more widespread outbreak caused extensive losses of sheep and cattle in 1974–76, while lesser outbreaks were recorded in 1952–53, 1955–59, 1969–71 and 1981. Severe outbreaks occurred in the predominantly sheep farming areas of southern Namibia in 1955 and 1974–76. Further extensive outbreaks of the disease in southern Africa occurred in areas where cattle farming predominates, in Zimbabwe in 1955, 1957, 1969–70 and 1978, in Mozambique in 1969, and in Zambia in 1973–74, 1978 and 1985. In addition, evidence of the occurrence of the infection was recorded in many other southern and east African countries.

It was realized from the time of the original investigations in Kenya that febrile illness in humans accompanied outbreaks of disease in livestock, and that some patients experienced transient loss of visual acuity, but the occurrence of serious ocular sequelae was first recognized in the 1950–51 epizootic in South Africa. The first known human fatality was recorded in 1934 in a laboratory worker in the USA, but since the infection was complicated by thrombophlebitis and the patient died from pulmonary embolism, the potential lethality of the virus for man was overlooked until seven deaths from encephalitis and/or haemorrhagic fever with necrotic hepatitis were ascribed to RVF during the 1974–76 epizootic in South Africa. Subsequently deaths were also observed in Zimbabwe.

Prior to the 1970s, the presence of the virus was known for decades in the Sudan and certain west African countries from antibody studies, and there were periodic isolations of the virus in West Africa, where it was sometimes reported as Zinga virus, which is now known to be identical to RVF virus. In 1973 and 1976, outbreaks of RVF affecting livestock were reported in the Sudan. These epizootics were followed in 1977–78 by a major outbreak which occurred along the Nile delta and valley in Egypt, causing an unprecedented number of human infections and deaths, as well as numerous deaths and abortions in sheep and cattle and some losses in goats, water buffaloes and camels. Estimates of the number of human infections range from 18,000 to more than 200,000 with at least 598 deaths occurring from encephalitis and/or haemorrhagic fever. Thereafter, a severe epizootic was reported in 1987 in the Senegal River basin of southern Mauritania and northern Senegal. In Mauritania alone an estimated 224 human patients died of the disease, and there was a high rate of abortion in sheep and goats. These outbreaks in North and West Africa differed in several respects from the pattern of disease which had hitherto been observed in sub-saharan Africa; in particular they occurred independently of rainfall in arid countries, apparently in association with vectors which breed in large rivers and dams.

Since the virus is capable of utilizing a wide range of mosquitoes as vectors (Turrell et al. 2008), the occurrence of the outbreak in Egypt raised the possibility that RVF could be introduced to the mainland of Eurasia, and extensive preventive vaccination of livestock was undertaken at the time in the Sinai Peninsula and Israel. Fears were also expressed that the virus could be transported to Saudi Arabia with animals exported from Africa for ritual slaughter on the annual Islamic pilgrimage to Mecca. In the event, only isolated outbreaks of RVF were recorded in Egypt in 1979 and 1980, and thereafter the country remained free of the disease for twelve years until it was again recognized in the Aswan Governate in May 1993. On this occasion there was not the same tendency for an explosive outbreak of the disease to occur as in 1977–78, but by October 1993 infections of humans and livestock, including sheep, cattle and water buffalo, had also been recognized in Sharqiya, Giza and El Faiyum Governates (Anon. 1993; Anon. 1994).

From late October 1997 to February 1998, a large outbreak of RVF occurred in northeastern Kenya and adjoining southern Somalia, following the occurrence of heavy rains and extensive flooding in what is essentially an arid area, and extensive outbreaks of the disease also occurred elsewhere in Kenya and Tanzania (Anon. 1998; Woods et al. 2002). There were heavy losses of livestock and an estimated 500 human deaths. An agent isolated from human blood was thought to be a new bunyavirus and given the name Garissa virus, but was later found to be Ngari virus, originally isolated from mosquitoes in West Africa and recently shown to be a recombinant bunyavirus (Bowen et al. 2001; Gerrard et al. 2004; Briese et al. 2006). Antibody to Ngari virus was found in people in both Kenya and Somalia, but the importance of the virus as a human pathogen remains to be determined. An outbreak of RVF was again recognized in the North Eastern Province of Kenya in November 2006 following the occurrence of heavy rains, and by the end of January 2007 it had appeared in the Coastal, Central, Rift Valley and Eastern Provinces. A total of 684 human cases of the disease were recorded, with a 20% death rate. The disease also occurred in neighbouring Somalia with 114 cases and a 45% death rate being reported. Outbreaks were recognized in Tanzania in January 2007, but investigations revealed that livestock and human disease had already occurred in late 2006, with a total of 191 cases and a 21% death rate being recorded. Following heavy rains an outbreak of RVF occurred in October–November 2007 in White Nile, Gezira and Sennar Provinces of Sudan, with 451 cases of the disease and an approximately 36% death rate being reported (Anon. 2007a; Anon. 2007b). The high death rates reported in the recent outbreaks were estimated from cases which were diagnosed mainly on clinical grounds, without reference to mild or inapparent infections. Small outbreaks of RVF were recognized in northeastern South Africa in 1999 and 2008, and also in central Madagascar in 2008.

In September, 2000, RVF broke out simultaneously in southwest Saudi Arabia and adjoining Yemen following heavy rains on the inland mountain range (Jupp et al. 2002; Al Hazmi et al. 2003; Madani et al. 2003; Abdo-Salem et al. 2006). This was the first known occurrence of outbreaks outside of the African region. The outbreaks lasted until early 2001, and resulted in 245 human deaths and the loss of thousands of sheep and goats. There was speculation that the virus may have been imported from the Horn of Africa with infected slaughter animals, possibly during the 1997–98 epidemic in East Africa. Subsequent detection of IgM antibody in sentinel sheep suggests that the virus may have become endemic on the Arabian Peninsula (Eldafil et al. 2006).

The Smithburn strain of RVF virus, which had been isolated from mosquitoes in Uganda in 1944 and passaged intracerebrally in mice, was subjected to further passaging in embryonated chicken eggs and mice in South Africa, and issued in the form of freeze-dried infected mouse brain for use as a partially-attenuated vaccine for livestock from 1951 onwards. In 1958, reversion was made to the use of a lower mouse passage level of the virus, and since 1971 the virus has been grown in cell cultures for the preparation of freeze-dried vaccine, recommended particularly for use in non-pregnant sheep (the virus retains abortigenic and teratogenic properties for a proportion of pregnant ewes). The same strain of virus is used at a slightly different level of mouse passage for the preparation of veterinary vaccine in Kenya when demand arises. The Smithburn virus was found to be inadequately immunogenic for cattle, and since 1975 a wild strain of virus grown in cell cultures has been used in South Africa for the preparation of a formalin-inactivated vaccine for use in cattle. An inactivated cell culture vaccine for veterinary use was also developed in Egypt in 1981. An experimental formalin-inactivated cell culture vaccine for use in humans was developed in the USA in 1962, and improvements to the vaccine were made in 1981, but it was made available for use in people with occupational exposure to infection on a very limited scale only.

The virus has the morphological and physicochemical properties typical of a member of the Phlebovirus genus of the family Bunyaviridae. It is spherical, approximately 100 nm in diameter, and has a host cell-derived bilipid-layer envelope through which virus-coded glycoprotein spikes project. The genome comprises three segments of single-stranded RNA with a total molecular weight of 4x10⁶ Da, and is in the negative-sense (complementary to mRNA), except that the small segment consists of ambisense RNA, i.e. has bi-directional coding. Each of the three RNA segments, L (large), M (medium) and S (small), is contained in a separate nucleocapsid within the virion. In common with other bunyaviruses, phlebovirus virions contain three major structural proteins: two envelope glycoproteins, G1 and G2, and a nucleocapsid protein N, plus minor quantities of viral transcriptase or L (large) protein as it is termed. The L RNA segment codes for the viral transcriptase, the M segment for the G proteins and a non-structural protein, NS_m, and the S segment for the N protein and a non-structural protein, NS_s. The glycoproteins are responsible for recognition of receptor sites on susceptible cells, manifestation of viral haemagglutinating ability, and inducing protective immune response. The N protein induces production of and reacts with complement-fixing antibody. The non-structural NS_s protein synthesized during the replication of RVF virus, enters the cell nucleus to form intranuclear inclusions which are seen histologically in infected tissues. The NS_S protein acts as a major determinant of virulence by antagonizing interferon β expression after infection (Le May et al. 2008). Virus which attaches to receptors on susceptible cells is internalized by endocytosis and replication occurs in the cytoplasm. Virions mature primarily by budding through endoplasmic reticulum in the Golgi region into cytoplasmic vesicles which are presumed to fuse with the plasma membrane to release virus, but particles can also bud directly from the plasma membrane.

No significant antigenic or genetic differences have been detected between RVF isolates and laboratory passaged strains originating from widely separated countries (Battles and Dalrymple 1988; Bird et al. 2007a), but differences have been demonstrated in pathogenicity for laboratory rodents. However, it is uncertain whether this finding is reflected in differences in virulence for humans and livestock. Zinga virus, originally isolated in the Central African Republic in 1969 and long thought to be a distinct virus, and Lunyo virus, isolated in Uganda in 1955 and described as a variant of RVF virus, have both been found to be indistinguishable from RVF virus.

By analogy with the course of events believed to follow natural infection with other arthropod-borne viruses, it can be surmised that the pathogenesis of the disease may involve some replication of virus at the site of inoculation, conveyance of infection by lymphatic drainage to regional lymph nodes where there is further replication with spill-over of virus into the circulation to produce primary viraemia, which in turn leads to systemic infection, and that intense viraemia then results from release of virus following replication in major target organs such as the liver and spleen. Wild RVF virus, which has not been subjected to serial passaging in laboratory host systems, is described as being hepato-, viscero- or pantropic, and immunofluorescence studies in laboratory animals indicate that replication occurs in littoral macrophages of lymph nodes, most areas of the spleen except T-dependent peri-arteriolar sheaths, foci of adrenocortical cells, virtually all cells of the liver, most renal glomeruli and some tubules, lung tissue and scattered small vessel walls, as well as in necrotic foci in the brains of individuals which develop the encephalitic form of the disease. These sites correspond to the lymphoid necrosis in lymph nodes and spleen, hepatic necrosis and adrenal, lung and glomerular lesions seen in humans and livestock, and the brain lesions in humans (encephalitis has not been described in natural disease of ruminants). Cell damage is ascribed directly to the lytic effects of the virus, but the inflammatory response seen in human brain tissue suggests that there may also be an immunopathological element to the pathogenesis of encephalitis. The same may be true for ocular lesions. Recovery is mediated by non-specific and specific host responses, and the clearance of viraemia correlates with the appearance of neutralizing antibody. Immunity appears to be lifelong.

The mechanisms involved in the pathogenesis of the haemostatic derangement which occurs in RVF remain speculative. It is postulated that the critical lesions are vasculitis and hepatic necrosis. Destruction of the antithrombotic properties of endothelial cells is thought to trigger intravascular coagulation, and the widespread necrosis of hepatocytes and other affected cells to result in the release of procoagulants into the circulation. Severe liver damage presumably limits or abolishes production of coagulation proteins and reduces clearance of activated coagulation factors, thereby further promoting the occurrence of disseminated intravascular coagulopathy, which in turn augments tissue injury by impairing blood flow. Vasculitis and haemostatic failure result in purpura and widespread haemorrhages.

The virus can be grown in and readily produces cytopathic effect and plaques in virtually all common continuous line and primary cell cultures, including Vero and BHK21 line cells, primary calf and lamb kidney or testis cells; the only exceptions being primary macrophages and lymphoblastoid cell lines. It can be grown in embryonated chicken eggs and a variety of laboratory animals including suckling or weaned mice and hamsters inoculated by intracerebral or intraperitoneal routes. Some laboratory strains of rat are resistant, as are rabbits, guinea-pigs, chickens and African primates, but a proportion of rhesus monkeys manifests severe or fatal disease.

The virus is stable in serum and can be recovered after several months storage at 4°C or after three hours at 56°C; viraemic blood collected in an oxalate–carbol–glycerin preservative retained its infectivity after eight years of storage under a variety of conditions of refrigeration, and the virus is very stable at temperatures lower than -60°C or after freeze drying, and in aerosols at 23°C and 50 to 85% relative humidity. It is inactivated by lipid solvents, such as ether and sodium deoxycholate, and low concentrations of formalin, and infectivity is rapidly lost below pH 6.8.

Newborn lambs and goat kids are extremely susceptible to the disease, and the incubation period is short, in the range of 12–36 hours. The disease is marked by the development of fever which may be biphasic, listlessness, hyperpnoea, and disinclination to move or feed. Evidence of abdominal pain can be elicited. The course is usually peracute and lambs rarely survive longer than 24–36 hours after the onset of illness; many are simply found dead. Mortality may exceed 90% in animals less than a week old. Lambs and kids older than two weeks and mature sheep and goats are significantly less susceptible to the disease. Nevertheless, following an incubation period of 24 to 72 hours, a few animals may die peracutely without exhibiting noteworthy signs of illness. Most develop an acute disease with fever of up to 42EC that lasts for 24 to 96 hours, anorexia, weakness, listlessness and hypernoea. Some animals may regurgitate ingesta, and develop melaena or foetid diarrhoea and a blood-tinged, mucopurulent nasal discharge. A few animals may be icteric. Many sheep and goats undergo inapparent infection. Reported death rates vary from 5 to 60% for sheep, with highest mortality generally occurring in pregnant animals. Non-pregnant goats were described as resistant to the disease in some outbreaks, but suffered similar mortality to sheep in other instances.

The disease in calves resembles that in lambs and sheep, with occurrence of fever, inappetence, weakness and a bloody or foetid diarrhoea, but a higher proportion of calves may develop icterus. Death generally occurs two to eight days after infection, and estimates of mortality range from less than 10% in some outbreaks to 70% in experimentally infected one week old calves. Infection is frequently inapparent in adult cattle, but some animals develop acute disease characterized by fever of 24 to 96 hours duration, anorexia, staring coat, lachrymation, salivation, nasal discharge, dysgalactia and a bloody or foetid diarrhoea. The death rate in cattle does not generally appear to exceed 5–10%, but was reported to be 30% among cattle which aborted in Egypt. Illness tended to run a prolonged course of 10 to 20 days in cattle in the Sudan in 1973, with severe icterus being a marked feature of the disease, although most animals recovered spontaneously.

Abortion appears to be the usual, if not invariable, outcome to infection in pregnant sheep, goats and cattle. Animals may abort at any stage of gestation, and the foetuses generally have an autolysed appearance. However, abortion rates vary with epidemiological circumstances, and have ranged from 15 to 100% in different outbreaks, or in separate herds and flocks in a single outbreak. Frequently, abortion may be the only overt manifestation of disease in a herd or flock. Factors which determine the pattern of disease which occurs include the immune status of the animals, the challenge rate in the particular locality (mosquito biting frequency), and timing of the outbreak relative to the livestock breeding cycle. The offspring of immune ruminants acquire protective maternal immunity through the uptake of antibody from colostrum, but it was observed in South Africa that lambs were sometimes subjected to attack by large numbers of mosquitoes as soon as they were born, and could undergo irreversible infection before colostral immunity became effective.

Viraemia is generally demonstrable in domestic ruminants at the onset of fever and may persist for up to a week, with maximum titres of infectivity recorded being 10^10.1 mouse intraperitoneal 50% lethal doses/ml (MIPLD₅₀/ml) in lambs, 10^8.2 in kids and 10^7.5 in calves, with somewhat lower maximum titres being recorded in adult animals.

Inoculation of pregnant ewes with the live Smithburn vaccine virus between about five and ten weeks of gestation may result in the occurrence of a range of anomalies of the central nervous system including porencephaly, hydranencephaly and micrencephaly, as well as arthrogryposis and other defects in foetuses, and prolonged gestation and hydrops amnii in the ewes. Inoculation at an earlier stage of pregnancy may result in unnoticed early loss of the conceptus, while inoculation at a later stage may result in abortion, stillbirth or birth of immune or viraemic progeny. Teratology following vaccination has been recorded in the progeny of up to 15% of pregnant ewes in flocks, but on average it appears to affect less than 2% of ewes and abortion probably occurs in less than 10% of pregnant ewes.

High prevalences of antibody were found in domesticated Asian water buffaloes during the 1977–78 epizootic in Egypt, and abortion and mortality rates of 7 to 12% were recorded on some farms. Horses develop only low grade viraemia following experimental infection, but during the Egyptian epizootic there was one isolation of virus from a horse and four abortions in donkeys were ascribed to RVF, while a low prevalence of antibody to the virus was detected in the two species. No pathogenicity tests have been conducted on camels, but antibody was detected in camels in Kenya. Although there was only one isolation of RVF virus from a camel during the 1977–78 Egyptian outbreak, 56 deaths and one abortion were ascribed to the disease on the basis of circumstantial evidence. Pigs and dogs are resistant to infection, i.e. undergo inapparent infection, and birds are refractory to the virus.

Experimental RVF infection of African buffaloes (Syncerus caffer) in Kenya resulted in transient fever and viraemia, and one of two pregnant females aborted. It was noted on some properties involved in the 1950–51 epizootic in South Africa that abortion occurred in farmed springbok (Antidorcas marsupialis) and blesbok (Damaliscus dorcas phillipsi) antelope, but this was not confirmed to be due to RVF. A low prevalence of antibody to RVF virus was found in African buffaloes and a few species of antelopes in Zimbabwe, but no evidence of disease was recorded. RVF was confirmed as a cause of abortion in captive-bred buffaloes in north eastern South Africa in 1999 and again in 2008. Some species of wild myomorph rodents (rats and mice) exhibit transient viraemia following peripheral infection, and those that circulate the highest levels of virus succumb to the disease.

Although age and underlying illness undoubtedly influence the course of infection, it has been shown in cross-breeding experiments with inbred strains of laboratory rodent that there is a genetic basis to susceptibility to RVF, and it was postulated that the innate mechanisms involved also operate in humans and livestock to determine the manifestation of disease. It has been suggested that indigenous African breeds of livestock may be more resistant to RVF than exotic breeds, possibly through natural selection, but it was shown in limited experiments in Nigeria that local sheep were highly susceptible, and indigenous sheep, cattle and goats were severely affected in the epizootics in Egypt and West Africa.

The majority of RVF infections in humans are inapparent or associated with moderate to severe, non-fatal, febrile illness. After an incubation period of two to six days, the onset of the benign illness is usually very sudden and the disease is characterized by rigor, fever that persists for several days and is often biphasic, headache with retro-orbital pain and photophobia, weakness, and muscle and joint pains. Sometimes there is nausea and vomiting, abdominal pain, vertigo, epistaxis and a petechial rash. Viraemia in humans lasts for up to a week, with a maximum recorded intensity of 10^8.6 mouse intracerebral 50% lethal doses/ml (MICLD₅₀/ml). Defervescence and symptomatic improvement occur in four to seven days in benign disease and recovery is often complete in two weeks, but in a minority of patients the disease is complicated by the development of ocular lesions at the time of the initial illness or up to four weeks later. Estimates for the incidence of ocular complications range from less than 1% to 20% of human infections, and possibly the differences stem from failure to record mild cases in populations where illiterate persons are less likely to report minor disturbances of vision. The ocular disease usually presents as a loss of acuity of central vision, sometimes with development of scotomas. The essential lesion appears to be focal retinal ischaemia, generally in the macular or paramacular area, associated with thrombotic occlusion of arterioles and capillaries, and is characterized by retinal oedema and loss of transparency caused by dense white exudate and haemorrhages. Sometimes there is severe haemorrhage and detachment of the retina. The lesions and the loss of visual acuity generally resolve over a period of months with variable residual scarring of the retina, but in instances of severe haemorrhage and detachment of the retina there may be permanent uni- or bilateral blindness.

Probably less than 1% of human patients develop the haemorrhagic and/or encephalitic forms of the disease. Underlying liver disease may predispose to the haemorrhagic form of the illness. The haemorrhagic syndrome starts with sudden onset of febrile illness similar to the benign disease, but within two to four days there may be development of a petechial rash, purpura, ecchymoses and extensive subcutaneous haemorrhages, bleeding from needle puncture sites, epistaxis, haematemesis, diarrhoea and melaena, sore and inflamed throat, gingival bleeding, epigastric pain, hepatomegaly or hepatosplenomegaly, tenderness of the right upper quadrant of the abdomen and deep jaundice. This is followed by pneumonitis, anaemia, shock with racing pulse and low blood pressure, hepatorenal failure, coma and cardiorespiratory arrest. Factors contributing to fatal outcome in the hepatic form of the disease include anaemia, shock and hepatorenal failure, with the kidney lesions possibly being as important as shock in producing anuria. A proportion of the less severely affected patients may make a protracted recovery without sequelae.

Encephalitis may occur in combination with the haemorrhagic syndrome. Otherwise, signs of encephalitis in humans may supervene during the acute illness, or up to four weeks later and include severe headache, vertigo, confusion, disorientation, amnesia, meningismus, hallucinations, hypersalivation, grinding of teeth, choreiform movements, convulsions, hemiparesis, lethargy, decerebrate posturing, locked-in syndrome, coma and death. A proportion of patients may recover completely, but others may be left with sequelae, such as hemiparesis.

An attempt to relate the occurrence of abortion in humans to serological evidence of RVF infection in Egypt produced inconclusive results. Diagnosis of the infection in a neonatal child in Saudi Arabia in 2000 implied that there had been vertical transmission of infection, but the mother was not tested (Arishi et al. 2006).

The information available on clinical pathology findings in humans is compatible with observations made in haematological and coagulation studies on rhesus monkeys, except that leucocytosis and anaemia may be more marked in severe human disease (Peters et al. 1980; Al Hazmi et al. 2003). Rhesus monkeys may have prolonged activated partial thromboplastin times and prothrombin times even in benign infection, and in severe liver disease there may be depletion of coagulation factors II, V, VII, IX, X and XII, thrombocytopenia and platelet dysfunction, increased schistocyte counts and depletion of fibrinogen together with raised fibrin degradation product levels. Raised serum aspartate aminotransferase and alanine aminotransferase levels have been recorded even in benign disease in humans.

The disease may be suspected when there is a sudden outbreak of febrile illness with headache and myalgia in humans, in association with the occurrence of abortions in domestic ruminants and deaths of young animals. Sometimes the human disease is only recognized from the occurrence of ocular complications, or haemorrhagic or encephalitic manifestations, and this is especially true in the rare instances where residents of other continents develop the illness following a visit to Africa. Frequently, outbreaks of RVF in livestock only become evident after investigations have been triggered by the recognition of the disease in humans.

Specimens to be submitted for laboratory confirmation of the diagnosis include blood from live patients, and tissue samples, particularly liver, but also spleen, kidney, lymph nodes and heart blood of deceased patients. Tissue samples should be submitted in duplicate in a viral transport medium, and in 10% buffered–formalin for histopathological examination.

Viral antigen can often be detected rapidly in blood and other tissues by a variety of immunological methods, including immunodiffusion, complement-fixation, immunofluorescence and enzyme-linked immunoassay, and viral nucleic acid can readily be detected by RT-PCR (Jupp et al. 2002; Drosten et al. 2002; Weidmann et al. 2007; Bird et al. 2007b). The virus is cytopathic and can be isolated readily in almost all cell cultures commonly used in diagnostic laboratories, and identified rapidly by immunofluorescence. Virus can also be isolated in suckling or weaned mice, or hamsters, inoculated intracerebrally or intraperitoneally, and antigen can be identified in harvested brain or liver by the immunological methods mentioned above. Definitive identification of isolates is achieved by performing neutralization tests with reference antiserum.

Antibody to RVF virus can be demonstrated in complement-fixation, enzyme-linked immunoassay, indirect immunofluorescence, haemagglutination-inhibition, or neutralization tests. Diagnosis of recent infection is confirmed by demonstrating seroconversion or a four-fold or greater rise in titre of antibody in paired serum samples, or by demonstrating IgM antibody activity in an enzyme-linked immunoassay (Paweska et al. 2005, 2007).

Benign RVF in humans must be distinguished from other febrile zoonotic diseases such as brucellosis and Q fever which can be acquired from contact with livestock carcases, while the fulminant hepatic disease must be distinguished from the so-called formidable viral haemorrhagic fevers of Africa: Lassa fever, Crimean–Congo haemorrhagic fever, Marburg disease, Ebola fever and, theoretically, the haemorrhagic fever with renal syndrome associated with hantavirus infections (there has been serological evidence of, but no virologically confirmed case of the latter syndrome in Africa).

Histopathological lesions, particularly those in the liver, are considered to be pathognomonic, and are essentially similar in humans and domestic ruminants. The severity of the lesions varies from primary foci of coagulative necrosis, consisting of clusters of hepatocytes with acidophilic cytoplasms and pyknotic nuclei, multifocally scattered throughout the parenchyma, to massive liver destruction in which the primary foci comprising dense aggregates of cytoplasmic and nuclear debris, some fibrin and a few neutrophils and macrophages, can be discerned against a background of parenchyma reduced by nuclear pyknosis, karyorrhexis and cytolysis to scattered fragments of cytoplasm and chromatin, with only narrow rims of degenerated hepatocytes remaining reasonably intact close to portal triads. Intensely acidophilic cytoplasmic bodies which resemble the Councilman bodies of yellow fever are common, and rod-shaped or oval eosinophilic intranuclear inclusions may be seen in intact nuclei. Icterus may be evident.

Treatment is essentially symptomatic, and supportive therapy in the haemorrhagic disease includes replacement of blood and coagulation factors. Results obtained in animal models suggest that the administration of immune plasma from recovered patients may be beneficial. The antiviral drug ribavirin inhibits virus replication in cell cultures and laboratory animals, and it was suggested that it could be used even in benign disease in order to obviate the potentially serious complications which may occur in humans. However, the drug but did not prevent the late occurrence of encephalitis in patients in Saudi Arabia, and its use is now considered contraindicated in RVF infection.

Despite the sudden and dramatic change perceived in the nature of the human disease in the mid-1970s, it was deduced from the 598 reported deaths and 200,000 estimated cases of disease that RVF had a case fatality rate of less than 1% in Egypt where a high prevalence of schistosomiasis may have predisposed the population to severe liver disease. The fatality rate may even have been lower in relation to total infections, since an antibody prevalence rate of approximately 30% was detected in the human population estimated at one to three million in the areas affected by the epizootic. Remarkably high estimates of approximately 5 and 14% were made for case fatality rates in two separate populations in the 1987 epizootic in Mauritania, on the basis of the proportion of IgM antibody-positive persons who actually reported illness considered to be compatible with RVF, but it can be deduced that the fatality rates in terms of total IgM antibody-positive persons are much closer to the corresponding fatality rate in Egypt. The death rates of 20–45% estimated for recent outbreaks in East Africa are based on clinically-apparent or hospitalized cases only and there is no true denominator available from seroconversion studies.

Kenya, Tanzania, Somalia, South Africa, Namibia, Mozambique, Zimbabwe, Zambia, Sudan, Egypt, Mauritania, and Senegal have experienced large outbreaks of RVF as outlined above, while lesser outbreaks, periodic isolations of virus or serological evidence of infection have been recorded in Angola, Botswana, Burkina Faso, Cameroon, Central African Republic, Chad, Gabon, Guinea, Madagascar, Malawi, Mali, Nigeria, Uganda and Democratic Republic of the Congo.

Outbreaks of RVF in eastern and southern Africa have tended to occur at irregular intervals of up to 15 years or longer, and the fate of the virus during inter-epizootic periods has long constituted a central enigma in the epidemiology of the disease. On the basis of early observations made in Uganda, Kenya and South Africa, it was accepted for decades that the virus was enzootic in indigenous forests which extend in broken fashion from East Africa to the eastern and southern coastal regions of South Africa. The virus was thought to circulate in Eretmapodites spp. mosquitoes and unknown vertebrates in the forests, and to spread in seasons of exceptionally heavy rainfall to livestock rearing areas where the vectors were believed to be floodwater-breeding aedine mosquitoes of the subgenera Aedimorphus and Neomelaniconion, which attach their eggs to vegetation at the edge of stagnant surface water. In contrast to other culicine mosquitoes, it is obligatory that the eggs of aedines be subjected to a period of drying as the water recedes before they will hatch on being wetted again when next the area floods. Thus, the aedine mosquitoes overwinter as eggs which can survive for long periods in dried mud, possibly for several seasons if the area remains dry.

On the inland plateau of South Africa, where sheep rearing predominates, surface water gathers after heavy rains in undrained shallow depressions (pans) and farm dams which afford ideal breeding environments for aedines. On the watershed plateau of Zimbabwe, where cattle farming predominates, aedines breed in vleis, low-lying grassy areas which constitute drainage channels for surrounding high ground, and which are flooded by seepage after heavy rains. Vleis correspond to what are termed dambos in the livestock rearing areas of central and eastern Africa. Sustained monitoring in Zimbabwe revealed that a low level of virus transmission to livestock occurred each year in the same areas where epizootics occurred. The generation of epizootics, therefore, was associated with the simultaneous intensification of virus activity over vast livestock rearing areas where it was already present, rather than lateral spread from cryptic enzootic foci: examination of satellite images and aerial photographs revealed that the enzootic areas coincided with savannah and grasslands with a high density of vleis, and not with canopy forests. Subsequently, RVF virus was isolated from unfed Aedes mcintoshi mosquitoes (= Aedes lineatopennis sensu lato) hatched in dambos on a ranch in Kenya during inter-epizootic periods in 1982 and 1984, confirming that the virus is enzootic in livestock rearing areas and indicating that it appears to be maintained by transovarial transmission in aedines. The available evidence suggests that in Zimbabwe, as in Kenya, Aedes mcintoshi is the most important maintenance vector of the virus while Aedes dentatus is probably also a maintenance vector; the same two species and possibly Aedes unidentatus and Aedes juppi are maintenance vectors on the inland plateau of South Africa.

In contrast to countries such as Zimbabwe and Kenya, or even the coastal areas of South Africa, the inland plateau of South Africa has harsh winters and prolonged droughts are not uncommon, with pans and small dams remaining dry for many years or even decades at a time, so it is possible that aedine mosquito populations could decline to the point where RVF virus activity becomes virtually undetectable or the virus entirely disappears from the area. Indeed, no outbreaks of RVF have been recorded on the interior plateau of South Africa since the major epizootic of 1974–76, although small outbreaks were recognized in a coastal bush area in northern Natal in 1981 and in the north eastern Mpumalanga and Limpopo Provinces in 1999 and 2008 (Swanepoel and Paweska 2008). This suggests either that virus activity has declined on the inland plateau to a level where considerable amplification must occur before the disease again becomes evident, or that the virus has disappeared from the area and must be reintroduced through a mechanism permitting its long range dispersal, as discussed below in relation to the appearance of the disease in Egypt in 1977.

Epizootics generally become evident in late summer after there has been an initial increase in vector populations and in circulation of the virus. Heavy rainfall and the humid conditions which prevail during epizootics favour the breeding of other biting insects besides aedine mosquitoes. Following extensive flooding of aedine breeding sites, significant numbers of livestock become infected and circulate high levels of virus in their blood during the acute stage of infection. Other culicines and anopheline mosquitoes then become infected and serve as epizootic vectors, particularly Culex theileri in southern Africa, and biting flies such as midges, phlebotomids, stomoxids and simulids serve as mechanical transmitters of infection. Although contagion has been demonstrated on occasion under artificial conditions, non-vectorial transmission is not considered to be important in livestock, as opposed to humans. Outbreaks generally terminate in late autumn when the onset of cold weather depresses vector activity, or when most animals are immune following natural infection, or after there has been successful intervention with vaccine.

It can be deduced, and in some instances has been demonstrated directly, that the intensity of viraemia attained in domestic ruminants, humans and many rodents is adequate for the infection of the mosquito vectors of RVF virus through the ingestion of bloodmeals: estimated threshold levels of viraemia required to infect 50% of mosquitoes range from 10^5,7 to 10^8,7 MICLD₅₀/ml for the various putative vectors of southern Africa. Although extensive studies have failed to prove that the virus is maintained in natural transmission cycles in rodents, birds, or other wild vertebrates, it is felt that wild ruminants could play a role similar to their domestic counterparts in areas where they predominate. Furthermore, it is believed that the possibility that the virus is also maintained by circulation in forest mosquitoes and unidentified vertebrates, cannot be dismissed entirely and merits further investigation.

In retrospect, it can be surmised the occurrence of the massive epizootic in Egypt in 1977–78 was probably facilitated by an increase in mosquito breeding sites brought about by agricultural developments which followed the building of the Aswan dam, although it remains necessary to explain the mechanisms responsible for the introduction of the virus into the country. Various theories were advanced to account for the first known appearance of the virus in Egypt in 1977, including the long distance carriage of infected vectors at high altitude by prevailing winds associated with the inter-tropical convergence zone; a mechanism which has been invoked to explain the spread of many other arboviruses in the past. The introduction of the virus through the transportation of infected sheep and cattle on the Nile or overland from northern Sudan to markets in southern Egypt was also considered to have been a strong possibility, and the movement of slaughter animals by sea could account for the evidence of infection detected in the northern and eastern coastal areas of Egypt. Although transportation on some routes would take a long time in relation to the course of the infection, RVF virus has been shown to persist for prolonged periods in various organs of sheep, particularly the spleen for up to 21 days after infection. The same could be true for goats and cattle, or even the camels brought in by overland caravan routes. It is believed that humans slaughtering or handling the tissues of such animals could have become infected and served as the amplifying hosts for the infection of mosquitoes since the main vector in the Egyptian epizootic, Culex pipiens, is known to be peridomestic and anthropophilic. In at least one instance there were indications that human infections centred on a location where introduced camels were slaughtered. The incidence of the disease declined with the onset of the cool season in 1977, but it is thought that hibernation of infected adult Culex pipiens or other vector species, or a continued low level of biting activity by a proportion of the mosquito population, could account for the overwintering of the virus and the continuance of the epizootic into 1978.

In West Africa, the construction of the large Manantali dam on the Senegal River in Mali and the Diama dam downstream on the border between Mauritania and Senegal increased potential mosquito breeding sites in an area where the virus was already known to active, and prevailing drought conditions led to the concentration of nomadic people and their livestock in proximity to the dams. However, virus activity has declined in the arid ‘Sahelian’ region since the epizootic of 1987, and RVF is thought to be enzootic in the more humid `Guinean’ areas of West Africa, where Aedes mcintoshi, Aedes dalzieli and Aedes vexans are considered to be potentially important vectors.

There was speculation that RVF virus may have been imported into Saudi Arabia and Yemen from Africa with slaughter animals, or carried from Africa by wind-borne mosquitoes in 2000, but there were no known epidemics in the Horn of Africa at the time. It is much more likely that infected animals were imported during the 1997–98 epidemic in East Africa, and that infection smouldered on the Arabian Peninsula until ideal circumstances for an epidemic occurred following heavy rains in 2000 (Jupp et al. 2002).

In contrast to the main vector in the Egyptian epizootic of 1977–78, the principal mosquito vectors of RVF virus in sub-Saharan Africa tend to be zoophilic and sylvatic, with the result that humans become infected mainly from contact with animal tissues, although there are instances where no such history can be obtained and it must be assumed that infection has resulted from mosquito bite. Occasional infections diagnosed in tourists from abroad who have visited countries in Africa fall into this category. Generally, persons who become infected are involved in the livestock industry, such as farmers who assist in dystocia of livestock, farm labourers who salvage carcases for human consumption, veterinarians and their assistants, and abattoir workers. The virus is notorious as a cause of laboratory infections, and there are numerous reports of humans becoming infected while investigating the disease in the field. The results of surveys following epizootics in southern Africa indicated that nine to 15% of farm residents became infected, with a slight preponderance of adult males, although it appeared that housewives also gained infection from handling fresh meat.

No outbreaks of the disease have been recognized in urban consumer populations and it is surmised that the fall in pH associated with the maturation of meat in abattoirs is deleterious to the virus. Moreover, highest infection rates were found in workers in the by-products sections of abattoirs in Zimbabwe and the implication is that the carcases of infected animals which reach abattoirs are generally recognized as being diseased and are condemned as unfit for human consumption, and are then sterilized in the process of preparing carcase meal which is incorporated in animal feeds.

Human infection presumably results from contact of virus with abraded skin, wounds or mucous membranes, but aerosol and intranasal infection have been demonstrated experimentally and circumstantial evidence suggests that aerosols have been involved in some human infections in the laboratory, and in the field during the Egyptian outbreak of 1977–78. Many infections in Egypt are thought to have resulted from the slaughter of infected animals outside of abattoirs, and the fact that the mosquito vector was anthropophilic is thought to explain the high incidence of infection which occurred in people of all ages and diverse occupations. Low concentrations of virus have been found in milk and body fluids such as saliva and nasal discharges of sheep and cattle, and it appears that there may have been a connection between human infection and consumption of raw milk in Mauritania. In view of the intense viraemia which occurs in humans and the fact that virus has been isolated from throat washings, it is curious that there are no confirmed records of person to person transmission of infection.

Measures such as biological or chemical control of vectors, movement of livestock from low-lying areas to well drained and wind-swept pastures at higher altitudes, or confining of animals to mosquito-proof stables, are usually impractical or at best palliative in the face of a RVF epizootic, and immunization remains the only effective method of controlling the disease.

In addition to the modified live Smithburn strain and the formalin-inactivated vaccines referred to above, trials have been reported with small plaque variant and mutagen-derived candidate veterinary vaccines, but these have not been brought into commercial production. Promising candidate vaccines include the naturally attenuated clone 13 virus plus a genetically-engineered virus which lacks non-structural protein genes (Muller et al. 1995; Bird et al. 2008).

The Smithburn vaccine strain confers lifelong immunity in sheep and goats, and it is recommended that they should be immunized on a single occasion in the first year of life, preferably at six months of age after maternal immunity has waned. The Smithburn strain protects cattle against infection, but does not induce adequate humoral response to ensure transfer of colostral immunity to calves. Cattle and other domestic ruminants can be immunized at any age after maternal immunity has waned with inactivated vaccine, but the immunity is not durable and the animals should receive a second dose of vaccine three to six months later, plus annual boosters. It is, however, usually very difficult to persuade farmers to vaccinate livestock during long inter-epizootic periods, and the occurrence of outbreaks is difficult to predict. The result is that vaccine has almost invariably been used too late in the course of outbreaks to be fully effective. A further problem is that during outbreaks there is a chance of spreading infection with wild virus through transferring viraemic blood on needles used to inoculate different animals in succession. Nevertheless, in the past it has been practice in Kenya and South Africa to vaccinate all livestock, including pregnant ewes, with the Smithburn strain in the face of outbreaks, since it is deemed that the abortigenic and teratogenic effects of the vaccine are outweighed by the potentially severe consequences of allowing the disease to run its natural course. It is considered theoretically possible, although not proven, that live vaccine strains could revert to full virulence if passaged through hosts, as for instance through mosquitoes which become infected as a result of feeding on animals in the viraemic stage following administration of the vaccine. Hence, it is advised that only the inactivated vaccine should be used in situations where it is considered necessary to immunize animals in countries where the presence of RVF virus has not been proven.

Veterinarians and others engaged in the livestock industry should be made aware of the potential dangers of exposure to zoonotic agents in handling tissues of diseased animals, and precautions should be heightened during RVF epizootics. These should include the use of suitable protective clothing, such as an impervious gown or apron, gloves, and face mask or visor. The carcases of sick animals should not be utilized for human consumption. No registered vaccines are available for mass use on susceptible human populations, nor would their use be practicable in view of logistic problems and the essentially unpredictable occurrence and variable nature of outbreaks of the disease. A formalin-inactivated cell culture vaccine produced in the USA, was made available for use on a limited experimental basis, with the informed consent of recipients, to immunize persons such as veterinarians and laboratory workers who are regularly exposed to RVF infection, but no longer appears to be available.

Outbreaks of RVF in the Horn of Africa over the last 60 years have been shown to be related to the abnormally high and widespread rainfall caused by the El Nino in Southern Oscillation (ENSO) phenomenon. The resulting increase in green vegetation can be detected by satellite imaging (Linthicum et al. 1999). An RVF mapping model using climate data predicted an outbreak from December 2006–May 2007 in the Horn of Africa several months early. The predictions were confirmed by subsequent field investigations. Early warning advisory notices were given by US authorities and surveillance and outbreak response was put in place and may have saved lives (Anyamba et al. 2009). Increasingly, satellite imaging and prediction modelling will be used in control of such vector borne diseases.